In [1]:
from sklearn.metrics import r2_score
%run helper_functions.py
%run prophet_helper.py #this runs the TS models for features
%run regression_ts_model.py #nested TS script 
%run btc_info_df.py #helps loan jup new BTC data
%autosave 120
%matplotlib inline
plt.style.use('fivethirtyeight')
plt.rcParams["figure.figsize"] = (15,10)
plt.rcParams["xtick.labelsize"] = 16
plt.rcParams["ytick.labelsize"] = 16
plt.rcParams["axes.labelsize"] = 20
plt.rcParams['legend.fontsize'] = 20
plt.style.use('fivethirtyeight')
pd.set_option('display.max_colwidth', -1)


Autosaving every 120 seconds

Notebook Overview

In this notebook, I will construct:

  • A naive model of bitcoin price prediction

  • A nested time series model.

What do I mean by a nested time series model?

I will illustrate with a simple example.

Let's say that I wish to predict the mkt_price on 2016-10-30. I could fit a Linear Regression on all the features from 2016-10-26 - 29-10-2016. However, in order to predict the price of mkt_price on 2016-10-30 I need to have values for the features on 2016-10-30. This presents a problem as all my features are time series! That is, I cannot simply plug in a value for all the features because I don't know what their values would be on this future date!

One possible remedy for this is to simply use the values of all the features on 2016-10-29. In fact, it is well know that the best predictor of a variable tomorrow is it's current state today. However, I wish to be more rigorous.

Instead of simply plugging in t-1 values for the features at time t, I construct a time series model for each feature in order to predict its value at time t based on the entire history of data that I have for the features!

These predicted values are then passed as inputs to our linear regression models!

Thus, if I have N features, I am creating N-Time Series models in order to do a single prediction with Linear Regression for the mkt_price variable.

Naive Baseline Model

I will construct a naive baseline model that will most likely outperorm any other model I build below.

The model will work as follows:

When predicting the price on Day 91, I will take the average price change between Day 90 and Day 0. Let's call this average price change alpha.

I will then take the price of Day 90 and add alpha to it. This will serve as the 'predicted' price for day 91.


In [2]:
df = unpickle_object("FINAL_DATAFRAME_PROJ_5.pkl")
df.head()


Out[2]:
gold_price eth_price pos_sent neg_sent tot_num_trans unique_addr hash_rate mempool_trans avg_trans_per_block mkt_price
date
2016-10-26 1273.9 11.455 0.143706 0.066297 225924.0 431781.0 1.752433e+06 69811.000000 1625.352518 672.221413
2016-10-27 1269.3 11.515 0.127429 0.064310 326388.0 457806.0 2.042404e+06 171358.250000 2014.740741 682.223962
2016-10-28 1265.9 11.280 0.128794 0.056393 250876.0 434269.0 1.916330e+06 131888.333333 1650.500000 687.688337
2016-10-29 1265.9 10.770 0.139114 0.063177 229689.0 416457.0 1.878507e+06 17765.750000 1541.536913 714.895450
2016-10-30 1265.9 10.805 0.131922 0.064369 209337.0 360327.0 2.130656e+06 20822.250000 1238.680473 698.003400

In [15]:
def linear_extrapolation(df, window):
    pred_lst = []
    true_lst = []

    cnt = 0

    all_rows = df.shape[0]

    while cnt < window:
        start = df.iloc[cnt:all_rows-window+cnt, :].index[0].date()
        end = df.iloc[cnt:all_rows-window+cnt, :].index[-1].date()
        predicting = df.iloc[all_rows-window+cnt, :].name.date()

        print("---- Running model from {} to {} and predicting on {} ----".format(start,end,predicting))

        training_df = df.iloc[cnt:all_rows-window+cnt, :]

        testing_df = df.iloc[all_rows-window+cnt, :]
        
        true_val = testing_df[-1]
        
        first_row_value = training_df.iloc[0, :]['mkt_price']
        first_row_date = training_df.iloc[0, :].name
        
        last_row_value = training_df.iloc[-1, :]['mkt_price']
        last_row_date = training_df.iloc[-1, :].name
        
        alpha = (last_row_value-first_row_value)/90
        
        prediction = last_row_value + alpha
        
        pred_lst.append(prediction)
        
        true_lst.append(true_val)
        
        
        cnt += 1
        
    return pred_lst, true_lst

In [16]:
pred_lst, true_lst = linear_extrapolation(df, 30)


---- Running model from 2016-10-26 to 2017-01-23 and predicting on 2017-01-24 ----
---- Running model from 2016-10-27 to 2017-01-24 and predicting on 2017-01-25 ----
---- Running model from 2016-10-28 to 2017-01-25 and predicting on 2017-01-26 ----
---- Running model from 2016-10-29 to 2017-01-26 and predicting on 2017-01-27 ----
---- Running model from 2016-10-30 to 2017-01-27 and predicting on 2017-01-28 ----
---- Running model from 2016-10-31 to 2017-01-28 and predicting on 2017-01-29 ----
---- Running model from 2016-11-01 to 2017-01-29 and predicting on 2017-01-30 ----
---- Running model from 2016-11-02 to 2017-01-30 and predicting on 2017-01-31 ----
---- Running model from 2016-11-03 to 2017-01-31 and predicting on 2017-02-01 ----
---- Running model from 2016-11-04 to 2017-02-01 and predicting on 2017-02-02 ----
---- Running model from 2016-11-05 to 2017-02-02 and predicting on 2017-02-03 ----
---- Running model from 2016-11-06 to 2017-02-03 and predicting on 2017-02-04 ----
---- Running model from 2016-11-07 to 2017-02-04 and predicting on 2017-02-05 ----
---- Running model from 2016-11-08 to 2017-02-05 and predicting on 2017-02-06 ----
---- Running model from 2016-11-09 to 2017-02-06 and predicting on 2017-02-07 ----
---- Running model from 2016-11-10 to 2017-02-07 and predicting on 2017-02-08 ----
---- Running model from 2016-11-11 to 2017-02-08 and predicting on 2017-02-09 ----
---- Running model from 2016-11-12 to 2017-02-09 and predicting on 2017-02-10 ----
---- Running model from 2016-11-13 to 2017-02-10 and predicting on 2017-02-11 ----
---- Running model from 2016-11-14 to 2017-02-11 and predicting on 2017-02-12 ----
---- Running model from 2016-11-15 to 2017-02-12 and predicting on 2017-02-13 ----
---- Running model from 2016-11-16 to 2017-02-13 and predicting on 2017-02-14 ----
---- Running model from 2016-11-17 to 2017-02-14 and predicting on 2017-02-15 ----
---- Running model from 2016-11-18 to 2017-02-15 and predicting on 2017-02-16 ----
---- Running model from 2016-11-19 to 2017-02-16 and predicting on 2017-02-17 ----
---- Running model from 2016-11-20 to 2017-02-17 and predicting on 2017-02-18 ----
---- Running model from 2016-11-21 to 2017-02-18 and predicting on 2017-02-19 ----
---- Running model from 2016-11-22 to 2017-02-19 and predicting on 2017-02-20 ----
---- Running model from 2016-11-23 to 2017-02-20 and predicting on 2017-02-21 ----
---- Running model from 2016-11-24 to 2017-02-21 and predicting on 2017-02-22 ----

In [17]:
r2_score(true_lst, pred_lst)


Out[17]:
0.86689253204369099

Naïve Model Caveats

We can see above that we can use this extremely basic model to obtain an $R^2$ of 0.86. In fact, this should be the baseline model score that we need to beat!

Let me mention some caveats to this result:

  • I only have 4 months of Bitcoin data. It should be obvious to the reader that such a naive model is NOT the appropriate way to forecast bitcoin price in general. For if it were this simple, we would all be millionaires.
  • Since I have 120 days worth of day, I am choosing to subset my data in 90 day periods, as such, I will produce 30 predictions. The variability of bitcoin prices around these 30 days will significantly impact the $R^2$ score. Again, more data is needed.
  • While bitcoin data itself is not hard to come by, twitter data is! It is the twitter data that is limiting a deeper analysis. I hope that this notebook serves as a starting point for further investigation in the relationship between tweets and bitcoin price fluctuations.
  • Lastly, I have made this notebook in Sept. 2017. The data for this project spans Oct 2016 - Feb 2017. Since that timeframe, bitcoin grew to unprecedented highs of \$4k/coin. Furthermore, media sound bites of CEOs such as James Dimon of JPMorgan have sent bitcoin prices tumbling by as much as $1k/coin. For me, this is what truly lies at the crux of the difficulty of cryptocurrency forecasting. I searched at great length for a free, searchable NEWS API, however, I could not find one. I think I great next step for this project would be to incorporate sentiment of news headlines concerning bitcoin!
  • Furthermore, with the aforementioned timeframe, the overall bitcoin trend was upward. That is, there was not that much volatility in the price - as such, it is expected that the Naïve Model would outperform the nested time series model. The next step would again, be to collect more data and re-run all the models.

Nested Time Series Model


In [18]:
df = unpickle_object("FINAL_DATAFRAME_PROJ_5.pkl")
df.head()


Out[18]:
gold_price eth_price pos_sent neg_sent tot_num_trans unique_addr hash_rate mempool_trans avg_trans_per_block mkt_price
date
2016-10-26 1273.9 11.455 0.143706 0.066297 225924.0 431781.0 1.752433e+06 69811.000000 1625.352518 672.221413
2016-10-27 1269.3 11.515 0.127429 0.064310 326388.0 457806.0 2.042404e+06 171358.250000 2014.740741 682.223962
2016-10-28 1265.9 11.280 0.128794 0.056393 250876.0 434269.0 1.916330e+06 131888.333333 1650.500000 687.688337
2016-10-29 1265.9 10.770 0.139114 0.063177 229689.0 416457.0 1.878507e+06 17765.750000 1541.536913 714.895450
2016-10-30 1265.9 10.805 0.131922 0.064369 209337.0 360327.0 2.130656e+06 20822.250000 1238.680473 698.003400

In [19]:
df.corr()


Out[19]:
gold_price eth_price pos_sent neg_sent tot_num_trans unique_addr hash_rate mempool_trans avg_trans_per_block mkt_price
gold_price 1.000000 0.757126 -0.144099 -0.253473 -0.145614 -0.179554 -0.078998 0.087767 -0.101640 -0.192922
eth_price 0.757126 1.000000 -0.164950 -0.116261 0.054744 0.015746 0.424836 0.272740 0.071475 0.354780
pos_sent -0.144099 -0.164950 1.000000 -0.331122 -0.364671 -0.314049 -0.046629 -0.110460 -0.323005 0.095882
neg_sent -0.253473 -0.116261 -0.331122 1.000000 0.232886 0.186956 0.159483 0.316399 0.217727 0.062722
tot_num_trans -0.145614 0.054744 -0.364671 0.232886 1.000000 0.912685 0.300830 0.405603 0.743069 0.256059
unique_addr -0.179554 0.015746 -0.314049 0.186956 0.912685 1.000000 0.303917 0.313126 0.692544 0.278896
hash_rate -0.078998 0.424836 -0.046629 0.159483 0.300830 0.303917 1.000000 0.289567 -0.023170 0.784233
mempool_trans 0.087767 0.272740 -0.110460 0.316399 0.405603 0.313126 0.289567 1.000000 0.362882 0.249801
avg_trans_per_block -0.101640 0.071475 -0.323005 0.217727 0.743069 0.692544 -0.023170 0.362882 1.000000 0.228785
mkt_price -0.192922 0.354780 0.095882 0.062722 0.256059 0.278896 0.784233 0.249801 0.228785 1.000000

In [20]:
plot_corr_matrix(df)



In [21]:
beta_values, pred, true = master(df, 30)


---- Running model from 2016-10-26 to 2017-01-23 and predicting on 2017-01-24 ----
Creating Time series models: 
The real value for gold_price on 2017-01-24 is 1213.2999999998965
The predicted value for gold_price on 2017-01-24 is 1221.11026869873
The real value for eth_price on 2017-01-24 is 10.680000000000259
The predicted value for eth_price on 2017-01-24 is 10.965185902923128
The real value for pos_sent on 2017-01-24 is 0.12018382032272137
The predicted value for pos_sent on 2017-01-24 is 0.13216508320142728
The real value for neg_sent on 2017-01-24 is 0.06455189707806362
The predicted value for neg_sent on 2017-01-24 is 0.06816904538398737
The real value for tot_num_trans on 2017-01-24 is 278177.0
The predicted value for tot_num_trans on 2017-01-24 is 282370.2436552672
The real value for unique_addr on 2017-01-24 is 497858.0
The predicted value for unique_addr on 2017-01-24 is 496028.5135409737
The real value for hash_rate on 2017-01-24 is 2910612.1491232086
The predicted value for hash_rate on 2017-01-24 is 2839892.3785451157
The real value for mempool_trans on 2017-01-24 is 21244.0
The predicted value for mempool_trans on 2017-01-24 is 35101.8901949932
The real value for avg_trans_per_block on 2017-01-24 is 1866.959731543624
The predicted value for avg_trans_per_block on 2017-01-24 is 1787.3252913256674
The real value of mkt_price was [[ 890.320225]] on 2017-01-24
The predicted value of mkt_price was [[ 913.22114805]] on 2017-01-24

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.688
Model:                            OLS   Adj. R-squared:                  0.653
Method:                 Least Squares   F-statistic:                     19.57
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           7.20e-17
Time:                        16:17:39   Log-Likelihood:                -486.47
No. Observations:                  90   AIC:                             992.9
Df Residuals:                      80   BIC:                             1018.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1060.6482    490.633      2.162      0.034      84.258    2037.039
x1            -1.1592      0.347     -3.342      0.001      -1.849      -0.469
x2            30.9973     11.117      2.788      0.007       8.874      53.121
x3          2596.2549    857.678      3.027      0.003     889.420    4303.089
x4          -973.9708    913.738     -1.066      0.290   -2792.368     844.427
x5            -0.0007      0.001     -1.199      0.234      -0.002       0.000
x6         -3.667e-05      0.000     -0.107      0.915      -0.001       0.001
x7             0.0002    3.8e-05      4.713      0.000       0.000       0.000
x8            -0.0004      0.000     -1.918      0.059      -0.001    1.41e-05
x9             0.2148      0.066      3.236      0.002       0.083       0.347
==============================================================================
Omnibus:                       17.935   Durbin-Watson:                   0.557
Prob(Omnibus):                  0.000   Jarque-Bera (JB):               22.387
Skew:                           0.989   Prob(JB):                     1.38e-05
Kurtosis:                       4.433   Cond. No.                     4.18e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.18e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-10-27 to 2017-01-24 and predicting on 2017-01-25 ----
Creating Time series models: 
The real value for gold_price on 2017-01-25 is 1203.5
The predicted value for gold_price on 2017-01-25 is 1224.4639867641067
The real value for eth_price on 2017-01-25 is 10.595000000000706
The predicted value for eth_price on 2017-01-25 is 11.294641018212522
The real value for pos_sent on 2017-01-25 is 0.1286969139587164
The predicted value for pos_sent on 2017-01-25 is 0.12430585971633812
The real value for neg_sent on 2017-01-25 is 0.07623523400776616
The predicted value for neg_sent on 2017-01-25 is 0.07051948424425407
The real value for tot_num_trans on 2017-01-25 is 305961.0
The predicted value for tot_num_trans on 2017-01-25 is 322171.8217809314
The real value for unique_addr on 2017-01-25 is 514638.0
The predicted value for unique_addr on 2017-01-25 is 499030.2456465836
The real value for hash_rate on 2017-01-25 is 2930146.458848869
The predicted value for hash_rate on 2017-01-25 is 2926339.1470993166
The real value for mempool_trans on 2017-01-25 is 217339.5
The predicted value for mempool_trans on 2017-01-25 is 78005.05677586656
The real value for avg_trans_per_block on 2017-01-25 is 2039.74
The predicted value for avg_trans_per_block on 2017-01-25 is 1971.258244895787
The real value of mkt_price was [[ 893.045625]] on 2017-01-25
The predicted value of mkt_price was [[ 902.51807628]] on 2017-01-25

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.684
Model:                            OLS   Adj. R-squared:                  0.648
Method:                 Least Squares   F-statistic:                     19.23
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           1.14e-16
Time:                        16:17:51   Log-Likelihood:                -486.47
No. Observations:                  90   AIC:                             992.9
Df Residuals:                      80   BIC:                             1018.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1095.2901    484.166      2.262      0.026     131.768    2058.812
x1            -1.1982      0.342     -3.499      0.001      -1.880      -0.517
x2            32.4377     11.213      2.893      0.005      10.122      54.753
x3          2707.0364    864.311      3.132      0.002     987.002    4427.071
x4          -877.4623    919.836     -0.954      0.343   -2707.995     953.070
x5            -0.0007      0.001     -1.150      0.254      -0.002       0.001
x6         -3.359e-05      0.000     -0.098      0.922      -0.001       0.001
x7             0.0002    3.6e-05      4.746      0.000    9.93e-05       0.000
x8            -0.0004      0.000     -1.914      0.059      -0.001    1.49e-05
x9             0.2040      0.064      3.211      0.002       0.078       0.330
==============================================================================
Omnibus:                       18.364   Durbin-Watson:                   0.560
Prob(Omnibus):                  0.000   Jarque-Bera (JB):               23.044
Skew:                           1.010   Prob(JB):                     9.91e-06
Kurtosis:                       4.437   Cond. No.                     4.26e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.26e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-10-28 to 2017-01-25 and predicting on 2017-01-26 ----
Creating Time series models: 
The real value for gold_price on 2017-01-26 is 1191.5499999998972
The predicted value for gold_price on 2017-01-26 is 1214.577889262119
The real value for eth_price on 2017-01-26 is 10.580000000001139
The predicted value for eth_price on 2017-01-26 is 11.256516397426093
The real value for pos_sent on 2017-01-26 is 0.13635754431474312
The predicted value for pos_sent on 2017-01-26 is 0.1258370614805339
The real value for neg_sent on 2017-01-26 is 0.06979420665801993
The predicted value for neg_sent on 2017-01-26 is 0.06932766812476852
The real value for tot_num_trans on 2017-01-26 is 317904.0
The predicted value for tot_num_trans on 2017-01-26 is 285819.4406849477
The real value for unique_addr on 2017-01-26 is 518118.0
The predicted value for unique_addr on 2017-01-26 is 475889.51102365635
The real value for hash_rate on 2017-01-26 is 3066886.6269284827
The predicted value for hash_rate on 2017-01-26 is 2962157.562481624
The real value for mempool_trans on 2017-01-26 is 194245.5
The predicted value for mempool_trans on 2017-01-26 is 79162.04703379996
The real value for avg_trans_per_block on 2017-01-26 is 2024.8662420382168
The predicted value for avg_trans_per_block on 2017-01-26 is 1753.697813134939
The real value of mkt_price was [[ 915.95625]] on 2017-01-26
The predicted value of mkt_price was [[ 907.24446877]] on 2017-01-26

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.681
Model:                            OLS   Adj. R-squared:                  0.645
Method:                 Least Squares   F-statistic:                     18.98
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           1.62e-16
Time:                        16:18:02   Log-Likelihood:                -486.45
No. Observations:                  90   AIC:                             992.9
Df Residuals:                      80   BIC:                             1018.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1088.9049    466.968      2.332      0.022     159.609    2018.201
x1            -1.1915      0.333     -3.583      0.001      -1.853      -0.530
x2            32.3820     11.116      2.913      0.005      10.260      54.503
x3          2707.4466    862.536      3.139      0.002     990.946    4423.947
x4          -892.3400    921.542     -0.968      0.336   -2726.266     941.586
x5            -0.0007      0.001     -1.094      0.277      -0.002       0.001
x6           -5.7e-05      0.000     -0.154      0.878      -0.001       0.001
x7             0.0002   3.31e-05      5.183      0.000       0.000       0.000
x8            -0.0004      0.000     -2.056      0.043      -0.001   -1.15e-05
x9             0.2053      0.060      3.439      0.001       0.086       0.324
==============================================================================
Omnibus:                       18.317   Durbin-Watson:                   0.563
Prob(Omnibus):                  0.000   Jarque-Bera (JB):               23.004
Skew:                           1.006   Prob(JB):                     1.01e-05
Kurtosis:                       4.444   Cond. No.                     4.28e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.28e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-10-29 to 2017-01-26 and predicting on 2017-01-27 ----
Creating Time series models: 
The real value for gold_price on 2017-01-27 is 1184.2000000001008
The predicted value for gold_price on 2017-01-27 is 1206.7016498116755
The real value for eth_price on 2017-01-27 is 10.559999999999654
The predicted value for eth_price on 2017-01-27 is 10.98330315012105
The real value for pos_sent on 2017-01-27 is 0.13552577092510995
The predicted value for pos_sent on 2017-01-27 is 0.13187147125493057
The real value for neg_sent on 2017-01-27 is 0.07624515418502215
The predicted value for neg_sent on 2017-01-27 is 0.07107484693700307
The real value for tot_num_trans on 2017-01-27 is 263332.0
The predicted value for tot_num_trans on 2017-01-27 is 278220.1203876258
The real value for unique_addr on 2017-01-27 is 462950.0
The predicted value for unique_addr on 2017-01-27 is 471967.86533922085
The real value for hash_rate on 2017-01-27 is 3320832.65336205
The predicted value for hash_rate on 2017-01-27 is 2908155.012503354
The real value for mempool_trans on 2017-01-27 is 33070.5
The predicted value for mempool_trans on 2017-01-27 is 51090.904699314735
The real value for avg_trans_per_block on 2017-01-27 is 1549.0117647058826
The predicted value for avg_trans_per_block on 2017-01-27 is 1794.5093092848067
The real value of mkt_price was [[ 919.27975]] on 2017-01-27
The predicted value of mkt_price was [[ 934.48726506]] on 2017-01-27

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.679
Model:                            OLS   Adj. R-squared:                  0.643
Method:                 Least Squares   F-statistic:                     18.81
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           2.02e-16
Time:                        16:18:13   Log-Likelihood:                -486.54
No. Observations:                  90   AIC:                             993.1
Df Residuals:                      80   BIC:                             1018.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1098.8992    470.648      2.335      0.022     162.280    2035.518
x1            -1.1985      0.333     -3.597      0.001      -1.861      -0.536
x2            32.1904     11.204      2.873      0.005       9.894      54.487
x3          2703.0036    869.824      3.108      0.003     971.998    4434.009
x4          -808.7942    936.845     -0.863      0.391   -2673.175    1055.587
x5            -0.0006      0.001     -1.058      0.293      -0.002       0.001
x6         -4.833e-05      0.000     -0.130      0.897      -0.001       0.001
x7             0.0002   3.24e-05      5.220      0.000       0.000       0.000
x8            -0.0004      0.000     -2.341      0.022      -0.001   -6.07e-05
x9             0.2010      0.059      3.422      0.001       0.084       0.318
==============================================================================
Omnibus:                       18.606   Durbin-Watson:                   0.571
Prob(Omnibus):                  0.000   Jarque-Bera (JB):               23.414
Skew:                           1.022   Prob(JB):                     8.24e-06
Kurtosis:                       4.438   Cond. No.                     4.41e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.41e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-10-30 to 2017-01-27 and predicting on 2017-01-28 ----
Creating Time series models: 
The real value for gold_price on 2017-01-28 is 1184.2000000000949
The predicted value for gold_price on 2017-01-28 is 1203.2079587454868
The real value for eth_price on 2017-01-28 is 10.55000000000041
The predicted value for eth_price on 2017-01-28 is 11.056536918371878
The real value for pos_sent on 2017-01-28 is 0.151849140974347
The predicted value for pos_sent on 2017-01-28 is 0.13568027908229693
The real value for neg_sent on 2017-01-28 is 0.060212285243586704
The predicted value for neg_sent on 2017-01-28 is 0.07043870103503579
The real value for tot_num_trans on 2017-01-28 is 246395.0
The predicted value for tot_num_trans on 2017-01-28 is 260730.73632839049
The real value for unique_addr on 2017-01-28 is 461291.0
The predicted value for unique_addr on 2017-01-28 is 453344.2814615224
The real value for hash_rate on 2017-01-28 is 2969215.0783001864
The predicted value for hash_rate on 2017-01-28 is 3019045.1569345775
The real value for mempool_trans on 2017-01-28 is 16016.5
The predicted value for mempool_trans on 2017-01-28 is 53653.96974985474
The real value for avg_trans_per_block on 2017-01-28 is 1621.0197368421052
The predicted value for avg_trans_per_block on 2017-01-28 is 1596.0791471049663
The real value of mkt_price was [[ 920.31225]] on 2017-01-28
The predicted value of mkt_price was [[ 933.83086855]] on 2017-01-28

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.675
Model:                            OLS   Adj. R-squared:                  0.638
Method:                 Least Squares   F-statistic:                     18.45
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           3.37e-16
Time:                        16:18:24   Log-Likelihood:                -487.27
No. Observations:                  90   AIC:                             994.5
Df Residuals:                      80   BIC:                             1020.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1179.5584    469.135      2.514      0.014     245.950    2113.167
x1            -1.2218      0.335     -3.644      0.000      -1.889      -0.554
x2            31.7225     11.340      2.797      0.006       9.155      54.290
x3          2563.1695    870.986      2.943      0.004     829.852    4296.487
x4         -1026.3403    933.286     -1.100      0.275   -2883.639     830.959
x5            -0.0006      0.001     -0.961      0.339      -0.002       0.001
x6         -5.806e-05      0.000     -0.155      0.877      -0.001       0.001
x7             0.0002   3.21e-05      5.032      0.000    9.75e-05       0.000
x8            -0.0004      0.000     -2.201      0.031      -0.001   -3.65e-05
x9             0.1950      0.059      3.298      0.001       0.077       0.313
==============================================================================
Omnibus:                       18.278   Durbin-Watson:                   0.557
Prob(Omnibus):                  0.000   Jarque-Bera (JB):               22.587
Skew:                           1.022   Prob(JB):                     1.25e-05
Kurtosis:                       4.358   Cond. No.                     4.37e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.37e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-10-31 to 2017-01-28 and predicting on 2017-01-29 ----
Creating Time series models: 
The real value for gold_price on 2017-01-29 is 1184.2000000000999
The predicted value for gold_price on 2017-01-29 is 1198.2221728480995
The real value for eth_price on 2017-01-29 is 10.519999999999284
The predicted value for eth_price on 2017-01-29 is 11.154009662816872
The real value for pos_sent on 2017-01-29 is 0.13990977946090463
The predicted value for pos_sent on 2017-01-29 is 0.13162127923735864
The real value for neg_sent on 2017-01-29 is 0.057756070394297095
The predicted value for neg_sent on 2017-01-29 is 0.06576693543331665
The real value for tot_num_trans on 2017-01-29 is 241627.0
The predicted value for tot_num_trans on 2017-01-29 is 242347.3733516791
The real value for unique_addr on 2017-01-29 is 438727.0
The predicted value for unique_addr on 2017-01-29 is 433853.43331979803
The real value for hash_rate on 2017-01-29 is 3262229.7241850733
The predicted value for hash_rate on 2017-01-29 is 3072365.4218318667
The real value for mempool_trans on 2017-01-29 is 34131.5
The predicted value for mempool_trans on 2017-01-29 is 26815.94445831565
The real value for avg_trans_per_block on 2017-01-29 is 1446.868263473054
The predicted value for avg_trans_per_block on 2017-01-29 is 1500.2264768365828
The real value of mkt_price was [[ 915.933]] on 2017-01-29
The predicted value of mkt_price was [[ 941.2981646]] on 2017-01-29

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.666
Model:                            OLS   Adj. R-squared:                  0.628
Method:                 Least Squares   F-statistic:                     17.70
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           9.83e-16
Time:                        16:18:34   Log-Likelihood:                -488.42
No. Observations:                  90   AIC:                             996.8
Df Residuals:                      80   BIC:                             1022.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1245.9109    478.621      2.603      0.011     293.426    2198.396
x1            -1.2122      0.341     -3.558      0.001      -1.890      -0.534
x2            29.8388     11.477      2.600      0.011       6.998      52.680
x3          2286.7302    862.763      2.650      0.010     569.776    4003.684
x4         -1079.5731    947.222     -1.140      0.258   -2964.605     805.458
x5            -0.0005      0.001     -0.730      0.468      -0.002       0.001
x6            -0.0001      0.000     -0.265      0.792      -0.001       0.001
x7             0.0002   3.25e-05      4.753      0.000    8.98e-05       0.000
x8            -0.0004      0.000     -2.078      0.041      -0.001   -1.55e-05
x9             0.1843      0.061      3.042      0.003       0.064       0.305
==============================================================================
Omnibus:                       18.883   Durbin-Watson:                   0.521
Prob(Omnibus):                  0.000   Jarque-Bera (JB):               23.590
Skew:                           1.048   Prob(JB):                     7.54e-06
Kurtosis:                       4.378   Cond. No.                     4.35e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.35e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-01 to 2017-01-29 and predicting on 2017-01-30 ----
Creating Time series models: 
The real value for gold_price on 2017-01-30 is 1189.849999999963
The predicted value for gold_price on 2017-01-30 is 1188.3489508618525
The real value for eth_price on 2017-01-30 is 10.529999999999326
The predicted value for eth_price on 2017-01-30 is 11.00264121846572
The real value for pos_sent on 2017-01-30 is 0.12725906623235608
The predicted value for pos_sent on 2017-01-30 is 0.13143704752745614
The real value for neg_sent on 2017-01-30 is 0.07339500542888144
The predicted value for neg_sent on 2017-01-30 is 0.0635436765881497
The real value for tot_num_trans on 2017-01-30 is 250829.0
The predicted value for tot_num_trans on 2017-01-30 is 263864.7633348682
The real value for unique_addr on 2017-01-30 is 449341.0
The predicted value for unique_addr on 2017-01-30 is 473593.5051955378
The real value for hash_rate on 2017-01-30 is 2852009.2199462317
The predicted value for hash_rate on 2017-01-30 is 3038363.9205564815
The real value for mempool_trans on 2017-01-30 is 30839.0
The predicted value for mempool_trans on 2017-01-30 is 21039.70359582609
The real value for avg_trans_per_block on 2017-01-30 is 1718.0068493150684
The predicted value for avg_trans_per_block on 2017-01-30 is 1677.3441009734142
The real value of mkt_price was [[ 921.179325]] on 2017-01-30
The predicted value of mkt_price was [[ 960.4465163]] on 2017-01-30

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.659
Model:                            OLS   Adj. R-squared:                  0.621
Method:                 Least Squares   F-statistic:                     17.21
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           1.98e-15
Time:                        16:18:45   Log-Likelihood:                -489.08
No. Observations:                  90   AIC:                             998.2
Df Residuals:                      80   BIC:                             1023.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1253.2179    485.551      2.581      0.012     286.942    2219.494
x1            -1.2134      0.344     -3.522      0.001      -1.899      -0.528
x2            29.6420     11.646      2.545      0.013       6.466      52.818
x3          2281.5370    874.797      2.608      0.011     540.635    4022.439
x4          -982.1304    962.660     -1.020      0.311   -2897.885     933.624
x5            -0.0004      0.001     -0.616      0.540      -0.002       0.001
x6            -0.0001      0.000     -0.263      0.793      -0.001       0.001
x7             0.0001   3.24e-05      4.555      0.000     8.3e-05       0.000
x8            -0.0004      0.000     -2.099      0.039      -0.001   -1.92e-05
x9             0.1773      0.061      2.915      0.005       0.056       0.298
==============================================================================
Omnibus:                       19.127   Durbin-Watson:                   0.502
Prob(Omnibus):                  0.000   Jarque-Bera (JB):               23.870
Skew:                           1.065   Prob(JB):                     6.56e-06
Kurtosis:                       4.351   Cond. No.                     4.43e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.43e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-02 to 2017-01-30 and predicting on 2017-01-31 ----
Creating Time series models: 
The real value for gold_price on 2017-01-31 is 1198.799999999894
The predicted value for gold_price on 2017-01-31 is 1191.2870096075244
The real value for eth_price on 2017-01-31 is 10.644999999999103
The predicted value for eth_price on 2017-01-31 is 11.133370751182005
The real value for pos_sent on 2017-01-31 is 0.12829487983281093
The predicted value for pos_sent on 2017-01-31 is 0.125619048540313
The real value for neg_sent on 2017-01-31 is 0.06551745036572629
The predicted value for neg_sent on 2017-01-31 is 0.06348677339163648
The real value for tot_num_trans on 2017-01-31 is 307683.0
The predicted value for tot_num_trans on 2017-01-31 is 292985.9535665864
The real value for unique_addr on 2017-01-31 is 523036.0
The predicted value for unique_addr on 2017-01-31 is 497532.25697522954
The real value for hash_rate on 2017-01-31 is 2832474.9102205737
The predicted value for hash_rate on 2017-01-31 is 3002809.6526979352
The real value for mempool_trans on 2017-01-31 is 57173.5
The predicted value for mempool_trans on 2017-01-31 is 36325.202373350716
The real value for avg_trans_per_block on 2017-01-31 is 2121.9517241379317
The predicted value for avg_trans_per_block on 2017-01-31 is 1904.9122990063809
The real value of mkt_price was [[ 964.706075]] on 2017-01-31
The predicted value of mkt_price was [[ 963.78948841]] on 2017-01-31

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.663
Model:                            OLS   Adj. R-squared:                  0.625
Method:                 Least Squares   F-statistic:                     17.49
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           1.32e-15
Time:                        16:18:56   Log-Likelihood:                -488.74
No. Observations:                  90   AIC:                             997.5
Df Residuals:                      80   BIC:                             1022.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1242.3099    483.889      2.567      0.012     279.341    2205.279
x1            -1.2225      0.343     -3.561      0.001      -1.906      -0.539
x2            29.5707     11.594      2.550      0.013       6.497      52.644
x3          2336.2697    873.893      2.673      0.009     597.167    4075.373
x4          -896.2310    960.264     -0.933      0.353   -2807.217    1014.755
x5            -0.0004      0.001     -0.626      0.533      -0.002       0.001
x6            -0.0001      0.000     -0.253      0.801      -0.001       0.001
x7             0.0001   3.21e-05      4.662      0.000    8.57e-05       0.000
x8            -0.0004      0.000     -2.147      0.035      -0.001   -2.77e-05
x9             0.1791      0.060      2.976      0.004       0.059       0.299
==============================================================================
Omnibus:                       19.484   Durbin-Watson:                   0.533
Prob(Omnibus):                  0.000   Jarque-Bera (JB):               24.625
Skew:                           1.072   Prob(JB):                     4.50e-06
Kurtosis:                       4.403   Cond. No.                     4.48e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.48e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-03 to 2017-01-31 and predicting on 2017-02-01 ----
Creating Time series models: 
The real value for gold_price on 2017-02-01 is 1210.0
The predicted value for gold_price on 2017-02-01 is 1189.8702368189324
The real value for eth_price on 2017-02-01 is 10.735000000000465
The predicted value for eth_price on 2017-02-01 is 11.38816559833023
The real value for pos_sent on 2017-02-01 is 0.1294325015994884
The predicted value for pos_sent on 2017-02-01 is 0.12446265451681146
The real value for neg_sent on 2017-02-01 is 0.055811473661761576
The predicted value for neg_sent on 2017-02-01 is 0.07407949605681498
The real value for tot_num_trans on 2017-02-01 is 350560.0
The predicted value for tot_num_trans on 2017-02-01 is 297561.519393831
The real value for unique_addr on 2017-02-01 is 599052.0
The predicted value for unique_addr on 2017-02-01 is 491752.5003970177
The real value for hash_rate on 2017-02-01 is 3750587.467326551
The predicted value for hash_rate on 2017-02-01 is 3085268.9895366626
The real value for mempool_trans on 2017-02-01 is 104770.0
The predicted value for mempool_trans on 2017-02-01 is 58497.52160568198
The real value for avg_trans_per_block on 2017-02-01 is 1825.833333333333
The predicted value for avg_trans_per_block on 2017-02-01 is 1844.9365915217177
The real value of mkt_price was [[ 979.703875]] on 2017-02-01
The predicted value of mkt_price was [[ 954.93609912]] on 2017-02-01

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.672
Model:                            OLS   Adj. R-squared:                  0.635
Method:                 Least Squares   F-statistic:                     18.21
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           4.74e-16
Time:                        16:19:07   Log-Likelihood:                -488.26
No. Observations:                  90   AIC:                             996.5
Df Residuals:                      80   BIC:                             1022.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1254.4909    478.420      2.622      0.010     302.404    2206.578
x1            -1.2676      0.345     -3.676      0.000      -1.954      -0.581
x2            30.6529     11.592      2.644      0.010       7.585      53.721
x3          2441.8112    874.358      2.793      0.007     701.784    4181.838
x4          -789.5994    962.073     -0.821      0.414   -2704.186    1124.988
x5            -0.0004      0.001     -0.664      0.509      -0.002       0.001
x6         -5.422e-05      0.000     -0.137      0.891      -0.001       0.001
x7             0.0001   3.12e-05      4.803      0.000    8.78e-05       0.000
x8            -0.0004      0.000     -2.153      0.034      -0.001   -2.86e-05
x9             0.1745      0.059      2.959      0.004       0.057       0.292
==============================================================================
Omnibus:                       19.214   Durbin-Watson:                   0.546
Prob(Omnibus):                  0.000   Jarque-Bera (JB):               24.194
Skew:                           1.059   Prob(JB):                     5.58e-06
Kurtosis:                       4.401   Cond. No.                     4.53e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.53e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-04 to 2017-02-01 and predicting on 2017-02-02 ----
Creating Time series models: 
The real value for gold_price on 2017-02-02 is 1224.0499999998854
The predicted value for gold_price on 2017-02-02 is 1188.8153391394376
The real value for eth_price on 2017-02-02 is 10.779999999999257
The predicted value for eth_price on 2017-02-02 is 11.492983877627266
The real value for pos_sent on 2017-02-02 is 0.1433531321619556
The predicted value for pos_sent on 2017-02-02 is 0.13188210955356286
The real value for neg_sent on 2017-02-02 is 0.07207448433919024
The predicted value for neg_sent on 2017-02-02 is 0.07201549960538375
The real value for tot_num_trans on 2017-02-02 is 259559.0
The predicted value for tot_num_trans on 2017-02-02 is 282514.0160699174
The real value for unique_addr on 2017-02-02 is 452585.0
The predicted value for unique_addr on 2017-02-02 is 481109.893161246
The real value for hash_rate on 2017-02-02 is 3066886.6269284827
The predicted value for hash_rate on 2017-02-02 is 3003713.6311950926
The real value for mempool_trans on 2017-02-02 is 56515.5
The predicted value for mempool_trans on 2017-02-02 is 53611.64179720073
The real value for avg_trans_per_block on 2017-02-02 is 1653.2420382165603
The predicted value for avg_trans_per_block on 2017-02-02 is 1919.793764288965
The real value of mkt_price was [[ 1007.6137125]] on 2017-02-02
The predicted value of mkt_price was [[ 988.81327802]] on 2017-02-02

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.676
Model:                            OLS   Adj. R-squared:                  0.640
Method:                 Least Squares   F-statistic:                     18.58
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           2.80e-16
Time:                        16:19:18   Log-Likelihood:                -487.94
No. Observations:                  90   AIC:                             995.9
Df Residuals:                      80   BIC:                             1021.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1311.8964    466.154      2.814      0.006     384.220    2239.573
x1            -1.3241      0.342     -3.872      0.000      -2.005      -0.644
x2            31.7035     11.449      2.769      0.007       8.920      54.487
x3          2475.9545    872.183      2.839      0.006     740.255    4211.654
x4          -908.1279    964.411     -0.942      0.349   -2827.367    1011.111
x5            -0.0005      0.001     -0.776      0.440      -0.002       0.001
x6         -1.476e-05      0.000     -0.038      0.970      -0.001       0.001
x7             0.0001   2.94e-05      5.072      0.000    9.07e-05       0.000
x8            -0.0004      0.000     -2.078      0.041      -0.001   -1.54e-05
x9             0.1764      0.059      3.012      0.003       0.060       0.293
==============================================================================
Omnibus:                       18.692   Durbin-Watson:                   0.554
Prob(Omnibus):                  0.000   Jarque-Bera (JB):               23.255
Skew:                           1.041   Prob(JB):                     8.92e-06
Kurtosis:                       4.368   Cond. No.                     4.60e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.6e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-05 to 2017-02-02 and predicting on 2017-02-03 ----
Creating Time series models: 
The real value for gold_price on 2017-02-03 is 1213.0499999998895
The predicted value for gold_price on 2017-02-03 is 1191.4987771477267
The real value for eth_price on 2017-02-03 is 10.959999999999262
The predicted value for eth_price on 2017-02-03 is 11.45075490057764
The real value for pos_sent on 2017-02-03 is 0.14119209265175744
The predicted value for pos_sent on 2017-02-03 is 0.1339665462629359
The real value for neg_sent on 2017-02-03 is 0.06476737220447282
The predicted value for neg_sent on 2017-02-03 is 0.0659917945638754
The real value for tot_num_trans on 2017-02-03 is 323563.0
The predicted value for tot_num_trans on 2017-02-03 is 274621.5470219509
The real value for unique_addr on 2017-02-03 is 560374.0
The predicted value for unique_addr on 2017-02-03 is 464913.8249098099
The real value for hash_rate on 2017-02-03 is 2910612.1491232086
The predicted value for hash_rate on 2017-02-03 is 3172480.2817198285
The real value for mempool_trans on 2017-02-03 is 67658.83333333333
The predicted value for mempool_trans on 2017-02-03 is 57483.328148777335
The real value for avg_trans_per_block on 2017-02-03 is 2171.5637583892617
The predicted value for avg_trans_per_block on 2017-02-03 is 1724.3294573087444
The real value of mkt_price was [[ 1013.027]] on 2017-02-03
The predicted value of mkt_price was [[ 993.16408312]] on 2017-02-03

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.676
Model:                            OLS   Adj. R-squared:                  0.640
Method:                 Least Squares   F-statistic:                     18.56
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           2.89e-16
Time:                        16:19:28   Log-Likelihood:                -488.92
No. Observations:                  90   AIC:                             997.8
Df Residuals:                      80   BIC:                             1023.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1219.9875    484.829      2.516      0.014     255.147    2184.828
x1            -1.2813      0.357     -3.588      0.001      -1.992      -0.571
x2            30.8555     11.647      2.649      0.010       7.677      54.034
x3          2651.7658    874.547      3.032      0.003     911.361    4392.170
x4          -755.5287    978.296     -0.772      0.442   -2702.399    1191.342
x5            -0.0005      0.001     -0.745      0.459      -0.002       0.001
x6         -3.818e-05      0.000     -0.097      0.923      -0.001       0.001
x7             0.0002   2.93e-05      5.367      0.000    9.88e-05       0.000
x8            -0.0004      0.000     -2.055      0.043      -0.001   -1.15e-05
x9             0.1804      0.062      2.910      0.005       0.057       0.304
==============================================================================
Omnibus:                       18.056   Durbin-Watson:                   0.595
Prob(Omnibus):                  0.000   Jarque-Bera (JB):               22.046
Skew:                           1.023   Prob(JB):                     1.63e-05
Kurtosis:                       4.301   Cond. No.                     4.63e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.63e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-06 to 2017-02-03 and predicting on 2017-02-04 ----
Creating Time series models: 
The real value for gold_price on 2017-02-04 is 1213.0499999998917
The predicted value for gold_price on 2017-02-04 is 1198.5661865535658
The real value for eth_price on 2017-02-04 is 11.27999999999927
The predicted value for eth_price on 2017-02-04 is 11.40087206466676
The real value for pos_sent on 2017-02-04 is 0.14581487569344584
The predicted value for pos_sent on 2017-02-04 is 0.1377773320663873
The real value for neg_sent on 2017-02-04 is 0.07151818368604897
The predicted value for neg_sent on 2017-02-04 is 0.06645992785838523
The real value for tot_num_trans on 2017-02-04 is 286326.0
The predicted value for tot_num_trans on 2017-02-04 is 261341.58228864297
The real value for unique_addr on 2017-02-04 is 500098.0
The predicted value for unique_addr on 2017-02-04 is 469502.0801759702
The real value for hash_rate on 2017-02-04 is 3043000.8428698913
The predicted value for hash_rate on 2017-02-04 is 3131865.8603586615
The real value for mempool_trans on 2017-02-04 is 306620.5
The predicted value for mempool_trans on 2017-02-04 is 50803.078614685866
The real value for avg_trans_per_block on 2017-02-04 is 1974.6620689655167
The predicted value for avg_trans_per_block on 2017-02-04 is 1652.0857070953557
The real value of mkt_price was [[ 1030.9994125]] on 2017-02-04
The predicted value of mkt_price was [[ 983.81957079]] on 2017-02-04

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.683
Model:                            OLS   Adj. R-squared:                  0.647
Method:                 Least Squares   F-statistic:                     19.14
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           1.29e-16
Time:                        16:19:39   Log-Likelihood:                -488.90
No. Observations:                  90   AIC:                             997.8
Df Residuals:                      80   BIC:                             1023.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1183.3188    470.890      2.513      0.014     246.219    2120.419
x1            -1.2905      0.359     -3.593      0.001      -2.005      -0.576
x2            31.1175     11.666      2.667      0.009       7.901      54.334
x3          2750.5597    850.449      3.234      0.002    1058.113    4443.007
x4          -685.7178    978.841     -0.701      0.486   -2633.674    1262.239
x5            -0.0005      0.001     -0.872      0.386      -0.002       0.001
x6          2.349e-05      0.000      0.060      0.952      -0.001       0.001
x7             0.0002   2.86e-05      5.597      0.000       0.000       0.000
x8            -0.0004      0.000     -2.104      0.039      -0.001   -2.02e-05
x9             0.1863      0.060      3.093      0.003       0.066       0.306
==============================================================================
Omnibus:                       17.427   Durbin-Watson:                   0.608
Prob(Omnibus):                  0.000   Jarque-Bera (JB):               20.935
Skew:                           1.002   Prob(JB):                     2.85e-05
Kurtosis:                       4.251   Cond. No.                     4.58e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.58e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-07 to 2017-02-04 and predicting on 2017-02-05 ----
Creating Time series models: 
The real value for gold_price on 2017-02-05 is 1213.049999999891
The predicted value for gold_price on 2017-02-05 is 1201.537061670478
The real value for eth_price on 2017-02-05 is 11.3949999999989
The predicted value for eth_price on 2017-02-05 is 11.455074255622865
The real value for pos_sent on 2017-02-05 is 0.14154602980200065
The predicted value for pos_sent on 2017-02-05 is 0.1335335613376216
The real value for neg_sent on 2017-02-05 is 0.07556256378852823
The predicted value for neg_sent on 2017-02-05 is 0.06581842653497204
The real value for tot_num_trans on 2017-02-05 is 292101.0
The predicted value for tot_num_trans on 2017-02-05 is 249085.57874815125
The real value for unique_addr on 2017-02-05 is 510277.0
The predicted value for unique_addr on 2017-02-05 is 442071.9611441848
The real value for hash_rate on 2017-02-05 is 3336807.8208021554
The predicted value for hash_rate on 2017-02-05 is 3219645.6315634246
The real value for mempool_trans on 2017-02-05 is 83431.5
The predicted value for mempool_trans on 2017-02-05 is 34325.944461652296
The real value for avg_trans_per_block on 2017-02-05 is 1837.1132075471694
The predicted value for avg_trans_per_block on 2017-02-05 is 1568.8658356712233
The real value of mkt_price was [[ 1014.837725]] on 2017-02-05
The predicted value of mkt_price was [[ 984.56080747]] on 2017-02-05

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.691
Model:                            OLS   Adj. R-squared:                  0.656
Method:                 Least Squares   F-statistic:                     19.89
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           4.66e-17
Time:                        16:19:49   Log-Likelihood:                -488.94
No. Observations:                  90   AIC:                             997.9
Df Residuals:                      80   BIC:                             1023.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1157.3338    470.089      2.462      0.016     221.827    2092.841
x1            -1.3511      0.364     -3.713      0.000      -2.075      -0.627
x2            32.3158     11.713      2.759      0.007       9.006      55.626
x3          3078.3672    861.664      3.573      0.001    1363.602    4793.133
x4          -612.1325    985.452     -0.621      0.536   -2573.244    1348.979
x5            -0.0007      0.001     -1.186      0.239      -0.002       0.000
x6             0.0001      0.000      0.264      0.792      -0.001       0.001
x7             0.0002   2.83e-05      5.885      0.000       0.000       0.000
x8            -0.0003      0.000     -1.677      0.097      -0.001    4.76e-05
x9             0.2015      0.059      3.397      0.001       0.083       0.320
==============================================================================
Omnibus:                       16.505   Durbin-Watson:                   0.631
Prob(Omnibus):                  0.000   Jarque-Bera (JB):               19.644
Skew:                           0.952   Prob(JB):                     5.42e-05
Kurtosis:                       4.269   Cond. No.                     4.67e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.67e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-08 to 2017-02-05 and predicting on 2017-02-06 ----
Creating Time series models: 
The real value for gold_price on 2017-02-06 is 1221.8499999999156
The predicted value for gold_price on 2017-02-06 is 1201.7490166977643
The real value for eth_price on 2017-02-06 is 11.37000000000099
The predicted value for eth_price on 2017-02-06 is 11.538847854770735
The real value for pos_sent on 2017-02-06 is 0.12933792839364372
The predicted value for pos_sent on 2017-02-06 is 0.13501412798050988
The real value for neg_sent on 2017-02-06 is 0.06744380624162347
The predicted value for neg_sent on 2017-02-06 is 0.062415864134545605
The real value for tot_num_trans on 2017-02-06 is 230027.0
The predicted value for tot_num_trans on 2017-02-06 is 280518.3618128884
The real value for unique_addr on 2017-02-06 is 440409.0
The predicted value for unique_addr on 2017-02-06 is 493152.60007159726
The real value for hash_rate on 2017-02-06 is 2623276.58868094
The predicted value for hash_rate on 2017-02-06 is 3244732.6029567956
The real value for mempool_trans on 2017-02-06 is 40363.0
The predicted value for mempool_trans on 2017-02-06 is 41846.04074731837
The real value for avg_trans_per_block on 2017-02-06 is 1840.2160000000001
The predicted value for avg_trans_per_block on 2017-02-06 is 1728.0230331612756
The real value of mkt_price was [[ 1024.01375]] on 2017-02-06
The predicted value of mkt_price was [[ 1010.15458947]] on 2017-02-06

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.697
Model:                            OLS   Adj. R-squared:                  0.663
Method:                 Least Squares   F-statistic:                     20.46
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           2.21e-17
Time:                        16:19:59   Log-Likelihood:                -488.75
No. Observations:                  90   AIC:                             997.5
Df Residuals:                      80   BIC:                             1023.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1204.1598    461.360      2.610      0.011     286.024    2122.295
x1            -1.3961      0.368     -3.797      0.000      -2.128      -0.664
x2            32.8431     11.719      2.802      0.006       9.521      56.165
x3          3097.2229    844.956      3.666      0.000    1415.707    4778.739
x4          -574.1435    970.860     -0.591      0.556   -2506.216    1357.929
x5            -0.0007      0.001     -1.163      0.248      -0.002       0.000
x6             0.0001      0.000      0.271      0.787      -0.001       0.001
x7             0.0002   2.78e-05      5.968      0.000       0.000       0.000
x8            -0.0003      0.000     -1.660      0.101      -0.001    4.98e-05
x9             0.1971      0.059      3.342      0.001       0.080       0.315
==============================================================================
Omnibus:                       16.476   Durbin-Watson:                   0.657
Prob(Omnibus):                  0.000   Jarque-Bera (JB):               19.629
Skew:                           0.949   Prob(JB):                     5.46e-05
Kurtosis:                       4.277   Cond. No.                     4.59e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.59e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-09 to 2017-02-06 and predicting on 2017-02-07 ----
Creating Time series models: 
The real value for gold_price on 2017-02-07 is 1231.0
The predicted value for gold_price on 2017-02-07 is 1217.5606398853683
The real value for eth_price on 2017-02-07 is 11.474999999998982
The predicted value for eth_price on 2017-02-07 is 11.701568188961424
The real value for pos_sent on 2017-02-07 is 0.13516037904523523
The predicted value for pos_sent on 2017-02-07 is 0.13373812665351478
The real value for neg_sent on 2017-02-07 is 0.058521723583050196
The predicted value for neg_sent on 2017-02-07 is 0.06658390158165654
The real value for tot_num_trans on 2017-02-07 is 325268.0
The predicted value for tot_num_trans on 2017-02-07 is 293375.86560350994
The real value for unique_addr on 2017-02-07 is 557390.0
The predicted value for unique_addr on 2017-02-07 is 505727.2593342647
The real value for hash_rate on 2017-02-07 is 3168918.1191265755
The predicted value for hash_rate on 2017-02-07 is 3113577.2653616993
The real value for mempool_trans on 2017-02-07 is 126099.0
The predicted value for mempool_trans on 2017-02-07 is 44713.74393071981
The real value for avg_trans_per_block on 2017-02-07 is 2154.092715231788
The predicted value for avg_trans_per_block on 2017-02-07 is 1902.8324530148684
The real value of mkt_price was [[ 1050.11]] on 2017-02-07
The predicted value of mkt_price was [[ 998.78996259]] on 2017-02-07

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.690
Model:                            OLS   Adj. R-squared:                  0.655
Method:                 Least Squares   F-statistic:                     19.75
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           5.64e-17
Time:                        16:20:10   Log-Likelihood:                -490.65
No. Observations:                  90   AIC:                             1001.
Df Residuals:                      80   BIC:                             1026.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1179.2948    474.362      2.486      0.015     235.284    2123.306
x1            -1.3659      0.380     -3.596      0.001      -2.122      -0.610
x2            33.2406     11.980      2.775      0.007       9.400      57.081
x3          2945.8833    859.155      3.429      0.001    1236.110    4655.656
x4          -614.6045   1004.916     -0.612      0.543   -2614.451    1385.242
x5            -0.0010      0.001     -1.684      0.096      -0.002       0.000
x6             0.0002      0.000      0.394      0.695      -0.001       0.001
x7             0.0002   2.81e-05      6.175      0.000       0.000       0.000
x8            -0.0003      0.000     -1.634      0.106      -0.001    5.49e-05
x9             0.2255      0.058      3.867      0.000       0.109       0.342
==============================================================================
Omnibus:                       15.353   Durbin-Watson:                   0.626
Prob(Omnibus):                  0.000   Jarque-Bera (JB):               17.571
Skew:                           0.925   Prob(JB):                     0.000153
Kurtosis:                       4.125   Cond. No.                     4.61e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.61e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-10 to 2017-02-07 and predicting on 2017-02-08 ----
Creating Time series models: 
The real value for gold_price on 2017-02-08 is 1235.5999999999187
The predicted value for gold_price on 2017-02-08 is 1222.7845695491146
The real value for eth_price on 2017-02-08 is 11.48500000000048
The predicted value for eth_price on 2017-02-08 is 11.920131746780498
The real value for pos_sent on 2017-02-08 is 0.12112255285826205
The predicted value for pos_sent on 2017-02-08 is 0.13427586542172162
The real value for neg_sent on 2017-02-08 is 0.06421534847298353
The predicted value for neg_sent on 2017-02-08 is 0.06530877972570348
The real value for tot_num_trans on 2017-02-08 is 308316.0
The predicted value for tot_num_trans on 2017-02-08 is 304602.8247344821
The real value for unique_addr on 2017-02-08 is 516490.0
The predicted value for unique_addr on 2017-02-08 is 511896.44594204356
The real value for hash_rate on 2017-02-08 is 3063987.0555793373
The predicted value for hash_rate on 2017-02-08 is 3318192.1027698983
The real value for mempool_trans on 2017-02-08 is 250372.0
The predicted value for mempool_trans on 2017-02-08 is 81228.80666520211
The real value for avg_trans_per_block on 2017-02-08 is 2111.7534246575337
The predicted value for avg_trans_per_block on 2017-02-08 is 1822.5486705957935
The real value of mkt_price was [[ 1052.3766125]] on 2017-02-08
The predicted value of mkt_price was [[ 1001.67875012]] on 2017-02-08

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.698
Model:                            OLS   Adj. R-squared:                  0.664
Method:                 Least Squares   F-statistic:                     20.58
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           1.89e-17
Time:                        16:20:20   Log-Likelihood:                -490.67
No. Observations:                  90   AIC:                             1001.
Df Residuals:                      80   BIC:                             1026.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1269.4936    515.781      2.461      0.016     243.057    2295.930
x1            -1.4360      0.409     -3.511      0.001      -2.250      -0.622
x2            34.1819     12.200      2.802      0.006       9.903      58.461
x3          2816.6145    885.780      3.180      0.002    1053.856    4579.373
x4          -810.7468   1005.120     -0.807      0.422   -2810.999    1189.506
x5            -0.0011      0.001     -1.827      0.071      -0.002    9.68e-05
x6             0.0002      0.000      0.494      0.623      -0.001       0.001
x7             0.0002   2.74e-05      6.452      0.000       0.000       0.000
x8            -0.0002      0.000     -1.459      0.148      -0.001    8.27e-05
x9             0.2293      0.057      3.994      0.000       0.115       0.344
==============================================================================
Omnibus:                       13.573   Durbin-Watson:                   0.633
Prob(Omnibus):                  0.001   Jarque-Bera (JB):               14.812
Skew:                           0.864   Prob(JB):                     0.000608
Kurtosis:                       3.981   Cond. No.                     4.77e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.77e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-11 to 2017-02-08 and predicting on 2017-02-09 ----
Creating Time series models: 
The real value for gold_price on 2017-02-09 is 1241.75
The predicted value for gold_price on 2017-02-09 is 1217.9835770318562
The real value for eth_price on 2017-02-09 is 11.235000000000479
The predicted value for eth_price on 2017-02-09 is 11.929466732783517
The real value for pos_sent on 2017-02-09 is 0.12047519783825551
The predicted value for pos_sent on 2017-02-09 is 0.13873007319333472
The real value for neg_sent on 2017-02-09 is 0.0929992279482725
The predicted value for neg_sent on 2017-02-09 is 0.06603732117329246
The real value for tot_num_trans on 2017-02-09 is 346544.0
The predicted value for tot_num_trans on 2017-02-09 is 282541.23219745717
The real value for unique_addr on 2017-02-09 is 575007.0
The predicted value for unique_addr on 2017-02-09 is 474092.53151871485
The real value for hash_rate on 2017-02-09 is 3546669.9478966305
The predicted value for hash_rate on 2017-02-09 is 3253099.9623809815
The real value for mempool_trans on 2017-02-09 is 170184.5
The predicted value for mempool_trans on 2017-02-09 is 72944.22521425379
The real value for avg_trans_per_block on 2017-02-09 is 2050.5562130177514
The predicted value for avg_trans_per_block on 2017-02-09 is 1737.422301989725
The real value of mkt_price was [[ 976.103]] on 2017-02-09
The predicted value of mkt_price was [[ 1011.19947387]] on 2017-02-09

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.691
Model:                            OLS   Adj. R-squared:                  0.656
Method:                 Least Squares   F-statistic:                     19.88
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           4.74e-17
Time:                        16:20:30   Log-Likelihood:                -492.86
No. Observations:                  90   AIC:                             1006.
Df Residuals:                      80   BIC:                             1031.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1335.9599    552.888      2.416      0.018     235.678    2436.242
x1            -1.3996      0.436     -3.212      0.002      -2.267      -0.532
x2            32.0775     12.523      2.561      0.012       7.155      57.000
x3          2339.6876    901.252      2.596      0.011     546.139    4133.236
x4         -1289.0424   1003.490     -1.285      0.203   -3286.052     707.967
x5            -0.0012      0.001     -2.033      0.045      -0.002    -2.6e-05
x6             0.0002      0.000      0.415      0.679      -0.001       0.001
x7             0.0002   2.77e-05      6.711      0.000       0.000       0.000
x8            -0.0001      0.000     -0.706      0.482      -0.000       0.000
x9             0.2482      0.058      4.268      0.000       0.132       0.364
==============================================================================
Omnibus:                       13.024   Durbin-Watson:                   0.575
Prob(Omnibus):                  0.001   Jarque-Bera (JB):               13.905
Skew:                           0.862   Prob(JB):                     0.000956
Kurtosis:                       3.857   Cond. No.                     4.69e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.69e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-12 to 2017-02-09 and predicting on 2017-02-10 ----
Creating Time series models: 
The real value for gold_price on 2017-02-10 is 1225.75
The predicted value for gold_price on 2017-02-10 is 1222.9287298429545
The real value for eth_price on 2017-02-10 is 11.16000000000126
The predicted value for eth_price on 2017-02-10 is 11.763190719201766
The real value for pos_sent on 2017-02-10 is 0.12421585903083711
The predicted value for pos_sent on 2017-02-10 is 0.13336901293534748
The real value for neg_sent on 2017-02-10 is 0.07764547933710933
The predicted value for neg_sent on 2017-02-10 is 0.07161976936967031
The real value for tot_num_trans on 2017-02-10 is 304456.0
The predicted value for tot_num_trans on 2017-02-10 is 277212.2727478497
The real value for unique_addr on 2017-02-10 is 507435.0
The predicted value for unique_addr on 2017-02-10 is 487543.91582922294
The real value for hash_rate on 2017-02-10 is 3105959.480998233
The predicted value for hash_rate on 2017-02-10 is 3302293.155281244
The real value for mempool_trans on 2017-02-10 is 96694.25
The predicted value for mempool_trans on 2017-02-10 is 69623.44118707502
The real value for avg_trans_per_block on 2017-02-10 is 2057.135135135135
The predicted value for avg_trans_per_block on 2017-02-10 is 1705.9863428396195
The real value of mkt_price was [[ 999.1035]] on 2017-02-10
The predicted value of mkt_price was [[ 992.66709451]] on 2017-02-10

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.689
Model:                            OLS   Adj. R-squared:                  0.654
Method:                 Least Squares   F-statistic:                     19.70
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           6.08e-17
Time:                        16:20:41   Log-Likelihood:                -493.02
No. Observations:                  90   AIC:                             1006.
Df Residuals:                      80   BIC:                             1031.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1230.1268    534.991      2.299      0.024     165.460    2294.793
x1            -1.3264      0.427     -3.106      0.003      -2.176      -0.476
x2            30.4427     12.183      2.499      0.015       6.199      54.687
x3          2393.3748    920.790      2.599      0.011     560.944    4225.805
x4         -1111.6953    949.157     -1.171      0.245   -3000.579     777.188
x5            -0.0012      0.001     -1.991      0.050      -0.002   -5.62e-07
x6             0.0002      0.000      0.421      0.675      -0.001       0.001
x7             0.0002    2.7e-05      7.005      0.000       0.000       0.000
x8            -0.0001      0.000     -0.737      0.463      -0.000       0.000
x9             0.2511      0.058      4.320      0.000       0.135       0.367
==============================================================================
Omnibus:                       13.917   Durbin-Watson:                   0.586
Prob(Omnibus):                  0.001   Jarque-Bera (JB):               15.218
Skew:                           0.890   Prob(JB):                     0.000496
Kurtosis:                       3.942   Cond. No.                     4.57e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.57e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-13 to 2017-02-10 and predicting on 2017-02-11 ----
Creating Time series models: 
The real value for gold_price on 2017-02-11 is 1225.75
The predicted value for gold_price on 2017-02-11 is 1229.42840879064
The real value for eth_price on 2017-02-11 is 11.31999999999974
The predicted value for eth_price on 2017-02-11 is 11.74101334430075
The real value for pos_sent on 2017-02-11 is 0.1391725534308213
The predicted value for pos_sent on 2017-02-11 is 0.1411884571624364
The real value for neg_sent on 2017-02-11 is 0.07083509561304824
The predicted value for neg_sent on 2017-02-11 is 0.06832626416907624
The real value for tot_num_trans on 2017-02-11 is 262686.0
The predicted value for tot_num_trans on 2017-02-11 is 262995.29412571224
The real value for unique_addr on 2017-02-11 is 474676.0
The predicted value for unique_addr on 2017-02-11 is 471893.9431816743
The real value for hash_rate on 2017-02-11 is 2917083.566613205
The predicted value for hash_rate on 2017-02-11 is 3167332.014851758
The real value for mempool_trans on 2017-02-11 is 53025.75
The predicted value for mempool_trans on 2017-02-11 is 91303.69370617386
The real value for avg_trans_per_block on 2017-02-11 is 1889.8273381294964
The predicted value for avg_trans_per_block on 2017-02-11 is 1720.2243466513605
The real value of mkt_price was [[ 1008.8466625]] on 2017-02-11
The predicted value of mkt_price was [[ 996.74010998]] on 2017-02-11

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.689
Model:                            OLS   Adj. R-squared:                  0.654
Method:                 Least Squares   F-statistic:                     19.72
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           5.89e-17
Time:                        16:20:51   Log-Likelihood:                -492.90
No. Observations:                  90   AIC:                             1006.
Df Residuals:                      80   BIC:                             1031.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1266.6511    540.031      2.346      0.021     191.955    2341.347
x1            -1.3569      0.431     -3.145      0.002      -2.216      -0.498
x2            30.1826     12.111      2.492      0.015       6.081      54.284
x3          2326.3923    923.907      2.518      0.014     487.758    4165.026
x4         -1170.6672    955.541     -1.225      0.224   -3072.255     730.920
x5            -0.0012      0.001     -1.974      0.052      -0.002    1.01e-05
x6             0.0001      0.000      0.285      0.776      -0.001       0.001
x7             0.0002   2.67e-05      7.311      0.000       0.000       0.000
x8            -0.0001      0.000     -0.747      0.457      -0.000       0.000
x9             0.2645      0.058      4.588      0.000       0.150       0.379
==============================================================================
Omnibus:                       12.939   Durbin-Watson:                   0.594
Prob(Omnibus):                  0.002   Jarque-Bera (JB):               13.849
Skew:                           0.847   Prob(JB):                     0.000983
Kurtosis:                       3.907   Cond. No.                     4.65e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.65e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-14 to 2017-02-11 and predicting on 2017-02-12 ----
Creating Time series models: 
The real value for gold_price on 2017-02-12 is 1225.75
The predicted value for gold_price on 2017-02-12 is 1227.7273835664598
The real value for eth_price on 2017-02-12 is 11.375
The predicted value for eth_price on 2017-02-12 is 11.76132681023095
The real value for pos_sent on 2017-02-12 is 0.13607000234907213
The predicted value for pos_sent on 2017-02-12 is 0.1321610576035094
The real value for neg_sent on 2017-02-12 is 0.06214752172891703
The predicted value for neg_sent on 2017-02-12 is 0.06634154206031238
The real value for tot_num_trans on 2017-02-12 is 238182.0
The predicted value for tot_num_trans on 2017-02-12 is 251894.00850893883
The real value for unique_addr on 2017-02-12 is 422084.0
The predicted value for unique_addr on 2017-02-12 is 450260.9991751898
The real value for hash_rate on 2017-02-12 is 2959055.9920321
The predicted value for hash_rate on 2017-02-12 is 3393499.477670781
The real value for mempool_trans on 2017-02-12 is 43050.0
The predicted value for mempool_trans on 2017-02-12 is 49537.27477769497
The real value for avg_trans_per_block on 2017-02-12 is 1689.2340425531916
The predicted value for avg_trans_per_block on 2017-02-12 is 1509.0720532751457
The real value of mkt_price was [[ 1000.604625]] on 2017-02-12
The predicted value of mkt_price was [[ 985.61294201]] on 2017-02-12

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.690
Model:                            OLS   Adj. R-squared:                  0.655
Method:                 Least Squares   F-statistic:                     19.74
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           5.69e-17
Time:                        16:21:01   Log-Likelihood:                -492.79
No. Observations:                  90   AIC:                             1006.
Df Residuals:                      80   BIC:                             1031.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1323.0646    554.175      2.387      0.019     220.221    2425.908
x1            -1.4204      0.450     -3.153      0.002      -2.317      -0.524
x2            30.8945     12.178      2.537      0.013       6.659      55.130
x3          2285.3655    927.115      2.465      0.016     440.347    4130.383
x4         -1163.7319    950.789     -1.224      0.225   -3055.863     728.399
x5            -0.0013      0.001     -2.080      0.041      -0.003   -5.59e-05
x6             0.0001      0.000      0.363      0.718      -0.001       0.001
x7             0.0002   2.66e-05      7.468      0.000       0.000       0.000
x8            -0.0001      0.000     -0.757      0.451      -0.000       0.000
x9             0.2696      0.057      4.767      0.000       0.157       0.382
==============================================================================
Omnibus:                       12.192   Durbin-Watson:                   0.599
Prob(Omnibus):                  0.002   Jarque-Bera (JB):               12.822
Skew:                           0.816   Prob(JB):                      0.00164
Kurtosis:                       3.871   Cond. No.                     4.67e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.67e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-15 to 2017-02-12 and predicting on 2017-02-13 ----
Creating Time series models: 
The real value for gold_price on 2017-02-13 is 1229.4000000000847
The predicted value for gold_price on 2017-02-13 is 1234.4084563287872
The real value for eth_price on 2017-02-13 is 11.324999999998997
The predicted value for eth_price on 2017-02-13 is 11.784065172147589
The real value for pos_sent on 2017-02-13 is 0.12478775038520834
The predicted value for pos_sent on 2017-02-13 is 0.1329142177335163
The real value for neg_sent on 2017-02-13 is 0.0589857473035439
The predicted value for neg_sent on 2017-02-13 is 0.06651401012926303
The real value for tot_num_trans on 2017-02-13 is 288290.0
The predicted value for tot_num_trans on 2017-02-13 is 265774.2146505417
The real value for unique_addr on 2017-02-13 is 503160.0
The predicted value for unique_addr on 2017-02-13 is 489550.68942486023
The real value for hash_rate on 2017-02-13 is 3420752.671639945
The predicted value for hash_rate on 2017-02-13 is 3137829.123915912
The real value for mempool_trans on 2017-02-13 is 19444.5
The predicted value for mempool_trans on 2017-02-13 is 33004.27058155562
The real value for avg_trans_per_block on 2017-02-13 is 1768.6503067484662
The predicted value for avg_trans_per_block on 2017-02-13 is 1723.381611663538
The real value of mkt_price was [[ 999.877375]] on 2017-02-13
The predicted value of mkt_price was [[ 977.05780287]] on 2017-02-13

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.687
Model:                            OLS   Adj. R-squared:                  0.652
Method:                 Least Squares   F-statistic:                     19.52
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           7.74e-17
Time:                        16:21:12   Log-Likelihood:                -492.95
No. Observations:                  90   AIC:                             1006.
Df Residuals:                      80   BIC:                             1031.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1307.8006    554.937      2.357      0.021     203.442    2412.160
x1            -1.4001      0.450     -3.114      0.003      -2.295      -0.505
x2            30.7963     12.244      2.515      0.014       6.431      55.162
x3          2278.4327    934.760      2.437      0.017     418.201    4138.665
x4         -1201.0636    967.130     -1.242      0.218   -3125.714     723.586
x5            -0.0013      0.001     -2.079      0.041      -0.003    -5.5e-05
x6             0.0001      0.000      0.301      0.764      -0.001       0.001
x7             0.0002    2.7e-05      7.413      0.000       0.000       0.000
x8            -0.0001      0.000     -0.776      0.440      -0.000       0.000
x9             0.2732      0.058      4.749      0.000       0.159       0.388
==============================================================================
Omnibus:                       11.678   Durbin-Watson:                   0.591
Prob(Omnibus):                  0.003   Jarque-Bera (JB):               12.136
Skew:                           0.792   Prob(JB):                      0.00232
Kurtosis:                       3.851   Cond. No.                     4.76e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.76e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-16 to 2017-02-13 and predicting on 2017-02-14 ----
Creating Time series models: 
The real value for gold_price on 2017-02-14 is 1229.6500000000815
The predicted value for gold_price on 2017-02-14 is 1240.1845587927392
The real value for eth_price on 2017-02-14 is 12.16000000000159
The predicted value for eth_price on 2017-02-14 is 11.973016690800781
The real value for pos_sent on 2017-02-14 is 0.13862701966128066
The predicted value for pos_sent on 2017-02-14 is 0.1284749280158903
The real value for neg_sent on 2017-02-14 is 0.0641516449289469
The predicted value for neg_sent on 2017-02-14 is 0.062465288713755986
The real value for tot_num_trans on 2017-02-14 is 315084.0
The predicted value for tot_num_trans on 2017-02-14 is 303649.3179385784
The real value for unique_addr on 2017-02-14 is 530239.0
The predicted value for unique_addr on 2017-02-14 is 515500.3358781535
The real value for hash_rate on 2017-02-14 is 3546669.9478966305
The predicted value for hash_rate on 2017-02-14 is 3333903.292610822
The real value for mempool_trans on 2017-02-14 is 28684.0
The predicted value for mempool_trans on 2017-02-14 is 51307.197991809386
The real value for avg_trans_per_block on 2017-02-14 is 1864.4023668639054
The predicted value for avg_trans_per_block on 2017-02-14 is 1849.1441812620203
The real value of mkt_price was [[ 1011.78025]] on 2017-02-14
The predicted value of mkt_price was [[ 995.08364038]] on 2017-02-14

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.685
Model:                            OLS   Adj. R-squared:                  0.650
Method:                 Least Squares   F-statistic:                     19.37
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           9.51e-17
Time:                        16:21:22   Log-Likelihood:                -492.94
No. Observations:                  90   AIC:                             1006.
Df Residuals:                      80   BIC:                             1031.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1306.3495    556.402      2.348      0.021     199.075    2413.624
x1            -1.3998      0.450     -3.113      0.003      -2.295      -0.505
x2            30.9112     12.225      2.529      0.013       6.583      55.240
x3          2286.0598    931.318      2.455      0.016     432.679    4139.441
x4         -1196.3410    960.687     -1.245      0.217   -3108.170     715.488
x5            -0.0013      0.001     -2.079      0.041      -0.003   -5.53e-05
x6             0.0001      0.000      0.307      0.760      -0.001       0.001
x7             0.0002   2.71e-05      7.360      0.000       0.000       0.000
x8            -0.0001      0.000     -0.776      0.440      -0.000       0.000
x9             0.2722      0.058      4.701      0.000       0.157       0.387
==============================================================================
Omnibus:                       11.695   Durbin-Watson:                   0.593
Prob(Omnibus):                  0.003   Jarque-Bera (JB):               12.153
Skew:                           0.794   Prob(JB):                      0.00230
Kurtosis:                       3.847   Cond. No.                     4.76e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.76e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-17 to 2017-02-14 and predicting on 2017-02-15 ----
Creating Time series models: 
The real value for gold_price on 2017-02-15 is 1225.150000000062
The predicted value for gold_price on 2017-02-15 is 1248.2139176260546
The real value for eth_price on 2017-02-15 is 12.980000000001331
The predicted value for eth_price on 2017-02-15 is 12.199115385063644
The real value for pos_sent on 2017-02-15 is 0.12774146139515621
The predicted value for pos_sent on 2017-02-15 is 0.12841452075146514
The real value for neg_sent on 2017-02-15 is 0.06009190643759044
The predicted value for neg_sent on 2017-02-15 is 0.06446054864730294
The real value for tot_num_trans on 2017-02-15 is 294786.0
The predicted value for tot_num_trans on 2017-02-15 is 304874.246664002
The real value for unique_addr on 2017-02-15 is 495896.0
The predicted value for unique_addr on 2017-02-15 is 515322.3099670253
The real value for hash_rate on 2017-02-15 is 2917083.566613205
The predicted value for hash_rate on 2017-02-15 is 3365223.6337515353
The real value for mempool_trans on 2017-02-15 is 35245.75
The predicted value for mempool_trans on 2017-02-15 is 93579.07487072317
The real value for avg_trans_per_block on 2017-02-15 is 2120.7625899280574
The predicted value for avg_trans_per_block on 2017-02-15 is 1853.402917895418
The real value of mkt_price was [[ 1012.3259875]] on 2017-02-15
The predicted value of mkt_price was [[ 984.38389444]] on 2017-02-15

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.684
Model:                            OLS   Adj. R-squared:                  0.648
Method:                 Least Squares   F-statistic:                     19.20
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           1.19e-16
Time:                        16:21:32   Log-Likelihood:                -493.38
No. Observations:                  90   AIC:                             1007.
Df Residuals:                      80   BIC:                             1032.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1388.8034    556.570      2.495      0.015     281.194    2496.413
x1            -1.4353      0.454     -3.160      0.002      -2.339      -0.532
x2            30.1193     12.315      2.446      0.017       5.611      54.627
x3          2120.2950    923.567      2.296      0.024     282.338    3958.252
x4         -1218.5926    966.225     -1.261      0.211   -3141.441     704.256
x5            -0.0014      0.001     -2.148      0.035      -0.003      -0.000
x6             0.0001      0.000      0.340      0.735      -0.001       0.001
x7             0.0002   2.82e-05      7.070      0.000       0.000       0.000
x8         -8.493e-05      0.000     -0.583      0.562      -0.000       0.000
x9             0.2750      0.060      4.567      0.000       0.155       0.395
==============================================================================
Omnibus:                       12.232   Durbin-Watson:                   0.568
Prob(Omnibus):                  0.002   Jarque-Bera (JB):               12.880
Skew:                           0.816   Prob(JB):                      0.00160
Kurtosis:                       3.876   Cond. No.                     4.75e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.75e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-18 to 2017-02-15 and predicting on 2017-02-16 ----
Creating Time series models: 
The real value for gold_price on 2017-02-16 is 1236.75
The predicted value for gold_price on 2017-02-16 is 1234.9856494949445
The real value for eth_price on 2017-02-16 is 12.910000000001563
The predicted value for eth_price on 2017-02-16 is 12.270495245566854
The real value for pos_sent on 2017-02-16 is 0.13962691259694007
The predicted value for pos_sent on 2017-02-16 is 0.13340775925050083
The real value for neg_sent on 2017-02-16 is 0.052306644309369155
The predicted value for neg_sent on 2017-02-16 is 0.06303227087393933
The real value for tot_num_trans on 2017-02-16 is 285329.0
The predicted value for tot_num_trans on 2017-02-16 is 292286.0793956967
The real value for unique_addr on 2017-02-16 is 489189.0
The predicted value for unique_addr on 2017-02-16 is 495696.30292128155
The real value for hash_rate on 2017-02-16 is 3189904.331836023
The predicted value for hash_rate on 2017-02-16 is 3376436.8448373787
The real value for mempool_trans on 2017-02-16 is 49404.75
The predicted value for mempool_trans on 2017-02-16 is 72275.70897395269
The real value for avg_trans_per_block on 2017-02-16 is 1877.1644736842104
The predicted value for avg_trans_per_block on 2017-02-16 is 1835.2789526073939
The real value of mkt_price was [[ 1035.208125]] on 2017-02-16
The predicted value of mkt_price was [[ 1028.54542906]] on 2017-02-16

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.684
Model:                            OLS   Adj. R-squared:                  0.649
Method:                 Least Squares   F-statistic:                     19.27
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           1.08e-16
Time:                        16:21:43   Log-Likelihood:                -493.38
No. Observations:                  90   AIC:                             1007.
Df Residuals:                      80   BIC:                             1032.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1396.0072    558.432      2.500      0.014     284.692    2507.322
x1            -1.4443      0.457     -3.160      0.002      -2.354      -0.535
x2            28.4235     11.948      2.379      0.020       4.646      52.201
x3          2110.8982    924.499      2.283      0.025     271.087    3950.710
x4         -1107.6111    967.013     -1.145      0.255   -3032.029     816.807
x5            -0.0014      0.001     -2.209      0.030      -0.003      -0.000
x6             0.0002      0.000      0.392      0.696      -0.001       0.001
x7             0.0002   2.92e-05      6.964      0.000       0.000       0.000
x8         -7.639e-05      0.000     -0.525      0.601      -0.000       0.000
x9             0.2790      0.061      4.543      0.000       0.157       0.401
==============================================================================
Omnibus:                       13.722   Durbin-Watson:                   0.546
Prob(Omnibus):                  0.001   Jarque-Bera (JB):               15.108
Skew:                           0.861   Prob(JB):                     0.000524
Kurtosis:                       4.030   Cond. No.                     4.74e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.74e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-19 to 2017-02-16 and predicting on 2017-02-17 ----
Creating Time series models: 
The real value for gold_price on 2017-02-17 is 1241.400000000035
The predicted value for gold_price on 2017-02-17 is 1234.3391385847397
The real value for eth_price on 2017-02-17 is 12.79000000000074
The predicted value for eth_price on 2017-02-17 is 12.371602448215759
The real value for pos_sent on 2017-02-17 is 0.14091070615034207
The predicted value for pos_sent on 2017-02-17 is 0.13593915200505755
The real value for neg_sent on 2017-02-17 is 0.05408769931662875
The predicted value for neg_sent on 2017-02-17 is 0.0657288716715314
The real value for tot_num_trans on 2017-02-17 is 319767.0
The predicted value for tot_num_trans on 2017-02-17 is 282750.54308849585
The real value for unique_addr on 2017-02-17 is 550072.0
The predicted value for unique_addr on 2017-02-17 is 493863.11530950456
The real value for hash_rate on 2017-02-17 is 3168918.1191265755
The predicted value for hash_rate on 2017-02-17 is 3297747.1576741175
The real value for mempool_trans on 2017-02-17 is 83308.0
The predicted value for mempool_trans on 2017-02-17 is 68292.15210257303
The real value for avg_trans_per_block on 2017-02-17 is 2117.662251655629
The predicted value for avg_trans_per_block on 2017-02-17 is 1854.1111137157168
The real value of mkt_price was [[ 1055.53685]] on 2017-02-17
The predicted value of mkt_price was [[ 1036.75093602]] on 2017-02-17

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.691
Model:                            OLS   Adj. R-squared:                  0.656
Method:                 Least Squares   F-statistic:                     19.86
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           4.87e-17
Time:                        16:21:53   Log-Likelihood:                -492.92
No. Observations:                  90   AIC:                             1006.
Df Residuals:                      80   BIC:                             1031.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1368.9969    556.246      2.461      0.016     262.033    2475.961
x1            -1.3984      0.457     -3.062      0.003      -2.307      -0.490
x2            28.4649     11.750      2.423      0.018       5.083      51.847
x3          2107.3565    917.695      2.296      0.024     281.086    3933.627
x4         -1335.0163    991.954     -1.346      0.182   -3309.068     639.035
x5            -0.0014      0.001     -2.148      0.035      -0.003      -0.000
x6             0.0002      0.000      0.395      0.694      -0.001       0.001
x7             0.0002   2.96e-05      6.695      0.000       0.000       0.000
x8         -7.206e-05      0.000     -0.498      0.620      -0.000       0.000
x9             0.2729      0.061      4.441      0.000       0.151       0.395
==============================================================================
Omnibus:                       13.484   Durbin-Watson:                   0.548
Prob(Omnibus):                  0.001   Jarque-Bera (JB):               14.695
Skew:                           0.859   Prob(JB):                     0.000644
Kurtosis:                       3.982   Cond. No.                     4.81e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.81e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-20 to 2017-02-17 and predicting on 2017-02-18 ----
Creating Time series models: 
The real value for gold_price on 2017-02-18 is 1241.399999999999
The predicted value for gold_price on 2017-02-18 is 1239.68401472713
The real value for eth_price on 2017-02-18 is 12.760000000000261
The predicted value for eth_price on 2017-02-18 is 12.452229411396504
The real value for pos_sent on 2017-02-18 is 0.14641145307769948
The predicted value for pos_sent on 2017-02-18 is 0.14084628180437464
The real value for neg_sent on 2017-02-18 is 0.057815085771947515
The predicted value for neg_sent on 2017-02-18 is 0.0657696700582197
The real value for tot_num_trans on 2017-02-18 is 288149.0
The predicted value for tot_num_trans on 2017-02-18 is 268496.6754313435
The real value for unique_addr on 2017-02-18 is 510385.0
The predicted value for unique_addr on 2017-02-18 is 475727.42162932764
The real value for hash_rate on 2017-02-18 is 3330515.7440737467
The predicted value for hash_rate on 2017-02-18 is 3196782.323225993
The real value for mempool_trans on 2017-02-18 is 109288.5
The predicted value for mempool_trans on 2017-02-18 is 87375.69918965909
The real value for avg_trans_per_block on 2017-02-18 is 1895.717105263158
The predicted value for avg_trans_per_block on 2017-02-18 is 1813.3483623358163
The real value of mkt_price was [[ 1056.6371375]] on 2017-02-18
The predicted value of mkt_price was [[ 1026.89119156]] on 2017-02-18

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.697
Model:                            OLS   Adj. R-squared:                  0.663
Method:                 Least Squares   F-statistic:                     20.48
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           2.15e-17
Time:                        16:22:05   Log-Likelihood:                -492.69
No. Observations:                  90   AIC:                             1005.
Df Residuals:                      80   BIC:                             1030.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1316.4499    560.350      2.349      0.021     201.318    2431.582
x1            -1.3569      0.460     -2.950      0.004      -2.272      -0.442
x2            28.5665     11.670      2.448      0.017       5.343      51.790
x3          2187.7253    919.998      2.378      0.020     356.871    4018.580
x4         -1327.7723    981.905     -1.352      0.180   -3281.825     626.281
x5            -0.0014      0.001     -2.180      0.032      -0.003      -0.000
x6             0.0002      0.000      0.469      0.640      -0.001       0.001
x7             0.0002   3.02e-05      6.434      0.000       0.000       0.000
x8         -7.654e-05      0.000     -0.530      0.598      -0.000       0.000
x9             0.2686      0.062      4.364      0.000       0.146       0.391
==============================================================================
Omnibus:                       13.861   Durbin-Watson:                   0.548
Prob(Omnibus):                  0.001   Jarque-Bera (JB):               15.255
Skew:                           0.873   Prob(JB):                     0.000487
Kurtosis:                       4.010   Cond. No.                     4.86e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.86e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-21 to 2017-02-18 and predicting on 2017-02-19 ----
Creating Time series models: 
The real value for gold_price on 2017-02-19 is 1241.4000000000037
The predicted value for gold_price on 2017-02-19 is 1244.0939358744595
The real value for eth_price on 2017-02-19 is 12.790000000000727
The predicted value for eth_price on 2017-02-19 is 12.52606587419646
The real value for pos_sent on 2017-02-19 is 0.13378336653386494
The predicted value for pos_sent on 2017-02-19 is 0.13417518212368543
The real value for neg_sent on 2017-02-19 is 0.05737773904382463
The predicted value for neg_sent on 2017-02-19 is 0.0637655018197072
The real value for tot_num_trans on 2017-02-19 is 251721.0
The predicted value for tot_num_trans on 2017-02-19 is 254111.70684565054
The real value for unique_addr on 2017-02-19 is 458998.0
The predicted value for unique_addr on 2017-02-19 is 452383.94390860526
The real value for hash_rate on 2017-02-19 is 2936112.5638544867
The predicted value for hash_rate on 2017-02-19 is 3406210.6177579034
The real value for mempool_trans on 2017-02-19 is 38492.5
The predicted value for mempool_trans on 2017-02-19 is 43119.98435863157
The real value for avg_trans_per_block on 2017-02-19 is 1878.5149253731345
The predicted value for avg_trans_per_block on 2017-02-19 is 1640.6301392255396
The real value of mkt_price was [[ 1052.77928571]] on 2017-02-19
The predicted value of mkt_price was [[ 1024.27941242]] on 2017-02-19

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.701
Model:                            OLS   Adj. R-squared:                  0.668
Method:                 Least Squares   F-statistic:                     20.87
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           1.29e-17
Time:                        16:22:17   Log-Likelihood:                -492.46
No. Observations:                  90   AIC:                             1005.
Df Residuals:                      80   BIC:                             1030.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1264.4869    564.924      2.238      0.028     140.253    2388.721
x1            -1.3009      0.466     -2.792      0.007      -2.228      -0.374
x2            28.2070     11.614      2.429      0.017       5.094      51.320
x3          2219.9623    914.496      2.428      0.017     400.058    4039.867
x4         -1276.4922    978.678     -1.304      0.196   -3224.123     671.138
x5            -0.0014      0.001     -2.117      0.037      -0.003   -8.21e-05
x6             0.0002      0.000      0.371      0.712      -0.001       0.001
x7             0.0002   3.06e-05      6.231      0.000       0.000       0.000
x8         -9.017e-05      0.000     -0.622      0.535      -0.000       0.000
x9             0.2687      0.061      4.377      0.000       0.146       0.391
==============================================================================
Omnibus:                       14.483   Durbin-Watson:                   0.537
Prob(Omnibus):                  0.001   Jarque-Bera (JB):               16.219
Skew:                           0.893   Prob(JB):                     0.000301
Kurtosis:                       4.065   Cond. No.                     4.91e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.91e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-22 to 2017-02-19 and predicting on 2017-02-20 ----
Creating Time series models: 
The real value for gold_price on 2017-02-20 is 1235.3499999999547
The predicted value for gold_price on 2017-02-20 is 1249.7391744591425
The real value for eth_price on 2017-02-20 is 12.599999999999035
The predicted value for eth_price on 2017-02-20 is 12.71813773058641
The real value for pos_sent on 2017-02-20 is 0.1352459881292591
The predicted value for pos_sent on 2017-02-20 is 0.13100583223519977
The real value for neg_sent on 2017-02-20 is 0.05263728291932286
The predicted value for neg_sent on 2017-02-20 is 0.06387564126992515
The real value for tot_num_trans on 2017-02-20 is 282728.0
The predicted value for tot_num_trans on 2017-02-20 is 267847.1313055606
The real value for unique_addr on 2017-02-20 is 503952.0
The predicted value for unique_addr on 2017-02-20 is 493904.96453395224
The real value for hash_rate on 2017-02-20 is 3242870.5929139117
The predicted value for hash_rate on 2017-02-20 is 3191024.7961319247
The real value for mempool_trans on 2017-02-20 is 45046.0
The predicted value for mempool_trans on 2017-02-20 is 33473.96008814016
The real value for avg_trans_per_block on 2017-02-20 is 1910.3243243243246
The predicted value for avg_trans_per_block on 2017-02-20 is 1894.654141264163
The real value of mkt_price was [[ 1084.7550125]] on 2017-02-20
The predicted value of mkt_price was [[ 1034.34281822]] on 2017-02-20

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.704
Model:                            OLS   Adj. R-squared:                  0.670
Method:                 Least Squares   F-statistic:                     21.10
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           9.63e-18
Time:                        16:22:28   Log-Likelihood:                -492.45
No. Observations:                  90   AIC:                             1005.
Df Residuals:                      80   BIC:                             1030.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1203.3989    577.787      2.083      0.040      53.566    2353.232
x1            -1.2419      0.480     -2.589      0.011      -2.197      -0.287
x2            28.3093     11.591      2.442      0.017       5.243      51.376
x3          2245.6788    916.894      2.449      0.017     421.002    4070.356
x4         -1324.1733    974.716     -1.359      0.178   -3263.920     615.573
x5            -0.0014      0.001     -2.102      0.039      -0.003   -7.25e-05
x6             0.0001      0.000      0.329      0.743      -0.001       0.001
x7             0.0002    3.1e-05      6.062      0.000       0.000       0.000
x8            -0.0001      0.000     -0.693      0.490      -0.000       0.000
x9             0.2710      0.061      4.433      0.000       0.149       0.393
==============================================================================
Omnibus:                       14.110   Durbin-Watson:                   0.538
Prob(Omnibus):                  0.001   Jarque-Bera (JB):               15.641
Skew:                           0.880   Prob(JB):                     0.000401
Kurtosis:                       4.034   Cond. No.                     4.93e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.93e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-23 to 2017-02-20 and predicting on 2017-02-21 ----
Creating Time series models: 
The real value for gold_price on 2017-02-21 is 1228.7000000001137
The predicted value for gold_price on 2017-02-21 is 1252.7901470999086
The real value for eth_price on 2017-02-21 is 12.50499999999888
The predicted value for eth_price on 2017-02-21 is 13.05797424406768
The real value for pos_sent on 2017-02-21 is 0.1353162078324935
The predicted value for pos_sent on 2017-02-21 is 0.13033217317122633
The real value for neg_sent on 2017-02-21 is 0.05788639007367194
The predicted value for neg_sent on 2017-02-21 is 0.06408810297600863
The real value for tot_num_trans on 2017-02-21 is 279090.0
The predicted value for tot_num_trans on 2017-02-21 is 299924.91812519566
The real value for unique_addr on 2017-02-21 is 487720.0
The predicted value for unique_addr on 2017-02-21 is 525310.2878852865
The real value for hash_rate on 2017-02-21 is 3023757.715014323
The predicted value for hash_rate on 2017-02-21 is 3369689.08544092
The real value for mempool_trans on 2017-02-21 is 93842.0
The predicted value for mempool_trans on 2017-02-21 is 93378.12022702458
The real value for avg_trans_per_block on 2017-02-21 is 2022.391304347826
The predicted value for avg_trans_per_block on 2017-02-21 is 1979.5645275680463
The real value of mkt_price was [[ 1123.78842857]] on 2017-02-21
The predicted value of mkt_price was [[ 1051.3290671]] on 2017-02-21

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.709
Model:                            OLS   Adj. R-squared:                  0.677
Method:                 Least Squares   F-statistic:                     21.69
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           4.51e-18
Time:                        16:22:39   Log-Likelihood:                -492.43
No. Observations:                  90   AIC:                             1005.
Df Residuals:                      80   BIC:                             1030.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1170.7635    587.057      1.994      0.050       2.482    2339.045
x1            -1.2121      0.489     -2.478      0.015      -2.186      -0.239
x2            28.3352     11.579      2.447      0.017       5.291      51.379
x3          2243.8555    916.600      2.448      0.017     419.763    4067.948
x4         -1361.5172    964.741     -1.411      0.162   -3281.414     558.379
x5            -0.0014      0.001     -2.147      0.035      -0.003      -0.000
x6             0.0002      0.000      0.376      0.708      -0.001       0.001
x7             0.0002   3.14e-05      5.928      0.000       0.000       0.000
x8            -0.0001      0.000     -0.714      0.477      -0.000       0.000
x9             0.2724      0.061      4.462      0.000       0.151       0.394
==============================================================================
Omnibus:                       13.877   Durbin-Watson:                   0.535
Prob(Omnibus):                  0.001   Jarque-Bera (JB):               15.282
Skew:                           0.873   Prob(JB):                     0.000480
Kurtosis:                       4.013   Cond. No.                     4.93e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.93e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

---- Running model from 2016-11-24 to 2017-02-21 and predicting on 2017-02-22 ----
Creating Time series models: 
The real value for gold_price on 2017-02-22 is 1237.5
The predicted value for gold_price on 2017-02-22 is 1243.133808798761
The real value for eth_price on 2017-02-22 is 12.64499999999882
The predicted value for eth_price on 2017-02-22 is 13.262367093330273
The real value for pos_sent on 2017-02-22 is 0.1297823529411765
The predicted value for pos_sent on 2017-02-22 is 0.13225691648660678
The real value for neg_sent on 2017-02-22 is 0.062284313725490155
The predicted value for neg_sent on 2017-02-22 is 0.06389608460641129
The real value for tot_num_trans on 2017-02-22 is 284024.0
The predicted value for tot_num_trans on 2017-02-22 is 325586.05845974287
The real value for unique_addr on 2017-02-22 is 489247.0
The predicted value for unique_addr on 2017-02-22 is 517499.1792197267
The real value for hash_rate on 2017-02-22 is 3045669.0028042817
The predicted value for hash_rate on 2017-02-22 is 3442911.116577757
The real value for mempool_trans on 2017-02-22 is 229240.5
The predicted value for mempool_trans on 2017-02-22 is 139660.40170705027
The real value for avg_trans_per_block on 2017-02-22 is 2043.3381294964029
The predicted value for avg_trans_per_block on 2017-02-22 is 2085.242314553381
The real value of mkt_price was [[ 1123.2231875]] on 2017-02-22
The predicted value of mkt_price was [[ 1078.0909001]] on 2017-02-22

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.714
Model:                            OLS   Adj. R-squared:                  0.682
Method:                 Least Squares   F-statistic:                     22.18
Date:                Sun, 17 Sep 2017   Prob (F-statistic):           2.46e-18
Time:                        16:22:50   Log-Likelihood:                -493.08
No. Observations:                  90   AIC:                             1006.
Df Residuals:                      80   BIC:                             1031.
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const       1188.4086    593.774      2.001      0.049       6.761    2370.056
x1            -1.2263      0.495     -2.476      0.015      -2.212      -0.241
x2            29.4184     11.661      2.523      0.014       6.211      52.626
x3          2225.2066    923.442      2.410      0.018     387.498    4062.915
x4         -1458.2536    968.243     -1.506      0.136   -3385.119     468.612
x5            -0.0014      0.001     -2.214      0.030      -0.003      -0.000
x6             0.0002      0.000      0.393      0.696      -0.001       0.001
x7             0.0002   3.29e-05      5.642      0.000       0.000       0.000
x8          -9.11e-05      0.000     -0.613      0.541      -0.000       0.000
x9             0.2773      0.062      4.469      0.000       0.154       0.401
==============================================================================
Omnibus:                       12.235   Durbin-Watson:                   0.519
Prob(Omnibus):                  0.002   Jarque-Bera (JB):               12.852
Skew:                           0.823   Prob(JB):                      0.00162
Kurtosis:                       3.849   Cond. No.                     4.95e+08
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
[2] The condition number is large, 4.95e+08. This might indicate that there are
strong multicollinearity or other numerical problems.

This model architecture with 10 features has an RMSE score of 1.4085185252157295

In [22]:
r2_score(true, pred)#blows our Prophet TS only model away!


Out[22]:
0.75234861747398374

Nested TS VS. FB Prophet TS

We see from the above that our model has an $R^2$ of 0.75! This greatly outperforms our baseline model of just using FaceBook Prophet to forecast the price of bitcoin! The RMSE is 1.40

This is quite impressive given that we only have 3 months of training data and are testing on one month!

The output above also shows regression output from statsmodels!

The following features were significant in all 30 models:

  • Gold Price

  • Ethereum Price

  • Positive Sentiment (Yay!)

  • Average Transactions Per Block

It is important, yet again, to note that this data does NOT take into account the wild fluctuations in price that bitcoin later experienced. We would need more data to affirm the significance of the above variables.


In [25]:
plt.plot(pred)
plt.plot(true)
plt.legend(["Prediction", 'Actual'], loc='upper left')
plt.xlabel("Prediction #")
plt.ylabel("Price")
plt.title("Nested TS - Price Prediction");



In [26]:
fig, ax = plt.subplots()
ax.scatter(true, pred, edgecolors=(0, 0, 0))
ax.plot([min(true), max(true)], [min(true), max(true)], 'k--', lw=3)
ax.set_xlabel('Actual')
ax.set_ylabel('Predicted')


Out[26]:
<matplotlib.text.Text at 0x1142f8a58>

In [27]:
plotting_dict_1 = {"eth_price": [], "pos_sent": [], "neg_sent": [], "unique_addr": [], "gold_price": [], "tot_num_trans": [], "mempool_trans":[], "hash_rate": [], "avg_trans_per_block":[]}

for index, sub_list in enumerate(beta_values):
    for tup in sub_list:
        plotting_dict_1[tup[0]].append(tup[1])

In [34]:
plot_key(plotting_dict_1, "pos_sent")# here we say the effect of positive sentiment through time!
plt.title("Positive Sentiment Effect on BTC Price")
plt.ylabel("Beta Value")
plt.xlabel("Model #")
plt.tight_layout()



In [35]:
plot_key(plotting_dict_1, "gold_price")
plt.title("Gold Price Effect on BTC Price")
plt.ylabel("Beta Value")
plt.xlabel("Model #")
plt.tight_layout()



In [36]:
plot_key(plotting_dict_1, "avg_trans_per_block")
plt.title("Avg. Trans per Block Effect on BTC Price")
plt.ylabel("Beta Value")
plt.xlabel("Model #")
plt.tight_layout()


Percent change model!

I will now run the same nested TS model as above, however, I will now make my 'target' variable the percent change in bitcoin price. In order to make this a log-og model, I will use the percentage change of all features as inputs into the TS model and thus the linear regression!

Since percent change will 'shift' our dataframe by one row, I omit the first row (which is all NaN's).

Thus, if we were to predict a percent change of $0.008010$ on 28-10-2017, then this would mean that the predicted price would be the price on 27-10-2017 $*predicted\_percent\_change$.


In [4]:
df_pct = df.copy(deep=True)
df_pct = df_pct.pct_change()
df_pct.rename(columns={"mkt_price": "percent_change"}, inplace=True)
df_pct = df_pct.iloc[1:, :] #first row is all NaN's
df_pct.head()


Out[4]:
gold_price eth_price pos_sent neg_sent tot_num_trans unique_addr hash_rate mempool_trans avg_trans_per_block percent_change
date
2016-10-27 -3.610958e-03 0.005238 -0.113268 -0.029983 0.444681 0.060274 0.165468 1.454602 0.239572 0.014880
2016-10-28 -2.678642e-03 -0.020408 0.010710 -0.123097 -0.231357 -0.051413 -0.061728 -0.230336 -0.180788 0.008010
2016-10-29 -4.996004e-15 -0.045213 0.080134 0.120302 -0.084452 -0.041016 -0.019737 -0.865297 -0.066018 0.039563
2016-10-30 -2.353673e-14 0.003250 -0.051700 0.018852 -0.088607 -0.134780 0.134228 0.172045 -0.196464 -0.023629
2016-10-31 6.556600e-03 0.028690 -0.004203 -0.260696 0.154970 0.273510 -0.130178 -0.317341 0.327823 0.005728

In [5]:
beta_values_p, pred_p, true_p = master(df_pct, 30)


---- Running model from 2016-10-27 to 2017-01-23 and predicting on 2017-01-24 ----
Creating Time series models: 
The real value for gold_price on 2017-01-24 is -0.00037075180235102145
The predicted value for gold_price on 2017-01-24 is 0.005760312059128465
The real value for eth_price on 2017-01-24 is -0.007895959126733021
The predicted value for eth_price on 2017-01-24 is 0.02477744372194353
The real value for pos_sent on 2017-01-24 is -0.10952281294819832
The predicted value for pos_sent on 2017-01-24 is -0.047601683404250084
The real value for neg_sent on 2017-01-24 is 0.14629332553048702
The predicted value for neg_sent on 2017-01-24 is -0.001670140565898595
The real value for tot_num_trans on 2017-01-24 is 0.17526997110168496
The predicted value for tot_num_trans on 2017-01-24 is 0.19600695623048575
The real value for unique_addr on 2017-01-24 is 0.16673072236074926
The predicted value for unique_addr on 2017-01-24 is 0.017314970604224898
The real value for hash_rate on 2017-01-24 is 0.07194244604316502
The predicted value for hash_rate on 2017-01-24 is 0.08136158053874175
The real value for mempool_trans on 2017-01-24 is 2.184052757793765
The predicted value for mempool_trans on 2017-01-24 is 1.3621781327289164
The real value for avg_trans_per_block on 2017-01-24 is 0.09639279183311555
The predicted value for avg_trans_per_block on 2017-01-24 is 0.09008812355549371
The real value of percent_change was [[-0.03443693]] on 2017-01-24
The predicted value of percent_change was [[ 0.01131492]] on 2017-01-24

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.153
Model:                            OLS   Adj. R-squared:                  0.057
Method:                 Least Squares   F-statistic:                     1.587
Date:                Sun, 17 Sep 2017   Prob (F-statistic):              0.133
Time:                        15:25:54   Log-Likelihood:                 187.36
No. Observations:                  89   AIC:                            -354.7
Df Residuals:                      79   BIC:                            -329.8
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0054      0.004      1.367      0.176      -0.002       0.013
x1             0.5265      0.453      1.161      0.249      -0.376       1.429
x2             0.2016      0.106      1.894      0.062      -0.010       0.413
x3             0.0177      0.053      0.336      0.738      -0.087       0.122
x4            -0.0733      0.029     -2.562      0.012      -0.130      -0.016
x5             0.0276      0.112      0.247      0.806      -0.195       0.250
x6             0.0195      0.059      0.330      0.742      -0.098       0.137
x7            -0.0109      0.106     -0.103      0.918      -0.222       0.200
x8            -0.0035      0.005     -0.711      0.479      -0.013       0.006
x9            -0.0161      0.111     -0.146      0.885      -0.236       0.204
==============================================================================
Omnibus:                       45.593   Durbin-Watson:                   2.066
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              177.618
Skew:                          -1.602   Prob(JB):                     2.70e-39
Kurtosis:                       9.134   Cond. No.                         141.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-10-28 to 2017-01-24 and predicting on 2017-01-25 ----
Creating Time series models: 
The real value for gold_price on 2017-01-25 is -0.008077144976425754
The predicted value for gold_price on 2017-01-25 is 0.004521228019330714
The real value for eth_price on 2017-01-25 is -0.007958801498085366
The predicted value for eth_price on 2017-01-25 is 0.025375980352773463
The real value for pos_sent on 2017-01-25 is 0.07083394098419737
The predicted value for pos_sent on 2017-01-25 is -0.0018205682445750601
The real value for neg_sent on 2017-01-25 is 0.1809913799368854
The predicted value for neg_sent on 2017-01-25 is 0.016737890085646433
The real value for tot_num_trans on 2017-01-25 is 0.09987885411087194
The predicted value for tot_num_trans on 2017-01-25 is -0.09722396918572462
The real value for unique_addr on 2017-01-25 is 0.033704389605068164
The predicted value for unique_addr on 2017-01-25 is -0.04419057061471744
The real value for hash_rate on 2017-01-25 is 0.006711409395973478
The predicted value for hash_rate on 2017-01-25 is -0.013126105671929468
The real value for mempool_trans on 2017-01-25 is 9.230629824891734
The predicted value for mempool_trans on 2017-01-25 is -0.07446360922889225
The real value for avg_trans_per_block on 2017-01-25 is 0.09254632841679955
The predicted value for avg_trans_per_block on 2017-01-25 is -0.06628025994436923
The real value of percent_change was [[ 0.00306115]] on 2017-01-25
The predicted value of percent_change was [[ 0.00990098]] on 2017-01-25

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.161
Model:                            OLS   Adj. R-squared:                  0.066
Method:                 Least Squares   F-statistic:                     1.689
Date:                Sun, 17 Sep 2017   Prob (F-statistic):              0.106
Time:                        15:26:07   Log-Likelihood:                 187.14
No. Observations:                  89   AIC:                            -354.3
Df Residuals:                      79   BIC:                            -329.4
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0051      0.004      1.276      0.206      -0.003       0.013
x1             0.5399      0.454      1.189      0.238      -0.364       1.444
x2             0.2071      0.106      1.946      0.055      -0.005       0.419
x3             0.0240      0.053      0.457      0.649      -0.081       0.129
x4            -0.0744      0.029     -2.591      0.011      -0.132      -0.017
x5             0.0155      0.119      0.130      0.897      -0.221       0.252
x6             0.0216      0.064      0.340      0.735      -0.105       0.148
x7            -0.0018      0.108     -0.016      0.987      -0.216       0.212
x8            -0.0046      0.005     -0.956      0.342      -0.014       0.005
x9            -0.0078      0.112     -0.070      0.944      -0.231       0.216
==============================================================================
Omnibus:                       43.108   Durbin-Watson:                   2.064
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              159.956
Skew:                          -1.519   Prob(JB):                     1.84e-35
Kurtosis:                       8.823   Cond. No.                         141.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-10-29 to 2017-01-25 and predicting on 2017-01-26 ----
Creating Time series models: 
The real value for gold_price on 2017-01-26 is -0.009929372663151526
The predicted value for gold_price on 2017-01-26 is -0.0026370245442384225
The real value for eth_price on 2017-01-26 is -0.0014157621519175212
The predicted value for eth_price on 2017-01-26 is 0.0036850316459171983
The real value for pos_sent on 2017-01-26 is 0.059524584703593675
The predicted value for pos_sent on 2017-01-26 is 0.027214068087432153
The real value for neg_sent on 2017-01-26 is -0.08448885129794548
The predicted value for neg_sent on 2017-01-26 is 0.08009034706630527
The real value for tot_num_trans on 2017-01-26 is 0.03903438673556425
The predicted value for tot_num_trans on 2017-01-26 is -0.08134280907822058
The real value for unique_addr on 2017-01-26 is 0.0067620346729158065
The predicted value for unique_addr on 2017-01-26 is -0.06542449142480526
The real value for hash_rate on 2017-01-26 is 0.046666666666666634
The predicted value for hash_rate on 2017-01-26 is 0.04026325440246546
The real value for mempool_trans on 2017-01-26 is -0.10625772121496557
The predicted value for mempool_trans on 2017-01-26 is -0.13637431662292118
The real value for avg_trans_per_block on 2017-01-26 is -0.007291987195320537
The predicted value for avg_trans_per_block on 2017-01-26 is -0.11229992137379204
The real value of percent_change was [[ 0.02565448]] on 2017-01-26
The predicted value of percent_change was [[-0.00290918]] on 2017-01-26

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.150
Model:                            OLS   Adj. R-squared:                  0.053
Method:                 Least Squares   F-statistic:                     1.548
Date:                Sun, 17 Sep 2017   Prob (F-statistic):              0.146
Time:                        15:26:17   Log-Likelihood:                 186.55
No. Observations:                  89   AIC:                            -353.1
Df Residuals:                      79   BIC:                            -328.2
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0044      0.004      1.094      0.277      -0.004       0.012
x1             0.4595      0.450      1.020      0.311      -0.437       1.356
x2             0.1998      0.107      1.865      0.066      -0.013       0.413
x3             0.0291      0.053      0.551      0.583      -0.076       0.134
x4            -0.0694      0.029     -2.413      0.018      -0.127      -0.012
x5             0.0065      0.119      0.054      0.957      -0.231       0.244
x6             0.0169      0.065      0.260      0.796      -0.113       0.147
x7         -8.864e-05      0.109     -0.001      0.999      -0.218       0.218
x8            -0.0006      0.003     -0.218      0.828      -0.006       0.005
x9             0.0045      0.114      0.040      0.968      -0.222       0.231
==============================================================================
Omnibus:                       43.289   Durbin-Watson:                   2.052
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              160.207
Skew:                          -1.528   Prob(JB):                     1.63e-35
Kurtosis:                       8.819   Cond. No.                         176.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-10-30 to 2017-01-26 and predicting on 2017-01-27 ----
Creating Time series models: 
The real value for gold_price on 2017-01-27 is -0.006168436070493888
The predicted value for gold_price on 2017-01-27 is -0.000674814802653788
The real value for eth_price on 2017-01-27 is -0.0018903591683820942
The predicted value for eth_price on 2017-01-27 is -0.0032648946615146893
The real value for pos_sent on 2017-01-27 is -0.006099944039130278
The predicted value for pos_sent on 2017-01-27 is 0.015458799490093779
The real value for neg_sent on 2017-01-27 is 0.09242812313364057
The predicted value for neg_sent on 2017-01-27 is -0.04943469194716527
The real value for tot_num_trans on 2017-01-27 is -0.17166188534903615
The predicted value for tot_num_trans on 2017-01-27 is -0.05267118803203162
The real value for unique_addr on 2017-01-27 is -0.10647767496979454
The predicted value for unique_addr on 2017-01-27 is -0.06730125117129324
The real value for hash_rate on 2017-01-27 is 0.08280254777070017
The predicted value for hash_rate on 2017-01-27 is 0.023031478318183474
The real value for mempool_trans on 2017-01-27 is -0.8297489517131671
The predicted value for mempool_trans on 2017-01-27 is 0.1992789265797904
The real value for avg_trans_per_block on 2017-01-27 is -0.23500538823410988
The predicted value for avg_trans_per_block on 2017-01-27 is -0.054834225476666665
The real value of percent_change was [[ 0.00362845]] on 2017-01-27
The predicted value of percent_change was [[ 0.00530485]] on 2017-01-27

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.169
Model:                            OLS   Adj. R-squared:                  0.075
Method:                 Least Squares   F-statistic:                     1.789
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0833
Time:                        15:26:29   Log-Likelihood:                 187.97
No. Observations:                  89   AIC:                            -355.9
Df Residuals:                      79   BIC:                            -331.0
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0038      0.004      0.972      0.334      -0.004       0.012
x1             0.4336      0.441      0.984      0.328      -0.443       1.311
x2             0.2225      0.106      2.098      0.039       0.011       0.434
x3             0.0212      0.052      0.405      0.686      -0.083       0.125
x4            -0.0754      0.028     -2.649      0.010      -0.132      -0.019
x5             0.0070      0.117      0.060      0.952      -0.226       0.240
x6             0.0115      0.064      0.179      0.858      -0.116       0.139
x7             0.0045      0.108      0.042      0.967      -0.210       0.219
x8            -0.0002      0.003     -0.064      0.949      -0.006       0.006
x9             0.0091      0.112      0.081      0.936      -0.213       0.231
==============================================================================
Omnibus:                       44.963   Durbin-Watson:                   2.043
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              168.667
Skew:                          -1.594   Prob(JB):                     2.37e-37
Kurtosis:                       8.943   Cond. No.                         174.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-10-31 to 2017-01-27 and predicting on 2017-01-28 ----
Creating Time series models: 
The real value for gold_price on 2017-01-28 is -4.9960036108132044e-15
The predicted value for gold_price on 2017-01-28 is 0.0037234345649768148
The real value for eth_price on 2017-01-28 is -0.0009469696968982255
The predicted value for eth_price on 2017-01-28 is 0.011650512327204163
The real value for pos_sent on 2017-01-28 is 0.12044476808958482
The predicted value for pos_sent on 2017-01-28 is 0.06234448018195863
The real value for neg_sent on 2017-01-28 is -0.21028049733533094
The predicted value for neg_sent on 2017-01-28 is -0.037663832003662905
The real value for tot_num_trans on 2017-01-28 is -0.06431804717998568
The predicted value for tot_num_trans on 2017-01-28 is -0.04906498243903882
The real value for unique_addr on 2017-01-28 is -0.003583540339129443
The predicted value for unique_addr on 2017-01-28 is 0.024747932809002268
The real value for hash_rate on 2017-01-28 is -0.10588235294117632
The predicted value for hash_rate on 2017-01-28 is 0.010113988708799638
The real value for mempool_trans on 2017-01-28 is -0.5156861855732451
The predicted value for mempool_trans on 2017-01-28 is 0.41519622253902577
The real value for avg_trans_per_block on 2017-01-28 is 0.04648639460133164
The predicted value for avg_trans_per_block on 2017-01-28 is -0.04142141482163712
The real value of percent_change was [[ 0.00112316]] on 2017-01-28
The predicted value of percent_change was [[ 0.01170557]] on 2017-01-28

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.167
Model:                            OLS   Adj. R-squared:                  0.072
Method:                 Least Squares   F-statistic:                     1.755
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0904
Time:                        15:26:41   Log-Likelihood:                 188.18
No. Observations:                  89   AIC:                            -356.4
Df Residuals:                      79   BIC:                            -331.5
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0044      0.004      1.105      0.272      -0.003       0.012
x1             0.4284      0.438      0.978      0.331      -0.443       1.300
x2             0.2235      0.106      2.113      0.038       0.013       0.434
x3             0.0161      0.052      0.307      0.760      -0.088       0.120
x4            -0.0754      0.028     -2.661      0.009      -0.132      -0.019
x5             0.0100      0.117      0.085      0.932      -0.223       0.243
x6             0.0055      0.065      0.086      0.932      -0.123       0.134
x7             0.0048      0.107      0.045      0.965      -0.209       0.218
x8            -0.0003      0.003     -0.113      0.911      -0.006       0.005
x9             0.0062      0.111      0.055      0.956      -0.216       0.228
==============================================================================
Omnibus:                       47.023   Durbin-Watson:                   2.049
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              184.555
Skew:                          -1.662   Prob(JB):                     8.40e-41
Kurtosis:                       9.223   Cond. No.                         174.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-01 to 2017-01-28 and predicting on 2017-01-29 ----
Creating Time series models: 
The real value for gold_price on 2017-01-29 is 4.218847493575595e-15
The predicted value for gold_price on 2017-01-29 is 0.004336710875604626
The real value for eth_price on 2017-01-29 is -0.002843601895841208
The predicted value for eth_price on 2017-01-29 is 0.007967994041826652
The real value for pos_sent on 2017-01-29 is -0.07862646727424938
The predicted value for pos_sent on 2017-01-29 is -0.02408069975637564
The real value for neg_sent on 2017-01-29 is -0.04079258641908501
The predicted value for neg_sent on 2017-01-29 is -0.0021829555450274644
The real value for tot_num_trans on 2017-01-29 is -0.01935104202601512
The predicted value for tot_num_trans on 2017-01-29 is -0.0610405450212488
The real value for unique_addr on 2017-01-29 is -0.04891489320190512
The predicted value for unique_addr on 2017-01-29 is -0.06617079227060356
The real value for hash_rate on 2017-01-29 is 0.09868421052631571
The predicted value for hash_rate on 2017-01-29 is 0.0026814266927012165
The real value for mempool_trans on 2017-01-29 is 1.1310211344550933
The predicted value for mempool_trans on 2017-01-29 is 0.05262244605003441
The real value for avg_trans_per_block on 2017-01-29 is -0.10743328376020533
The predicted value for avg_trans_per_block on 2017-01-29 is -0.05860261733740651
The real value of percent_change was [[-0.00475844]] on 2017-01-29
The predicted value of percent_change was [[ 0.00601819]] on 2017-01-29

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.176
Model:                            OLS   Adj. R-squared:                  0.082
Method:                 Least Squares   F-statistic:                     1.870
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0687
Time:                        15:26:55   Log-Likelihood:                 188.66
No. Observations:                  89   AIC:                            -357.3
Df Residuals:                      79   BIC:                            -332.4
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0043      0.004      1.105      0.272      -0.003       0.012
x1             0.4750      0.438      1.084      0.282      -0.397       1.347
x2             0.2300      0.105      2.183      0.032       0.020       0.440
x3             0.0113      0.052      0.219      0.827      -0.092       0.114
x4            -0.0792      0.028     -2.781      0.007      -0.136      -0.023
x5            -0.0168      0.118     -0.141      0.888      -0.252       0.219
x6             0.0277      0.068      0.408      0.684      -0.107       0.163
x7             0.0190      0.107      0.178      0.859      -0.194       0.232
x8            -0.0002      0.003     -0.086      0.932      -0.006       0.005
x9             0.0223      0.111      0.200      0.842      -0.199       0.244
==============================================================================
Omnibus:                       46.546   Durbin-Watson:                   1.987
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              179.975
Skew:                          -1.649   Prob(JB):                     8.30e-40
Kurtosis:                       9.137   Cond. No.                         175.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-02 to 2017-01-29 and predicting on 2017-01-30 ----
Creating Time series models: 
The real value for gold_price on 2017-01-30 is 0.004771153521248683
The predicted value for gold_price on 2017-01-30 is 0.0030349043299289983
The real value for eth_price on 2017-01-30 is 0.0009505703422094669
The predicted value for eth_price on 2017-01-30 is -0.0020797340045430835
The real value for pos_sent on 2017-01-30 is -0.090420507253273
The predicted value for pos_sent on 2017-01-30 is -0.046370924778520825
The real value for neg_sent on 2017-01-30 is 0.27077560727068706
The predicted value for neg_sent on 2017-01-30 is -0.03760191636276943
The real value for tot_num_trans on 2017-01-30 is 0.03808349232494712
The predicted value for tot_num_trans on 2017-01-30 is 0.18383180827984177
The real value for unique_addr on 2017-01-30 is 0.0241927212138755
The predicted value for unique_addr on 2017-01-30 is 0.1615705131354522
The real value for hash_rate on 2017-01-30 is -0.12574850299401197
The predicted value for hash_rate on 2017-01-30 is -0.026993423611225337
The real value for mempool_trans on 2017-01-30 is -0.0964651421707221
The predicted value for mempool_trans on 2017-01-30 is -0.14752085616271354
The real value for avg_trans_per_block on 2017-01-30 is 0.18739687135798744
The predicted value for avg_trans_per_block on 2017-01-30 is 0.243363900584849
The real value of percent_change was [[ 0.00572785]] on 2017-01-30
The predicted value of percent_change was [[ 0.01368477]] on 2017-01-30

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.189
Model:                            OLS   Adj. R-squared:                  0.096
Method:                 Least Squares   F-statistic:                     2.039
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0455
Time:                        15:27:07   Log-Likelihood:                 189.90
No. Observations:                  89   AIC:                            -359.8
Df Residuals:                      79   BIC:                            -334.9
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0036      0.004      0.929      0.356      -0.004       0.011
x1             0.3947      0.434      0.909      0.366      -0.470       1.259
x2             0.2469      0.104      2.366      0.020       0.039       0.455
x3             0.0145      0.050      0.288      0.774      -0.086       0.115
x4            -0.0805      0.028     -2.882      0.005      -0.136      -0.025
x5            -0.0297      0.117     -0.254      0.800      -0.262       0.203
x6             0.0384      0.067      0.573      0.568      -0.095       0.172
x7             0.0259      0.105      0.246      0.807      -0.184       0.236
x8            -0.0004      0.003     -0.146      0.885      -0.006       0.005
x9             0.0287      0.110      0.262      0.794      -0.189       0.247
==============================================================================
Omnibus:                       45.668   Durbin-Watson:                   2.028
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              175.642
Skew:                          -1.612   Prob(JB):                     7.24e-39
Kurtosis:                       9.080   Cond. No.                         176.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-03 to 2017-01-30 and predicting on 2017-01-31 ----
Creating Time series models: 
The real value for gold_price on 2017-01-31 is 0.007521956549086939
The predicted value for gold_price on 2017-01-31 is 0.005281111023783189
The real value for eth_price on 2017-01-31 is 0.010921177587823738
The predicted value for eth_price on 2017-01-31 is 0.02402852630315866
The real value for pos_sent on 2017-01-31 is 0.008139409089829464
The predicted value for pos_sent on 2017-01-31 is -0.05494743046163883
The real value for neg_sent on 2017-01-31 is -0.10733094189615366
The predicted value for neg_sent on 2017-01-31 is 0.10300473108946542
The real value for tot_num_trans on 2017-01-31 is 0.22666438091289276
The predicted value for tot_num_trans on 2017-01-31 is 0.07111870189399788
The real value for unique_addr on 2017-01-31 is 0.16400684558052792
The predicted value for unique_addr on 2017-01-31 is 0.01819329340935112
The real value for hash_rate on 2017-01-31 is -0.006849315068492734
The predicted value for hash_rate on 2017-01-31 is 0.0700527358325116
The real value for mempool_trans on 2017-01-31 is 0.8539349524952171
The predicted value for mempool_trans on 2017-01-31 is 1.4315715791437014
The real value for avg_trans_per_block on 2017-01-31 is 0.23512413526401676
The predicted value for avg_trans_per_block on 2017-01-31 is -0.014308647357786548
The real value of percent_change was [[ 0.04725111]] on 2017-01-31
The predicted value of percent_change was [[ 0.00228994]] on 2017-01-31

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.184
Model:                            OLS   Adj. R-squared:                  0.091
Method:                 Least Squares   F-statistic:                     1.978
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0528
Time:                        15:27:19   Log-Likelihood:                 189.65
No. Observations:                  89   AIC:                            -359.3
Df Residuals:                      79   BIC:                            -334.4
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0037      0.004      0.961      0.339      -0.004       0.011
x1             0.3981      0.440      0.904      0.369      -0.478       1.274
x2             0.2461      0.105      2.346      0.022       0.037       0.455
x3             0.0118      0.050      0.235      0.815      -0.088       0.112
x4            -0.0770      0.028     -2.795      0.007      -0.132      -0.022
x5            -0.0345      0.117     -0.295      0.769      -0.267       0.198
x6             0.0376      0.067      0.558      0.578      -0.096       0.172
x7             0.0289      0.106      0.274      0.785      -0.181       0.239
x8            -0.0004      0.003     -0.156      0.876      -0.006       0.005
x9             0.0345      0.110      0.315      0.754      -0.183       0.252
==============================================================================
Omnibus:                       46.567   Durbin-Watson:                   2.002
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              179.995
Skew:                          -1.650   Prob(JB):                     8.21e-40
Kurtosis:                       9.136   Cond. No.                         178.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-04 to 2017-01-31 and predicting on 2017-02-01 ----
Creating Time series models: 
The real value for gold_price on 2017-02-01 is 0.00934267600943195
The predicted value for gold_price on 2017-02-01 is 0.0072390604222625065
The real value for eth_price on 2017-02-01 is 0.008454673555788661
The predicted value for eth_price on 2017-02-01 is 0.03630622509178755
The real value for pos_sent on 2017-02-01 is 0.008867242154636079
The predicted value for pos_sent on 2017-02-01 is 0.04149764553301076
The real value for neg_sent on 2017-02-01 is -0.14814338240857639
The predicted value for neg_sent on 2017-02-01 is 0.013827408083275164
The real value for tot_num_trans on 2017-02-01 is 0.13935446547258046
The predicted value for tot_num_trans on 2017-02-01 is -0.0506839823032646
The real value for unique_addr on 2017-02-01 is 0.14533607629302758
The predicted value for unique_addr on 2017-02-01 is -0.027293435065750153
The real value for hash_rate on 2017-02-01 is 0.3241379310344823
The predicted value for hash_rate on 2017-02-01 is -0.08117112769702206
The real value for mempool_trans on 2017-02-01 is 0.8324923259901877
The predicted value for mempool_trans on 2017-02-01 is 1.0391172982596015
The real value for avg_trans_per_block on 2017-02-01 is -0.13955001305456194
The predicted value for avg_trans_per_block on 2017-02-01 is 0.05980271188800578
The real value of percent_change was [[ 0.0155465]] on 2017-02-01
The predicted value of percent_change was [[ 0.01532338]] on 2017-02-01

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.171
Model:                            OLS   Adj. R-squared:                  0.077
Method:                 Least Squares   F-statistic:                     1.817
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0781
Time:                        15:27:32   Log-Likelihood:                 190.35
No. Observations:                  89   AIC:                            -360.7
Df Residuals:                      79   BIC:                            -335.8
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0043      0.004      1.126      0.264      -0.003       0.012
x1             0.3944      0.434      0.908      0.366      -0.470       1.259
x2             0.2525      0.104      2.423      0.018       0.045       0.460
x3             0.0164      0.050      0.328      0.743      -0.083       0.116
x4            -0.0677      0.028     -2.378      0.020      -0.124      -0.011
x5            -0.0307      0.116     -0.265      0.792      -0.261       0.200
x6             0.0365      0.067      0.547      0.586      -0.096       0.170
x7             0.0338      0.105      0.323      0.748      -0.175       0.242
x8            -0.0006      0.003     -0.227      0.821      -0.006       0.005
x9             0.0350      0.109      0.322      0.748      -0.181       0.251
==============================================================================
Omnibus:                       50.078   Durbin-Watson:                   1.978
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              210.785
Skew:                          -1.761   Prob(JB):                     1.69e-46
Kurtosis:                       9.666   Cond. No.                         178.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-05 to 2017-02-01 and predicting on 2017-02-02 ----
Creating Time series models: 
The real value for gold_price on 2017-02-02 is 0.011611570247839209
The predicted value for gold_price on 2017-02-02 is 0.0011753499200797208
The real value for eth_price on 2017-02-02 is 0.004191895668261836
The predicted value for eth_price on 2017-02-02 is 0.016116098364192116
The real value for pos_sent on 2017-02-02 is 0.10755127491503425
The predicted value for pos_sent on 2017-02-02 is 0.016474892325463102
The real value for neg_sent on 2017-02-02 is 0.29139188791158954
The predicted value for neg_sent on 2017-02-02 is -0.11149301286461073
The real value for tot_num_trans on 2017-02-02 is -0.25958751711547234
The predicted value for tot_num_trans on 2017-02-02 is -0.06059262820435519
The real value for unique_addr on 2017-02-02 is -0.244497973464741
The predicted value for unique_addr on 2017-02-02 is -0.05739767179533191
The real value for hash_rate on 2017-02-02 is -0.18229166666666652
The predicted value for hash_rate on 2017-02-02 is 0.11072370999171455
The real value for mempool_trans on 2017-02-02 is -0.46057554643504817
The predicted value for mempool_trans on 2017-02-02 is 0.4311597130224351
The real value for avg_trans_per_block on 2017-02-02 is -0.09452740946605531
The predicted value for avg_trans_per_block on 2017-02-02 is -0.1375986891188492
The real value of percent_change was [[ 0.02848803]] on 2017-02-02
The predicted value of percent_change was [[ 0.01452091]] on 2017-02-02

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.168
Model:                            OLS   Adj. R-squared:                  0.073
Method:                 Least Squares   F-statistic:                     1.773
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0866
Time:                        15:27:44   Log-Likelihood:                 190.33
No. Observations:                  89   AIC:                            -360.7
Df Residuals:                      79   BIC:                            -335.8
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0042      0.004      1.104      0.273      -0.003       0.012
x1             0.3555      0.427      0.832      0.408      -0.495       1.206
x2             0.2517      0.104      2.413      0.018       0.044       0.459
x3             0.0138      0.051      0.272      0.786      -0.087       0.115
x4            -0.0663      0.028     -2.353      0.021      -0.122      -0.010
x5            -0.0198      0.119     -0.166      0.868      -0.256       0.217
x6             0.0315      0.067      0.474      0.637      -0.101       0.164
x7             0.0250      0.106      0.237      0.813      -0.185       0.235
x8            -0.0007      0.003     -0.236      0.814      -0.006       0.005
x9             0.0265      0.111      0.238      0.812      -0.195       0.248
==============================================================================
Omnibus:                       50.076   Durbin-Watson:                   1.988
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              213.109
Skew:                          -1.755   Prob(JB):                     5.30e-47
Kurtosis:                       9.719   Cond. No.                         176.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-06 to 2017-02-02 and predicting on 2017-02-03 ----
Creating Time series models: 
The real value for gold_price on 2017-02-03 is -0.00898656100649231
The predicted value for gold_price on 2017-02-03 is 0.003979985399693414
The real value for eth_price on 2017-02-03 is 0.016697588126161067
The predicted value for eth_price on 2017-02-03 is -0.008010835087858464
The real value for pos_sent on 2017-02-03 is -0.01507493751693334
The predicted value for pos_sent on 2017-02-03 is -0.012767739263535385
The real value for neg_sent on 2017-02-03 is -0.10138278756639274
The predicted value for neg_sent on 2017-02-03 is 0.006724431752906848
The real value for tot_num_trans on 2017-02-03 is 0.24658748107366724
The predicted value for tot_num_trans on 2017-02-03 is -0.07584099659425053
The real value for unique_addr on 2017-02-03 is 0.23816299700608723
The predicted value for unique_addr on 2017-02-03 is -0.029836619409549292
The real value for hash_rate on 2017-02-03 is -0.050955414012739175
The predicted value for hash_rate on 2017-02-03 is 0.007405718993625705
The real value for mempool_trans on 2017-02-03 is 0.1971730469222306
The predicted value for mempool_trans on 2017-02-03 is 0.00021333007480683897
The real value for avg_trans_per_block on 2017-02-03 is 0.3135183525407099
The predicted value for avg_trans_per_block on 2017-02-03 is -0.05873335613187458
The real value of percent_change was [[ 0.00537238]] on 2017-02-03
The predicted value of percent_change was [[ 0.00321369]] on 2017-02-03

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.158
Model:                            OLS   Adj. R-squared:                  0.063
Method:                 Least Squares   F-statistic:                     1.652
Date:                Sun, 17 Sep 2017   Prob (F-statistic):              0.115
Time:                        15:27:56   Log-Likelihood:                 189.53
No. Observations:                  89   AIC:                            -359.1
Df Residuals:                      79   BIC:                            -334.2
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0050      0.004      1.313      0.193      -0.003       0.013
x1             0.4388      0.432      1.016      0.313      -0.421       1.299
x2             0.2540      0.105      2.414      0.018       0.045       0.463
x3             0.0240      0.051      0.475      0.636      -0.077       0.125
x4            -0.0625      0.029     -2.189      0.032      -0.119      -0.006
x5            -0.0133      0.120     -0.111      0.912      -0.252       0.225
x6             0.0126      0.067      0.189      0.851      -0.120       0.146
x7             0.0259      0.108      0.241      0.810      -0.188       0.240
x8            -0.0007      0.003     -0.264      0.793      -0.006       0.005
x9             0.0273      0.113      0.242      0.809      -0.197       0.252
==============================================================================
Omnibus:                       51.621   Durbin-Watson:                   1.958
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              225.117
Skew:                          -1.812   Prob(JB):                     1.31e-49
Kurtosis:                       9.897   Cond. No.                         175.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-07 to 2017-02-03 and predicting on 2017-02-04 ----
Creating Time series models: 
The real value for gold_price on 2017-02-04 is 1.7763568394002505e-15
The predicted value for gold_price on 2017-02-04 is 0.0009089806319667478
The real value for eth_price on 2017-02-04 is 0.029197080291973432
The predicted value for eth_price on 2017-02-04 is 0.0048427752592801075
The real value for pos_sent on 2017-02-04 is 0.032741090204606804
The predicted value for pos_sent on 2017-02-04 is 0.06477805247364787
The real value for neg_sent on 2017-02-04 is 0.10423167177855541
The predicted value for neg_sent on 2017-02-04 is 0.015789276316006556
The real value for tot_num_trans on 2017-02-04 is -0.11508423398225387
The predicted value for tot_num_trans on 2017-02-04 is -0.027356841494433487
The real value for unique_addr on 2017-02-04 is -0.1075638769821583
The predicted value for unique_addr on 2017-02-04 is 0.01954727296536892
The real value for hash_rate on 2017-02-04 is 0.04548482826424105
The predicted value for hash_rate on 2017-02-04 is -0.01363061430264948
The real value for mempool_trans on 2017-02-04 is 3.5318620628496404
The predicted value for mempool_trans on 2017-02-04 is 0.3612014095187891
The real value for avg_trans_per_block on 2017-02-04 is -0.09067276457486795
The predicted value for avg_trans_per_block on 2017-02-04 is -0.013297112204939177
The real value of percent_change was [[ 0.0177413]] on 2017-02-04
The predicted value of percent_change was [[ 0.00649027]] on 2017-02-04

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.158
Model:                            OLS   Adj. R-squared:                  0.062
Method:                 Least Squares   F-statistic:                     1.643
Date:                Sun, 17 Sep 2017   Prob (F-statistic):              0.117
Time:                        15:28:09   Log-Likelihood:                 189.51
No. Observations:                  89   AIC:                            -359.0
Df Residuals:                      79   BIC:                            -334.1
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0047      0.004      1.222      0.225      -0.003       0.012
x1             0.4544      0.430      1.058      0.293      -0.401       1.309
x2             0.2527      0.105      2.404      0.019       0.044       0.462
x3             0.0262      0.051      0.511      0.611      -0.076       0.128
x4            -0.0617      0.029     -2.156      0.034      -0.119      -0.005
x5            -0.0120      0.120     -0.100      0.920      -0.251       0.227
x6             0.0064      0.066      0.096      0.923      -0.126       0.138
x7             0.0285      0.108      0.264      0.793      -0.186       0.243
x8            -0.0007      0.003     -0.248      0.805      -0.006       0.005
x9             0.0294      0.113      0.260      0.796      -0.196       0.255
==============================================================================
Omnibus:                       51.380   Durbin-Watson:                   1.981
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              224.347
Skew:                          -1.801   Prob(JB):                     1.92e-49
Kurtosis:                       9.894   Cond. No.                         174.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-08 to 2017-02-04 and predicting on 2017-02-05 ----
Creating Time series models: 
The real value for gold_price on 2017-02-05 is -5.551115123125783e-16
The predicted value for gold_price on 2017-02-05 is 0.002390808812072436
The real value for eth_price on 2017-02-05 is 0.010195035460960788
The predicted value for eth_price on 2017-02-05 is 0.01623795598235315
The real value for pos_sent on 2017-02-05 is -0.029275791452305633
The predicted value for pos_sent on 2017-02-05 is -0.038608573623958496
The real value for neg_sent on 2017-02-05 is 0.05655037494007553
The predicted value for neg_sent on 2017-02-05 is -0.02479170916502444
The real value for tot_num_trans on 2017-02-05 is 0.02016931749125117
The predicted value for tot_num_trans on 2017-02-05 is -0.030302046860973497
The real value for unique_addr on 2017-02-05 is 0.020354010613919726
The predicted value for unique_addr on 2017-02-05 is -0.053921259245508515
The real value for hash_rate on 2017-02-05 is 0.09655172413793056
The predicted value for hash_rate on 2017-02-05 is 0.08608113786496827
The real value for mempool_trans on 2017-02-05 is -0.7278997979587145
The predicted value for mempool_trans on 2017-02-05 is 0.845301830400483
The real value for avg_trans_per_block on 2017-02-05 is -0.06965691172181487
The predicted value for avg_trans_per_block on 2017-02-05 is -0.0855494225411361
The real value of percent_change was [[-0.01567575]] on 2017-02-05
The predicted value of percent_change was [[ 0.01005782]] on 2017-02-05

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.157
Model:                            OLS   Adj. R-squared:                  0.060
Method:                 Least Squares   F-statistic:                     1.629
Date:                Sun, 17 Sep 2017   Prob (F-statistic):              0.121
Time:                        15:28:20   Log-Likelihood:                 189.49
No. Observations:                  89   AIC:                            -359.0
Df Residuals:                      79   BIC:                            -334.1
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0047      0.004      1.220      0.226      -0.003       0.012
x1             0.4241      0.435      0.974      0.333      -0.443       1.291
x2             0.2573      0.105      2.455      0.016       0.049       0.466
x3             0.0285      0.051      0.555      0.580      -0.074       0.131
x4            -0.0597      0.029     -2.077      0.041      -0.117      -0.002
x5            -0.0212      0.118     -0.180      0.857      -0.255       0.213
x6             0.0046      0.066      0.070      0.944      -0.127       0.136
x7             0.0380      0.106      0.358      0.721      -0.173       0.249
x8            -0.0005      0.003     -0.172      0.864      -0.006       0.005
x9             0.0407      0.111      0.367      0.715      -0.180       0.261
==============================================================================
Omnibus:                       51.545   Durbin-Watson:                   1.988
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              224.589
Skew:                          -1.809   Prob(JB):                     1.70e-49
Kurtosis:                       9.890   Cond. No.                         184.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-09 to 2017-02-05 and predicting on 2017-02-06 ----
Creating Time series models: 
The real value for gold_price on 2017-02-06 is 0.007254441284386681
The predicted value for gold_price on 2017-02-06 is 0.009555411255081782
The real value for eth_price on 2017-02-06 is -0.0021939447124099676
The predicted value for eth_price on 2017-02-06 is 0.0045166054767428005
The real value for pos_sent on 2017-02-06 is -0.08624827856658379
The predicted value for pos_sent on 2017-02-06 is -0.002611160042591168
The real value for neg_sent on 2017-02-06 is -0.10744417790833793
The predicted value for neg_sent on 2017-02-06 is 0.09429391139753357
The real value for tot_num_trans on 2017-02-06 is -0.21250868706372106
The predicted value for tot_num_trans on 2017-02-06 is 0.13325330970284055
The real value for unique_addr on 2017-02-06 is -0.1369217111490426
The predicted value for unique_addr on 2017-02-06 is 0.1267611594596011
The real value for hash_rate on 2017-02-06 is -0.21383647798742123
The predicted value for hash_rate on 2017-02-06 is -0.06175458581947332
The real value for mempool_trans on 2017-02-06 is -0.5162139000257697
The predicted value for mempool_trans on 2017-02-06 is -0.05986926522367897
The real value for avg_trans_per_block on 2017-02-06 is 0.0016889500549470693
The predicted value for avg_trans_per_block on 2017-02-06 is 0.21635457484427884
The real value of percent_change was [[ 0.00904186]] on 2017-02-06
The predicted value of percent_change was [[ 0.00822468]] on 2017-02-06

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.157
Model:                            OLS   Adj. R-squared:                  0.060
Method:                 Least Squares   F-statistic:                     1.629
Date:                Sun, 17 Sep 2017   Prob (F-statistic):              0.121
Time:                        15:28:31   Log-Likelihood:                 189.28
No. Observations:                  89   AIC:                            -358.6
Df Residuals:                      79   BIC:                            -333.7
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0044      0.004      1.154      0.252      -0.003       0.012
x1             0.4293      0.437      0.983      0.329      -0.440       1.298
x2             0.2543      0.105      2.423      0.018       0.045       0.463
x3             0.0301      0.052      0.584      0.561      -0.073       0.133
x4            -0.0610      0.029     -2.106      0.038      -0.119      -0.003
x5            -0.0240      0.118     -0.203      0.839      -0.259       0.211
x6             0.0036      0.066      0.054      0.957      -0.129       0.136
x7             0.0401      0.106      0.378      0.706      -0.171       0.251
x8            -0.0002      0.003     -0.060      0.952      -0.006       0.005
x9             0.0443      0.111      0.399      0.691      -0.177       0.265
==============================================================================
Omnibus:                       50.483   Durbin-Watson:                   1.981
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              214.746
Skew:                          -1.774   Prob(JB):                     2.34e-47
Kurtosis:                       9.732   Cond. No.                         178.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-10 to 2017-02-06 and predicting on 2017-02-07 ----
Creating Time series models: 
The real value for gold_price on 2017-02-07 is 0.007488644269006084
The predicted value for gold_price on 2017-02-07 is 0.0016890021855147392
The real value for eth_price on 2017-02-07 is 0.009234828495864766
The predicted value for eth_price on 2017-02-07 is 0.016138406409366618
The real value for pos_sent on 2017-02-07 is 0.04501734892390363
The predicted value for pos_sent on 2017-02-07 is -0.021940929906275622
The real value for neg_sent on 2017-02-07 is -0.1322891330689302
The predicted value for neg_sent on 2017-02-07 is -0.031490073011630745
The real value for tot_num_trans on 2017-02-07 is 0.41404269933529547
The predicted value for tot_num_trans on 2017-02-07 is 0.060279549123236416
The real value for unique_addr on 2017-02-07 is 0.2656190041529578
The predicted value for unique_addr on 2017-02-07 is 0.030934197417965693
The real value for hash_rate on 2017-02-07 is 0.20799999999999996
The predicted value for hash_rate on 2017-02-07 is 0.0634532869995536
The real value for mempool_trans on 2017-02-07 is 2.1241235785248866
The predicted value for mempool_trans on 2017-02-07 is 1.332147378558802
The real value for avg_trans_per_block on 2017-02-07 is 0.17056514845637039
The predicted value for avg_trans_per_block on 2017-02-07 is 0.016997592696567858
The real value of percent_change was [[ 0.02548428]] on 2017-02-07
The predicted value of percent_change was [[ 0.01263898]] on 2017-02-07

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.164
Model:                            OLS   Adj. R-squared:                  0.069
Method:                 Least Squares   F-statistic:                     1.723
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0976
Time:                        15:28:42   Log-Likelihood:                 189.75
No. Observations:                  89   AIC:                            -359.5
Df Residuals:                      79   BIC:                            -334.6
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0042      0.004      1.104      0.273      -0.003       0.012
x1             0.3745      0.435      0.860      0.392      -0.492       1.241
x2             0.2583      0.104      2.476      0.015       0.051       0.466
x3             0.0233      0.050      0.464      0.644      -0.077       0.123
x4            -0.0677      0.029     -2.326      0.023      -0.126      -0.010
x5            -0.0192      0.117     -0.164      0.870      -0.253       0.215
x6             0.0021      0.066      0.032      0.975      -0.129       0.134
x7             0.0376      0.105      0.358      0.721      -0.172       0.247
x8          6.239e-05      0.003      0.023      0.982      -0.005       0.006
x9             0.0395      0.110      0.358      0.721      -0.180       0.259
==============================================================================
Omnibus:                       49.482   Durbin-Watson:                   2.001
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              209.439
Skew:                          -1.732   Prob(JB):                     3.32e-46
Kurtosis:                       9.670   Cond. No.                         178.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-11 to 2017-02-07 and predicting on 2017-02-08 ----
Creating Time series models: 
The real value for gold_price on 2017-02-08 is 0.003736799350055886
The predicted value for gold_price on 2017-02-08 is 0.0007376829793081517
The real value for eth_price on 2017-02-08 is 0.0008714596951198139
The predicted value for eth_price on 2017-02-08 is 0.034180218555881534
The real value for pos_sent on 2017-02-08 is -0.10386051212741143
The predicted value for pos_sent on 2017-02-08 is 0.02786167160831258
The real value for neg_sent on 2017-02-08 is 0.09729079291134202
The predicted value for neg_sent on 2017-02-08 is -0.033058016672654775
The real value for tot_num_trans on 2017-02-08 is -0.052117023500620974
The predicted value for tot_num_trans on 2017-02-08 is -0.011213472836674784
The real value for unique_addr on 2017-02-08 is -0.07337770681210642
The predicted value for unique_addr on 2017-02-08 is -0.02319779156631299
The real value for hash_rate on 2017-02-08 is -0.03311258278145712
The predicted value for hash_rate on 2017-02-08 is 0.05407903260006898
The real value for mempool_trans on 2017-02-08 is 0.9855193141896446
The predicted value for mempool_trans on 2017-02-08 is 1.1626066164868865
The real value for avg_trans_per_block on 2017-02-08 is -0.019655277730094567
The predicted value for avg_trans_per_block on 2017-02-08 is -0.049732218149905666
The real value of percent_change was [[ 0.00215845]] on 2017-02-08
The predicted value of percent_change was [[ 0.01657069]] on 2017-02-08

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.167
Model:                            OLS   Adj. R-squared:                  0.073
Method:                 Least Squares   F-statistic:                     1.765
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0884
Time:                        15:28:53   Log-Likelihood:                 189.82
No. Observations:                  89   AIC:                            -359.6
Df Residuals:                      79   BIC:                            -334.7
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0043      0.004      1.141      0.257      -0.003       0.012
x1             0.3400      0.440      0.772      0.442      -0.537       1.217
x2             0.2582      0.104      2.477      0.015       0.051       0.466
x3             0.0223      0.050      0.445      0.658      -0.077       0.122
x4            -0.0681      0.029     -2.362      0.021      -0.125      -0.011
x5            -0.0180      0.115     -0.156      0.876      -0.247       0.211
x6             0.0020      0.066      0.030      0.976      -0.129       0.133
x7             0.0368      0.104      0.353      0.725      -0.171       0.244
x8          5.284e-05      0.003      0.019      0.985      -0.005       0.005
x9             0.0376      0.109      0.345      0.731      -0.180       0.255
==============================================================================
Omnibus:                       49.518   Durbin-Watson:                   2.005
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              210.730
Skew:                          -1.730   Prob(JB):                     1.74e-46
Kurtosis:                       9.697   Cond. No.                         184.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-12 to 2017-02-08 and predicting on 2017-02-09 ----
Creating Time series models: 
The real value for gold_price on 2017-02-09 is 0.004977338944708309
The predicted value for gold_price on 2017-02-09 is 0.00048364740837425957
The real value for eth_price on 2017-02-09 is -0.021767522855898225
The predicted value for eth_price on 2017-02-09 is 0.008258750654897059
The real value for pos_sent on 2017-02-09 is -0.0053446282688912605
The predicted value for pos_sent on 2017-02-09 is 0.0009231578394281494
The real value for neg_sent on 2017-02-09 is 0.44823987036991375
The predicted value for neg_sent on 2017-02-09 is 0.09829247798854278
The real value for tot_num_trans on 2017-02-09 is 0.12398967293296481
The predicted value for tot_num_trans on 2017-02-09 is -0.09968306101304696
The real value for unique_addr on 2017-02-09 is 0.11329745009583925
The predicted value for unique_addr on 2017-02-09 is -0.07128774063003206
The real value for hash_rate on 2017-02-09 is 0.15753424657534265
The predicted value for hash_rate on 2017-02-09 is -0.019674105218911896
The real value for mempool_trans on 2017-02-09 is -0.32027343313150036
The predicted value for mempool_trans on 2017-02-09 is 0.1589225168732446
The real value for avg_trans_per_block on 2017-02-09 is -0.02897933580939116
The predicted value for avg_trans_per_block on 2017-02-09 is -0.12107604069104491
The real value of percent_change was [[-0.07247749]] on 2017-02-09
The predicted value of percent_change was [[-0.00385795]] on 2017-02-09

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.168
Model:                            OLS   Adj. R-squared:                  0.073
Method:                 Least Squares   F-statistic:                     1.773
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0866
Time:                        15:29:05   Log-Likelihood:                 189.85
No. Observations:                  89   AIC:                            -359.7
Df Residuals:                      79   BIC:                            -334.8
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0042      0.004      1.115      0.268      -0.003       0.012
x1             0.3844      0.457      0.841      0.403      -0.526       1.294
x2             0.2575      0.104      2.472      0.016       0.050       0.465
x3             0.0204      0.049      0.413      0.681      -0.078       0.118
x4            -0.0684      0.029     -2.375      0.020      -0.126      -0.011
x5            -0.0199      0.115     -0.173      0.863      -0.249       0.209
x6             0.0019      0.066      0.028      0.978      -0.129       0.133
x7             0.0387      0.105      0.370      0.712      -0.169       0.247
x8             0.0002      0.003      0.058      0.954      -0.005       0.006
x9             0.0395      0.109      0.361      0.719      -0.178       0.257
==============================================================================
Omnibus:                       50.021   Durbin-Watson:                   1.996
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              214.364
Skew:                          -1.749   Prob(JB):                     2.83e-47
Kurtosis:                       9.750   Cond. No.                         192.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-13 to 2017-02-09 and predicting on 2017-02-10 ----
Creating Time series models: 
The real value for gold_price on 2017-02-10 is -0.012885041272397868
The predicted value for gold_price on 2017-02-10 is 0.001209295150159607
The real value for eth_price on 2017-02-10 is -0.006675567423161222
The predicted value for eth_price on 2017-02-10 is -0.004651137360806166
The real value for pos_sent on 2017-02-10 is 0.031049222244097452
The predicted value for pos_sent on 2017-02-10 is 0.005506283861076858
The real value for neg_sent on 2017-02-10 is -0.16509544164929146
The predicted value for neg_sent on 2017-02-10 is -0.015523910503048511
The real value for tot_num_trans on 2017-02-10 is -0.12145066715914865
The predicted value for tot_num_trans on 2017-02-10 is 0.005511837885388303
The real value for unique_addr on 2017-02-10 is -0.11751509112062986
The predicted value for unique_addr on 2017-02-10 is -0.0267686593328271
The real value for hash_rate on 2017-02-10 is -0.12426035502958566
The predicted value for hash_rate on 2017-02-10 is -0.0416126967983437
The real value for mempool_trans on 2017-02-10 is -0.4318269290093987
The predicted value for mempool_trans on 2017-02-10 is 0.22885964254121927
The real value for avg_trans_per_block on 2017-02-10 is 0.0032083597979992717
The predicted value for avg_trans_per_block on 2017-02-10 is 0.03718360154919892
The real value of percent_change was [[ 0.0235636]] on 2017-02-10
The predicted value of percent_change was [[ 0.00459717]] on 2017-02-10

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.201
Model:                            OLS   Adj. R-squared:                  0.110
Method:                 Least Squares   F-statistic:                     2.210
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0298
Time:                        15:29:15   Log-Likelihood:                 188.96
No. Observations:                  89   AIC:                            -357.9
Df Residuals:                      79   BIC:                            -333.0
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0036      0.004      0.941      0.350      -0.004       0.011
x1             0.3886      0.462      0.841      0.403      -0.531       1.308
x2             0.2678      0.106      2.537      0.013       0.058       0.478
x3             0.0155      0.050      0.311      0.757      -0.084       0.115
x4            -0.0845      0.027     -3.084      0.003      -0.139      -0.030
x5            -0.0192      0.116     -0.165      0.869      -0.251       0.212
x6            -0.0110      0.067     -0.164      0.870      -0.145       0.123
x7             0.0395      0.106      0.374      0.710      -0.171       0.250
x8             0.0007      0.003      0.268      0.789      -0.005       0.006
x9             0.0447      0.111      0.404      0.688      -0.176       0.265
==============================================================================
Omnibus:                       45.061   Durbin-Watson:                   1.992
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              170.918
Skew:                          -1.593   Prob(JB):                     7.68e-38
Kurtosis:                       8.995   Cond. No.                         192.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-14 to 2017-02-10 and predicting on 2017-02-11 ----
Creating Time series models: 
The real value for gold_price on 2017-02-11 is 0.0
The predicted value for gold_price on 2017-02-11 is -0.0022081679934849462
The real value for eth_price on 2017-02-11 is 0.01433691756258626
The predicted value for eth_price on 2017-02-11 is 0.008130226636322117
The real value for pos_sent on 2017-02-11 is 0.12040889558450929
The predicted value for pos_sent on 2017-02-11 is 0.06221217927301015
The real value for neg_sent on 2017-02-11 is -0.08771127156666525
The predicted value for neg_sent on 2017-02-11 is -0.018875289009422445
The real value for tot_num_trans on 2017-02-11 is -0.1371955225057151
The predicted value for tot_num_trans on 2017-02-11 is -0.04135185048948676
The real value for unique_addr on 2017-02-11 is -0.06455802220974116
The predicted value for unique_addr on 2017-02-11 is -0.006019472502879729
The real value for hash_rate on 2017-02-11 is -0.06081081081081097
The predicted value for hash_rate on 2017-02-11 is 0.02164094405325656
The real value for mempool_trans on 2017-02-11 is -0.4516142376614949
The predicted value for mempool_trans on 2017-02-11 is 0.8608541551357389
The real value for avg_trans_per_block on 2017-02-11 is -0.08133048439457424
The predicted value for avg_trans_per_block on 2017-02-11 is -0.0689068972463127
The real value of percent_change was [[ 0.00975191]] on 2017-02-11
The predicted value of percent_change was [[ 0.00706827]] on 2017-02-11

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.206
Model:                            OLS   Adj. R-squared:                  0.115
Method:                 Least Squares   F-statistic:                     2.276
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0253
Time:                        15:29:25   Log-Likelihood:                 189.07
No. Observations:                  89   AIC:                            -358.1
Df Residuals:                      79   BIC:                            -333.3
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0040      0.004      1.061      0.292      -0.004       0.012
x1             0.3695      0.456      0.810      0.420      -0.538       1.277
x2             0.2650      0.106      2.512      0.014       0.055       0.475
x3             0.0169      0.050      0.340      0.735      -0.082       0.116
x4            -0.0875      0.027     -3.198      0.002      -0.142      -0.033
x5            -0.0143      0.116     -0.123      0.902      -0.245       0.216
x6            -0.0199      0.068     -0.293      0.770      -0.155       0.115
x7             0.0368      0.105      0.350      0.727      -0.173       0.246
x8             0.0007      0.003      0.260      0.795      -0.005       0.006
x9             0.0453      0.110      0.410      0.683      -0.175       0.265
==============================================================================
Omnibus:                       45.904   Durbin-Watson:                   2.036
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              177.792
Skew:                          -1.619   Prob(JB):                     2.47e-39
Kurtosis:                       9.120   Cond. No.                         189.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-15 to 2017-02-11 and predicting on 2017-02-12 ----
Creating Time series models: 
The real value for gold_price on 2017-02-12 is 0.0
The predicted value for gold_price on 2017-02-12 is 0.0029500244935774167
The real value for eth_price on 2017-02-12 is 0.004858657243839382
The predicted value for eth_price on 2017-02-12 is 0.007629078599273193
The real value for pos_sent on 2017-02-12 is -0.022292837238855112
The predicted value for pos_sent on 2017-02-12 is -0.028894386867419405
The real value for neg_sent on 2017-02-12 is -0.12264505057759689
The predicted value for neg_sent on 2017-02-12 is 0.04173400527851562
The real value for tot_num_trans on 2017-02-12 is -0.09328247413261459
The predicted value for tot_num_trans on 2017-02-12 is -0.044387419974355254
The real value for unique_addr on 2017-02-12 is -0.11079557424432662
The predicted value for unique_addr on 2017-02-12 is -0.03688098539636035
The real value for hash_rate on 2017-02-12 is 0.014388489208633226
The predicted value for hash_rate on 2017-02-12 is 0.06478068404705742
The real value for mempool_trans on 2017-02-12 is -0.1881302951867725
The predicted value for mempool_trans on 2017-02-12 is 0.10452588271728602
The real value for avg_trans_per_block on 2017-02-12 is -0.1061437156342796
The predicted value for avg_trans_per_block on 2017-02-12 is -0.06479439403746852
The real value of percent_change was [[-0.00816976]] on 2017-02-12
The predicted value of percent_change was [[ 0.00387023]] on 2017-02-12

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.206
Model:                            OLS   Adj. R-squared:                  0.115
Method:                 Least Squares   F-statistic:                     2.276
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0252
Time:                        15:29:37   Log-Likelihood:                 189.06
No. Observations:                  89   AIC:                            -358.1
Df Residuals:                      79   BIC:                            -333.2
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0040      0.004      1.047      0.299      -0.004       0.012
x1             0.3576      0.493      0.726      0.470      -0.623       1.338
x2             0.2646      0.106      2.505      0.014       0.054       0.475
x3             0.0152      0.050      0.301      0.764      -0.085       0.116
x4            -0.0875      0.027     -3.187      0.002      -0.142      -0.033
x5            -0.0136      0.117     -0.116      0.908      -0.247       0.219
x6            -0.0201      0.068     -0.295      0.769      -0.156       0.116
x7             0.0369      0.106      0.348      0.729      -0.174       0.248
x8             0.0007      0.003      0.268      0.790      -0.005       0.006
x9             0.0450      0.112      0.400      0.690      -0.179       0.268
==============================================================================
Omnibus:                       45.785   Durbin-Watson:                   2.038
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              177.211
Skew:                          -1.614   Prob(JB):                     3.30e-39
Kurtosis:                       9.113   Cond. No.                         204.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-16 to 2017-02-12 and predicting on 2017-02-13 ----
Creating Time series models: 
The real value for gold_price on 2017-02-13 is 0.002977768713101847
The predicted value for gold_price on 2017-02-13 is 0.0026708311640897277
The real value for eth_price on 2017-02-13 is -0.004395604395692532
The predicted value for eth_price on 2017-02-13 is 0.008021364258016831
The real value for pos_sent on 2017-02-13 is -0.08291505672881849
The predicted value for pos_sent on 2017-02-13 is -0.03079435313822866
The real value for neg_sent on 2017-02-13 is -0.05087530986616895
The predicted value for neg_sent on 2017-02-13 is -0.04509932457521735
The real value for tot_num_trans on 2017-02-13 is 0.21037693864355833
The predicted value for tot_num_trans on 2017-02-13 is 0.12110611080178911
The real value for unique_addr on 2017-02-13 is 0.1920849878223292
The predicted value for unique_addr on 2017-02-13 is 0.07916242005291207
The real value for hash_rate on 2017-02-13 is 0.15602836879432602
The predicted value for hash_rate on 2017-02-13 is -0.024940640003463754
The real value for mempool_trans on 2017-02-13 is -0.5483275261324042
The predicted value for mempool_trans on 2017-02-13 is -0.23814213545270346
The real value for avg_trans_per_block on 2017-02-13 is 0.04701318005362998
The predicted value for avg_trans_per_block on 2017-02-13 is 0.1408706124412079
The real value of percent_change was [[-0.00072681]] on 2017-02-13
The predicted value of percent_change was [[ 0.01203572]] on 2017-02-13

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.203
Model:                            OLS   Adj. R-squared:                  0.112
Method:                 Least Squares   F-statistic:                     2.239
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0277
Time:                        15:29:50   Log-Likelihood:                 188.84
No. Observations:                  89   AIC:                            -357.7
Df Residuals:                      79   BIC:                            -332.8
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0035      0.004      0.926      0.357      -0.004       0.011
x1             0.3426      0.495      0.693      0.491      -0.642       1.327
x2             0.2643      0.106      2.497      0.015       0.054       0.475
x3             0.0167      0.051      0.330      0.742      -0.084       0.118
x4            -0.0862      0.028     -3.125      0.002      -0.141      -0.031
x5            -0.0189      0.117     -0.161      0.872      -0.252       0.214
x6            -0.0168      0.068     -0.246      0.806      -0.152       0.119
x7             0.0413      0.106      0.389      0.698      -0.170       0.253
x8             0.0008      0.003      0.281      0.780      -0.005       0.006
x9             0.0493      0.112      0.439      0.662      -0.174       0.273
==============================================================================
Omnibus:                       44.043   Durbin-Watson:                   2.031
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              164.669
Skew:                          -1.556   Prob(JB):                     1.75e-36
Kurtosis:                       8.893   Cond. No.                         204.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-17 to 2017-02-13 and predicting on 2017-02-14 ----
Creating Time series models: 
The real value for gold_price on 2017-02-14 is 0.0002033512282388905
The predicted value for gold_price on 2017-02-14 is 0.007301334055510984
The real value for eth_price on 2017-02-14 is 0.0737306843269463
The predicted value for eth_price on 2017-02-14 is 0.013870242839755964
The real value for pos_sent on 2017-02-14 is 0.11090246625451439
The predicted value for pos_sent on 2017-02-14 is -0.026032442806909473
The real value for neg_sent on 2017-02-14 is 0.08757874336691884
The predicted value for neg_sent on 2017-02-14 is -0.03734493874040487
The real value for tot_num_trans on 2017-02-14 is 0.0929411356620069
The predicted value for tot_num_trans on 2017-02-14 is 0.08802604689036414
The real value for unique_addr on 2017-02-14 is 0.05381787105493285
The predicted value for unique_addr on 2017-02-14 is 0.05095602778764486
The real value for hash_rate on 2017-02-14 is 0.03680981595092048
The predicted value for hash_rate on 2017-02-14 is 0.03284560316863672
The real value for mempool_trans on 2017-02-14 is 0.47517292807734846
The predicted value for mempool_trans on 2017-02-14 is 1.5358133295442926
The real value for avg_trans_per_block on 2017-02-14 is 0.054138491792349974
The predicted value for avg_trans_per_block on 2017-02-14 is 0.07140822065847156
The real value of percent_change was [[ 0.01190433]] on 2017-02-14
The predicted value of percent_change was [[ 0.01577612]] on 2017-02-14

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.203
Model:                            OLS   Adj. R-squared:                  0.112
Method:                 Least Squares   F-statistic:                     2.239
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0277
Time:                        15:30:03   Log-Likelihood:                 189.33
No. Observations:                  89   AIC:                            -358.7
Df Residuals:                      79   BIC:                            -333.8
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0029      0.004      0.771      0.443      -0.005       0.010
x1             0.3611      0.492      0.733      0.465      -0.619       1.341
x2             0.2627      0.105      2.496      0.015       0.053       0.472
x3             0.0184      0.050      0.369      0.713      -0.081       0.118
x4            -0.0846      0.027     -3.091      0.003      -0.139      -0.030
x5            -0.0350      0.117     -0.299      0.765      -0.268       0.198
x6            -0.0090      0.068     -0.132      0.896      -0.145       0.127
x7             0.0474      0.105      0.451      0.654      -0.162       0.257
x8             0.0011      0.003      0.395      0.694      -0.004       0.006
x9             0.0596      0.111      0.536      0.594      -0.162       0.281
==============================================================================
Omnibus:                       44.097   Durbin-Watson:                   2.035
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              167.933
Skew:                          -1.548   Prob(JB):                     3.42e-37
Kurtosis:                       8.974   Cond. No.                         204.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-18 to 2017-02-14 and predicting on 2017-02-15 ----
Creating Time series models: 
The real value for gold_price on 2017-02-15 is -0.003659577928694535
The predicted value for gold_price on 2017-02-15 is -0.0018086357034827946
The real value for eth_price on 2017-02-15 is 0.06743421052628573
The predicted value for eth_price on 2017-02-15 is 0.027105354814292666
The real value for pos_sent on 2017-02-15 is -0.078524073392922
The predicted value for pos_sent on 2017-02-15 is 0.04135923510419166
The real value for neg_sent on 2017-02-15 is -0.06328346678955699
The predicted value for neg_sent on 2017-02-15 is -0.043408892349795086
The real value for tot_num_trans on 2017-02-15 is -0.0644209163270747
The predicted value for tot_num_trans on 2017-02-15 is -0.007138443529484803
The real value for unique_addr on 2017-02-15 is -0.06476890609706187
The predicted value for unique_addr on 2017-02-15 is 0.008504820481417286
The real value for hash_rate on 2017-02-15 is -0.1775147928994083
The predicted value for hash_rate on 2017-02-15 is 0.04195397521395513
The real value for mempool_trans on 2017-02-15 is 0.22875993585274013
The predicted value for mempool_trans on 2017-02-15 is 1.2982690515535835
The real value for avg_trans_per_block on 2017-02-15 is 0.13750262691168613
The predicted value for avg_trans_per_block on 2017-02-15 is -0.01657737160381892
The real value of percent_change was [[ 0.00053938]] on 2017-02-15
The predicted value of percent_change was [[ 0.01621433]] on 2017-02-15

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.204
Model:                            OLS   Adj. R-squared:                  0.113
Method:                 Least Squares   F-statistic:                     2.247
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0271
Time:                        15:30:15   Log-Likelihood:                 189.34
No. Observations:                  89   AIC:                            -358.7
Df Residuals:                      79   BIC:                            -333.8
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0029      0.004      0.779      0.438      -0.005       0.010
x1             0.3797      0.494      0.769      0.444      -0.603       1.363
x2             0.2552      0.102      2.502      0.014       0.052       0.458
x3             0.0160      0.049      0.328      0.744      -0.081       0.113
x4            -0.0859      0.027     -3.153      0.002      -0.140      -0.032
x5            -0.0363      0.117     -0.311      0.757      -0.269       0.196
x6            -0.0085      0.068     -0.124      0.902      -0.144       0.127
x7             0.0470      0.105      0.447      0.656      -0.163       0.257
x8             0.0011      0.003      0.423      0.674      -0.004       0.007
x9             0.0597      0.111      0.537      0.593      -0.162       0.281
==============================================================================
Omnibus:                       44.484   Durbin-Watson:                   2.034
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              171.566
Skew:                          -1.559   Prob(JB):                     5.56e-38
Kurtosis:                       9.045   Cond. No.                         205.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-19 to 2017-02-15 and predicting on 2017-02-16 ----
Creating Time series models: 
The real value for gold_price on 2017-02-16 is 0.00946822838014727
The predicted value for gold_price on 2017-02-16 is -0.001116486102348295
The real value for eth_price on 2017-02-16 is -0.005392912172554776
The predicted value for eth_price on 2017-02-16 is 0.009176661815676303
The real value for pos_sent on 2017-02-16 is 0.09304301885992472
The predicted value for pos_sent on 2017-02-16 is 0.0545586661554608
The real value for neg_sent on 2017-02-16 is -0.1295559184215067
The predicted value for neg_sent on 2017-02-16 is 0.1699740019600633
The real value for tot_num_trans on 2017-02-16 is -0.0320808993642846
The predicted value for tot_num_trans on 2017-02-16 is -0.08555507746783711
The real value for unique_addr on 2017-02-16 is -0.013525013309242273
The predicted value for unique_addr on 2017-02-16 is -0.0841958478591242
The real value for hash_rate on 2017-02-16 is 0.09352517985611519
The predicted value for hash_rate on 2017-02-16 is 0.0012441323878789128
The real value for mempool_trans on 2017-02-16 is 0.4017221934559485
The predicted value for mempool_trans on 2017-02-16 is -0.048369011525873806
The real value for avg_trans_per_block on 2017-02-16 is -0.11486345402391818
The predicted value for avg_trans_per_block on 2017-02-16 is -0.08066256515147247
The real value of percent_change was [[ 0.02260353]] on 2017-02-16
The predicted value of percent_change was [[-0.00992137]] on 2017-02-16

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.197
Model:                            OLS   Adj. R-squared:                  0.106
Method:                 Least Squares   F-statistic:                     2.154
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0342
Time:                        15:30:28   Log-Likelihood:                 189.01
No. Observations:                  89   AIC:                            -358.0
Df Residuals:                      79   BIC:                            -333.1
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0027      0.004      0.720      0.473      -0.005       0.010
x1             0.4299      0.509      0.845      0.401      -0.583       1.443
x2             0.2404      0.101      2.391      0.019       0.040       0.441
x3             0.0204      0.049      0.418      0.677      -0.077       0.117
x4            -0.0843      0.027     -3.077      0.003      -0.139      -0.030
x5            -0.0291      0.121     -0.241      0.810      -0.270       0.211
x6            -0.0067      0.068     -0.097      0.923      -0.143       0.129
x7             0.0426      0.110      0.389      0.699      -0.176       0.261
x8             0.0011      0.003      0.397      0.693      -0.004       0.007
x9             0.0519      0.116      0.449      0.655      -0.178       0.282
==============================================================================
Omnibus:                       43.931   Durbin-Watson:                   2.007
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              167.639
Skew:                          -1.540   Prob(JB):                     3.96e-37
Kurtosis:                       8.976   Cond. No.                         210.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-20 to 2017-02-16 and predicting on 2017-02-17 ----
Creating Time series models: 
The real value for gold_price on 2017-02-17 is 0.0037598544572752246
The predicted value for gold_price on 2017-02-17 is 0.002905478692173356
The real value for eth_price on 2017-02-17 is -0.009295120062030149
The predicted value for eth_price on 2017-02-17 is -0.001736334419482564
The real value for pos_sent on 2017-02-17 is 0.009194456351748759
The predicted value for pos_sent on 2017-02-17 is -0.0030516675932439324
The real value for neg_sent on 2017-02-17 is 0.034050263227086264
The predicted value for neg_sent on 2017-02-17 is 0.006684575465428715
The real value for tot_num_trans on 2017-02-17 is 0.12069575822997303
The predicted value for tot_num_trans on 2017-02-17 is -0.032935465469521676
The real value for unique_addr on 2017-02-17 is 0.12445700945851201
The predicted value for unique_addr on 2017-02-17 is -0.030488945562394273
The real value for hash_rate on 2017-02-17 is -0.006578947368421018
The predicted value for hash_rate on 2017-02-17 is -0.038594173673690085
The real value for mempool_trans on 2017-02-17 is 0.6862346231890659
The predicted value for mempool_trans on 2017-02-17 is 0.07516907638261894
The real value for avg_trans_per_block on 2017-02-17 is 0.1281175844434166
The predicted value for avg_trans_per_block on 2017-02-17 is 0.01777830417375514
The real value of percent_change was [[ 0.01963733]] on 2017-02-17
The predicted value of percent_change was [[ 0.00306839]] on 2017-02-17

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.216
Model:                            OLS   Adj. R-squared:                  0.126
Method:                 Least Squares   F-statistic:                     2.412
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0179
Time:                        15:30:41   Log-Likelihood:                 189.88
No. Observations:                  89   AIC:                            -359.8
Df Residuals:                      79   BIC:                            -334.9
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0022      0.004      0.592      0.555      -0.005       0.010
x1             0.4731      0.497      0.953      0.344      -0.515       1.462
x2             0.2438      0.099      2.450      0.016       0.046       0.442
x3             0.0142      0.048      0.293      0.770      -0.082       0.110
x4            -0.0967      0.028     -3.392      0.001      -0.153      -0.040
x5            -0.0264      0.120     -0.221      0.826      -0.265       0.212
x6            -0.0177      0.068     -0.259      0.796      -0.154       0.118
x7             0.0502      0.109      0.462      0.645      -0.166       0.267
x8             0.0013      0.003      0.498      0.620      -0.004       0.007
x9             0.0574      0.115      0.501      0.618      -0.171       0.285
==============================================================================
Omnibus:                       43.689   Durbin-Watson:                   2.002
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              170.125
Skew:                          -1.519   Prob(JB):                     1.14e-37
Kurtosis:                       9.053   Cond. No.                         207.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-21 to 2017-02-17 and predicting on 2017-02-18 ----
Creating Time series models: 
The real value for gold_price on 2017-02-18 is -2.90878432451791e-14
The predicted value for gold_price on 2017-02-18 is 0.004084620741183704
The real value for eth_price on 2017-02-18 is -0.002345582486354747
The predicted value for eth_price on 2017-02-18 is 0.008968186248915634
The real value for pos_sent on 2017-02-18 is 0.03903711135680843
The predicted value for pos_sent on 2017-02-18 is 0.08280075881788088
The real value for neg_sent on 2017-02-18 is 0.06891375492787533
The predicted value for neg_sent on 2017-02-18 is -0.005317770065096799
The real value for tot_num_trans on 2017-02-18 is -0.09887824572266679
The predicted value for tot_num_trans on 2017-02-18 is -0.054634205943183245
The real value for unique_addr on 2017-02-18 is -0.07214873689262502
The predicted value for unique_addr on 2017-02-18 is -0.02075892424197979
The real value for hash_rate on 2017-02-18 is 0.05099457255516304
The predicted value for hash_rate on 2017-02-18 is -0.0023840566715290154
The real value for mempool_trans on 2017-02-18 is 0.31186080568492813
The predicted value for mempool_trans on 2017-02-18 is 0.7436403561208043
The real value for avg_trans_per_block on 2017-02-18 is -0.10480667831659662
The predicted value for avg_trans_per_block on 2017-02-18 is -0.030093400701045554
The real value of percent_change was [[ 0.0010424]] on 2017-02-18
The predicted value of percent_change was [[ 0.00925666]] on 2017-02-18

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.210
Model:                            OLS   Adj. R-squared:                  0.120
Method:                 Least Squares   F-statistic:                     2.330
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0220
Time:                        15:30:53   Log-Likelihood:                 189.95
No. Observations:                  89   AIC:                            -359.9
Df Residuals:                      79   BIC:                            -335.0
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0027      0.004      0.722      0.473      -0.005       0.010
x1             0.4833      0.496      0.975      0.332      -0.503       1.470
x2             0.2408      0.099      2.425      0.018       0.043       0.438
x3             0.0135      0.048      0.280      0.780      -0.083       0.110
x4            -0.0943      0.029     -3.302      0.001      -0.151      -0.037
x5            -0.0219      0.120     -0.182      0.856      -0.260       0.217
x6            -0.0207      0.069     -0.302      0.764      -0.157       0.116
x7             0.0464      0.109      0.427      0.671      -0.170       0.263
x8             0.0013      0.003      0.484      0.630      -0.004       0.007
x9             0.0560      0.114      0.489      0.626      -0.172       0.284
==============================================================================
Omnibus:                       45.827   Durbin-Watson:                   1.997
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              184.222
Skew:                          -1.597   Prob(JB):                     9.92e-41
Kurtosis:                       9.283   Cond. No.                         208.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-22 to 2017-02-18 and predicting on 2017-02-19 ----
Creating Time series models: 
The real value for gold_price on 2017-02-19 is 3.774758283725532e-15
The predicted value for gold_price on 2017-02-19 is 0.0036595286035374913
The real value for eth_price on 2017-02-19 is 0.002351097178719952
The predicted value for eth_price on 2017-02-19 is 0.014975933425405291
The real value for pos_sent on 2017-02-19 is -0.08625067423607158
The predicted value for pos_sent on 2017-02-19 is -0.046082344896733816
The real value for neg_sent on 2017-02-19 is -0.007564578038472658
The predicted value for neg_sent on 2017-02-19 is 0.011981741151610397
The real value for tot_num_trans on 2017-02-19 is -0.12642070595421118
The predicted value for tot_num_trans on 2017-02-19 is -0.050686158094634864
The real value for unique_addr on 2017-02-19 is -0.10068281787278233
The predicted value for unique_addr on 2017-02-19 is -0.050164309207817476
The real value for hash_rate on 2017-02-19 is -0.1184210526315791
The predicted value for hash_rate on 2017-02-19 is 0.07365234298756476
The real value for mempool_trans on 2017-02-19 is -0.6477900236529919
The predicted value for mempool_trans on 2017-02-19 is 0.1494717218969216
The real value for avg_trans_per_block on 2017-02-19 is -0.009074233619702166
The predicted value for avg_trans_per_block on 2017-02-19 is -0.094633589778181
The real value of percent_change was [[-0.00365107]] on 2017-02-19
The predicted value of percent_change was [[ 0.00677633]] on 2017-02-19

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.209
Model:                            OLS   Adj. R-squared:                  0.119
Method:                 Least Squares   F-statistic:                     2.322
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0224
Time:                        15:31:05   Log-Likelihood:                 189.94
No. Observations:                  89   AIC:                            -359.9
Df Residuals:                      79   BIC:                            -335.0
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0027      0.004      0.712      0.479      -0.005       0.010
x1             0.4777      0.498      0.959      0.341      -0.514       1.470
x2             0.2411      0.099      2.425      0.018       0.043       0.439
x3             0.0145      0.048      0.299      0.766      -0.082       0.111
x4            -0.0938      0.029     -3.282      0.002      -0.151      -0.037
x5            -0.0246      0.120     -0.206      0.838      -0.263       0.214
x6            -0.0207      0.069     -0.302      0.763      -0.157       0.116
x7             0.0487      0.108      0.450      0.654      -0.167       0.264
x8             0.0013      0.003      0.477      0.635      -0.004       0.007
x9             0.0585      0.114      0.512      0.610      -0.169       0.286
==============================================================================
Omnibus:                       45.736   Durbin-Watson:                   1.999
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              183.578
Skew:                          -1.593   Prob(JB):                     1.37e-40
Kurtosis:                       9.273   Cond. No.                         208.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-23 to 2017-02-19 and predicting on 2017-02-20 ----
Creating Time series models: 
The real value for gold_price on 2017-02-20 is -0.004873529885652483
The predicted value for gold_price on 2017-02-20 is 0.0025512762019781264
The real value for eth_price on 2017-02-20 is -0.014855355746808518
The predicted value for eth_price on 2017-02-20 is 0.009933346007107957
The real value for pos_sent on 2017-02-20 is 0.010932761174192063
The predicted value for pos_sent on 2017-02-20 is -0.017569007036137122
The real value for neg_sent on 2017-02-20 is -0.08261838482135109
The predicted value for neg_sent on 2017-02-20 is 0.002752762076574743
The real value for tot_num_trans on 2017-02-20 is 0.12318002868254929
The predicted value for tot_num_trans on 2017-02-20 is 0.09734299956834049
The real value for unique_addr on 2017-02-20 is 0.09793942457265614
The predicted value for unique_addr on 2017-02-20 is 0.09768826287855675
The real value for hash_rate on 2017-02-20 is 0.10447761194029881
The predicted value for hash_rate on 2017-02-20 is -0.047657558345958716
The real value for mempool_trans on 2017-02-20 is 0.17025394557381301
The predicted value for mempool_trans on 2017-02-20 is 0.27331545218882847
The real value for avg_trans_per_block on 2017-02-20 is 0.01693326921257854
The predicted value for avg_trans_per_block on 2017-02-20 is 0.14244092837158442
The real value of percent_change was [[ 0.03037268]] on 2017-02-20
The predicted value of percent_change was [[ 0.00738052]] on 2017-02-20

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.210
Model:                            OLS   Adj. R-squared:                  0.120
Method:                 Least Squares   F-statistic:                     2.335
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0217
Time:                        15:31:18   Log-Likelihood:                 190.00
No. Observations:                  89   AIC:                            -360.0
Df Residuals:                      79   BIC:                            -335.1
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0025      0.004      0.659      0.512      -0.005       0.010
x1             0.4769      0.498      0.958      0.341      -0.514       1.468
x2             0.2400      0.099      2.415      0.018       0.042       0.438
x3             0.0167      0.048      0.348      0.729      -0.079       0.112
x4            -0.0940      0.029     -3.291      0.001      -0.151      -0.037
x5            -0.0248      0.119     -0.208      0.836      -0.262       0.213
x6            -0.0238      0.069     -0.344      0.732      -0.162       0.114
x7             0.0523      0.108      0.484      0.630      -0.163       0.267
x8             0.0012      0.003      0.456      0.649      -0.004       0.007
x9             0.0615      0.114      0.540      0.591      -0.165       0.288
==============================================================================
Omnibus:                       45.199   Durbin-Watson:                   1.999
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              181.019
Skew:                          -1.571   Prob(JB):                     4.92e-40
Kurtosis:                       9.240   Cond. No.                         207.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-24 to 2017-02-20 and predicting on 2017-02-21 ----
Creating Time series models: 
The real value for gold_price on 2017-02-21 is -0.00538308981247515
The predicted value for gold_price on 2017-02-21 is -0.002936936102805177
The real value for eth_price on 2017-02-21 is -0.0075396825396953915
The predicted value for eth_price on 2017-02-21 is 0.017203094548921005
The real value for pos_sent on 2017-02-21 is 0.0005191998979467982
The predicted value for pos_sent on 2017-02-21 is 0.018817977253818723
The real value for neg_sent on 2017-02-21 is 0.09972222848953627
The predicted value for neg_sent on 2017-02-21 is -0.04826016436201865
The real value for tot_num_trans on 2017-02-21 is -0.012867491016100296
The predicted value for tot_num_trans on 2017-02-21 is 0.16852971548188977
The real value for unique_addr on 2017-02-21 is -0.03220941676985112
The predicted value for unique_addr on 2017-02-21 is 0.05379544010743445
The real value for hash_rate on 2017-02-21 is -0.06756756756756754
The predicted value for hash_rate on 2017-02-21 is 0.07426655179523217
The real value for mempool_trans on 2017-02-21 is 1.083248235137415
The predicted value for mempool_trans on 2017-02-21 is 1.1853443434545965
The real value for avg_trans_per_block on 2017-02-21 is 0.05866385021461684
The predicted value for avg_trans_per_block on 2017-02-21 is 0.07983681754264715
The real value of percent_change was [[ 0.03598362]] on 2017-02-21
The predicted value of percent_change was [[ 0.01568364]] on 2017-02-21

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.208
Model:                            OLS   Adj. R-squared:                  0.118
Method:                 Least Squares   F-statistic:                     2.309
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0232
Time:                        15:31:30   Log-Likelihood:                 189.68
No. Observations:                  89   AIC:                            -359.4
Df Residuals:                      79   BIC:                            -334.5
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0029      0.004      0.777      0.439      -0.005       0.010
x1             0.4392      0.500      0.878      0.382      -0.556       1.434
x2             0.2380      0.100      2.385      0.019       0.039       0.437
x3             0.0166      0.048      0.344      0.732      -0.079       0.112
x4            -0.0951      0.029     -3.312      0.001      -0.152      -0.038
x5            -0.0177      0.119     -0.148      0.883      -0.255       0.220
x6            -0.0220      0.069     -0.317      0.752      -0.160       0.116
x7             0.0470      0.108      0.434      0.665      -0.168       0.262
x8             0.0012      0.003      0.453      0.652      -0.004       0.007
x9             0.0537      0.114      0.471      0.639      -0.174       0.281
==============================================================================
Omnibus:                       44.783   Durbin-Watson:                   2.005
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              176.424
Skew:                          -1.561   Prob(JB):                     4.90e-39
Kurtosis:                       9.150   Cond. No.                         205.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

---- Running model from 2016-11-25 to 2017-02-21 and predicting on 2017-02-22 ----
Creating Time series models: 
The real value for gold_price on 2017-02-22 is 0.007162041181643541
The predicted value for gold_price on 2017-02-22 is 0.0018893875747791496
The real value for eth_price on 2017-02-22 is 0.011195521791279806
The predicted value for eth_price on 2017-02-22 is 0.029775605921630362
The real value for pos_sent on 2017-02-22 is -0.04089572845676648
The predicted value for pos_sent on 2017-02-22 is -0.025217525555405074
The real value for neg_sent on 2017-02-22 is 0.07597508924327423
The predicted value for neg_sent on 2017-02-22 is 0.017624908002871864
The real value for tot_num_trans on 2017-02-22 is 0.017678884947508022
The predicted value for tot_num_trans on 2017-02-22 is -0.01266259243512348
The real value for unique_addr on 2017-02-22 is 0.003130894775690951
The predicted value for unique_addr on 2017-02-22 is -0.0030356923410002883
The real value for hash_rate on 2017-02-22 is 0.007246376811594013
The predicted value for hash_rate on 2017-02-22 is -0.007996434991105656
The real value for mempool_trans on 2017-02-22 is 1.442834764817459
The predicted value for mempool_trans on 2017-02-22 is 1.385966705377505
The real value for avg_trans_per_block on 2017-02-22 is 0.01035745412054756
The predicted value for avg_trans_per_block on 2017-02-22 is 0.015776052352259866
The real value of percent_change was [[-0.00050298]] on 2017-02-22
The predicted value of percent_change was [[ 0.01141195]] on 2017-02-22

                            OLS Regression Results                            
==============================================================================
Dep. Variable:                      y   R-squared:                       0.196
Model:                            OLS   Adj. R-squared:                  0.105
Method:                 Least Squares   F-statistic:                     2.145
Date:                Sun, 17 Sep 2017   Prob (F-statistic):             0.0350
Time:                        15:31:41   Log-Likelihood:                 188.61
No. Observations:                  89   AIC:                            -357.2
Df Residuals:                      79   BIC:                            -332.3
Df Model:                           9                                         
Covariance Type:            nonrobust                                         
==============================================================================
                 coef    std err          t      P>|t|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0036      0.004      0.952      0.344      -0.004       0.011
x1             0.3088      0.537      0.575      0.567      -0.761       1.379
x2             0.2322      0.101      2.295      0.024       0.031       0.434
x3             0.0195      0.049      0.395      0.694      -0.079       0.117
x4            -0.0919      0.029     -3.173      0.002      -0.150      -0.034
x5             0.0013      0.127      0.010      0.992      -0.251       0.254
x6            -0.0364      0.078     -0.468      0.641      -0.191       0.118
x7             0.0374      0.110      0.341      0.734      -0.181       0.256
x8             0.0014      0.003      0.489      0.626      -0.004       0.007
x9             0.0462      0.116      0.398      0.692      -0.185       0.277
==============================================================================
Omnibus:                       43.773   Durbin-Watson:                   1.981
Prob(Omnibus):                  0.000   Jarque-Bera (JB):              168.578
Skew:                          -1.529   Prob(JB):                     2.48e-37
Kurtosis:                       9.009   Cond. No.                         219.
==============================================================================

Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.

This model architecture with 10 features has an RMSE score of 0.6835307711047399

In [21]:
r2_score(true_p, pred_p) # this is expected due to the range of values on the y-axis!


Out[21]:
-0.034659636867694132

In [22]:
#very good!
plt.plot(pred_p)
plt.plot(true_p)
plt.legend(["Prediction", 'Actual'], loc='upper left')
plt.xlabel("Prediction #")
plt.ylabel("Price")
plt.title("Nested TS - % Change Prediction");


From the above, it seems that our model is not tuned well enough to anticipate the large dip shown above. This is due to a lack of training data. However, while our model might not be the best in predicting percent change how does it fair when we turn the percent change into prices.


In [23]:
fig, ax = plt.subplots()
ax.scatter(true_p, pred_p, edgecolors=(0, 0, 0))
ax.plot([min(true), max(true)], [min(true), max(true)], 'k--', lw=3)
ax.set_xlabel('Actual')
ax.set_ylabel('Predicted');



In [28]:
df.set_index('date', inplace=True)
prices_to_be_multiplied = df.loc[pd.date_range(start="2017-01-23", end="2017-02-21"), "mkt_price"]
forecast_price_lst = []
for index, price in enumerate(prices_to_be_multiplied):
    predicted_percent_change = 1+float(pred_p[index])
    forecasted_price = (predicted_percent_change)*price
    forecast_price_lst.append(forecasted_price)
ground_truth_prices = df.loc[pd.date_range(start="2017-01-24", end="2017-02-22"), "mkt_price"]
ground_truth_prices = list(ground_truth_prices)
r2_score(ground_truth_prices, forecast_price_lst)


Out[28]:
0.87042670228813868

We have an $R^2$ of 0.87!

This surpasses the baseline model and the nested TS model!

The caveats of the baseline model also apply here, however, it seems that the addition of additional variables have helped us slightly improve with regards to the $R^2$


In [29]:
plt.plot(forecast_price_lst)
plt.plot(ground_truth_prices)
plt.legend(["Prediction", 'Actual'], loc='upper left')
plt.xlabel("Prediction #")
plt.ylabel("Price")
plt.title("Nested TS - % Change Prediction");