In [1]:

    

%run ../../code/version_check.py


    

    




Python: 3.5.2 |Continuum Analytics, Inc.| (default, Jul  2 2016, 17:53:06) 
[GCC 4.4.7 20120313 (Red Hat 4.4.7-1)]

matplotlib:	1.5.3
numpy:		1.11.2
pandas:		0.19.0
quandl:		3.0.1
sklearn:	0.18
scipy:		0.18.1

In [2]:

    

%matplotlib inline
import matplotlib.pyplot as plt
import pandas as pd
import quandl


    

In [3]:

    

df = quandl.get("CFTC/FV_FO_ALL")


    

In [4]:

    

cols = ['Open Interest','Dealer Longs','Dealer Shorts',]
com = df[cols]
com.head()


    

    
Out[4]:






  

    

      

      
Open Interest
      
Dealer Longs
      
Dealer Shorts
    
    

      
Date
      

      

      

    
  
  

    

      
2006-06-13
      
1381214.0
      
132291.0
      
146427.0
    
    

      
2006-06-20
      
1466695.0
      
166343.0
      
146151.0
    
    

      
2006-06-27
      
1439152.0
      
153632.0
      
151381.0
    
    

      
2006-07-03
      
1396721.0
      
152118.0
      
127671.0
    
    

      
2006-07-11
      
1390896.0
      
151812.0
      
152756.0
    
  

In [5]:

    

com.plot()


    

    
Out[5]:





<matplotlib.axes._subplots.AxesSubplot at 0x7f1774698470>






    

In [6]:

    

# adj = com[['Dealer Longs', 'Dealer Shorts']] / com['Open Interest']
adj = com.div(com['Open Interest'], axis='index').drop('Open Interest', 1)
adj.plot(title='Adjusted Commerical Longs and Shorts')


    

    
Out[6]:





<matplotlib.axes._subplots.AxesSubplot at 0x7f17719f16d8>






    

In [7]:

    

adj_net = (com['Dealer Longs'] - com['Dealer Shorts']) / com['Open Interest']
print(adj_net.head(1))
adj_net.plot()


    

    




Date
2006-06-13   -0.010234
dtype: float64






    
Out[7]:





<matplotlib.axes._subplots.AxesSubplot at 0x7f1771726f60>






    

In [8]:

    

def cot(df):
    max = df.max()
    min = df.min()
    current = df[len(df) -1]
    
    return (current - min) / (max - min)


    

In [9]:

    

com_index = adj_net.rolling(window=26, min_periods=26).apply(func=cot)


    

In [10]:

    

producer_index = adj['Dealer Shorts'].rolling(window=26, min_periods=26).apply(func=cot)


    

In [11]:

    

consumer_index = adj['Dealer Longs'].rolling(window=26, min_periods=26).apply(func=cot)


    

In [12]:

    

mu = adj_net.mean()
mu


    

    
Out[12]:





-0.008084933986470812

In [13]:

    

sigma = adj_net.std()
sigma


    

    
Out[13]:





0.07104747517353989

In [14]:

    

u_limit = mu + (2 * sigma)
u_limit


    

    
Out[14]:





0.13401001636060897

In [15]:

    

l_limit = mu - (2 * sigma)
l_limit


    

    
Out[15]:





-0.15017988433355059

In [16]:

    

fig, ax = plt.subplots(nrows=4, ncols=1, figsize=(12,8), sharex=True)

adj_net.plot(color='blue',
          grid=True,
          linewidth=1,
          yticks=[u_limit, mu, l_limit],
          title='Commercial Control Chart (2 s.d.)',
          ax=ax[0])

com_index.plot(color='blue',
               grid=True,
               linewidth=1,
               yticks=[0.2, 0.8],
               title='Commercial Index (Net Pressure)',
               ax=ax[1])

producer_index.plot(color='blue',
               grid=True,
               linewidth=1,
               yticks=[0.2, 0.8],
               title='Producer Index (Selling Pressure)',
               ax=ax[2])

consumer_index.plot(color='blue',
               grid=True,
               linewidth=1,
               yticks=[0.2, 0.8],
               title='Consumer Index (Buying Pressure)',
               ax=ax[3])

for i in range(len(ax)):
    ax[i].spines['top'].set_visible(False)
    ax[i].spines['right'].set_visible(False)
    ax[i].legend().set_visible(False)


    

In [17]:

    

diff = adj_net.diff()
diff.plot(figsize=(12,2))


    

    
Out[17]:





<matplotlib.axes._subplots.AxesSubplot at 0x7f17717c7438>






    

In [18]:

    

from pymc3 import Exponential, StudentT, Deterministic, Model
from pymc3.math import exp
from pymc3.distributions.timeseries import GaussianRandomWalk

with Model() as sp500_model:
    nu = Exponential('nu', 1./10, testval=5.)
    sigma = Exponential('sigma', 1./.02, testval=.1)
    s = GaussianRandomWalk('s', sigma**-2, shape=len(diff))
    volatility_process = Deterministic('volatility_process', exp(-2*s))
    r = StudentT('r', nu, lam=1/volatility_process, observed=diff)


    

    




---------------------------------------------------------------------------
ImportError                               Traceback (most recent call last)
<ipython-input-18-28a09f5dbed9> in <module>()
----> 1 from pymc3 import Exponential, StudentT, Deterministic, Model
      2 from pymc3.math import exp
      3 from pymc3.distributions.timeseries import GaussianRandomWalk
      4 
      5 with Model() as sp500_model:

ImportError: No module named 'pymc3'

In [ ]:

    

from pymc3 import variational

import scipy
with sp500_model:
    mu, sds, elbo = variational.advi(n=100000)
    step = NUTS(scaling=sp500_model.dict_to_array(sds)**2, is_cov=True)

    trace = sample(2000, step, start=mu, progressbar=True)


    

In [ ]:

    

traceplot(trace[200:], [nu, sigma]);


    

In [ ]:

    

import quandl.version as v


    

In [ ]:

    

quandl.version.VERSION