In [46]:
%matplotlib inline
from pymongo import MongoClient
import math
import datetime
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
sns.set(color_codes=True)
In [47]:
def divtd(td1, td2):
us1 = td1.microseconds + 1000000 * (td1.seconds + 86400 * td1.days)
us2 = td2.microseconds + 1000000 * (td2.seconds + 86400 * td2.days)
return float(us1) / us2
def is_outlier(points, thresh=3.5):
"""
Returns a boolean array with True if points are outliers and False
otherwise.
Parameters:
-----------
points : An numobservations by numdimensions array of observations
thresh : The modified z-score to use as a threshold. Observations with
a modified z-score (based on the median absolute deviation) greater
than this value will be classified as outliers.
Returns:
--------
mask : A numobservations-length boolean array.
References:
----------
Boris Iglewicz and David Hoaglin (1993), "Volume 16: How to Detect and
Handle Outliers", The ASQC Basic References in Quality Control:
Statistical Techniques, Edward F. Mykytka, Ph.D., Editor.
"""
if len(points.shape) == 1:
points = points[:,None]
median = np.median(points, axis=0)
diff = np.sum((points - median)**2, axis=1)
diff = np.sqrt(diff)
med_abs_deviation = np.median(diff)
modified_z_score = 0.6745 * diff / med_abs_deviation
return modified_z_score > thresh
try:
import config
print('using local config file')
mongodb_uri = config.mongodb_uri
except:
print('using environment variable')
mongodb_uri = os.getenv('MONGODB_URI')
In [48]:
print("Analyzing Data")
success = True
if success:
try:
client = MongoClient(mongodb_uri)
db = client.get_database()
except Exception as e:
print('mongo login error ', str(e))
success = False
if success:
try:
dataCursor = db.percentageMove.find(projection={'_id': False},sort=[("timestamp", 1)])
except Exception as e:
print('mongo data cursor error ', str(e))
success = False
if success:
try:
df = pd.DataFrame(list(dataCursor)).set_index(['timestamp'])
filteredDf = df[~is_outlier(df)]
increase = df+1
cumulative = increase.cumprod(axis=0)
cumulative = cumulative - cumulative.iloc[0]
except Exception as e:
print('mongo data cursor error ', str(e))
success = False
if success:
try:
dataCursor = db.tracking.find(projection={'_id': False},sort=[("timestamp", 1)])
except Exception as e:
print('mongo data cursor error ', str(e))
success = False
if success:
try:
tracking = pd.DataFrame(list(dataCursor)).set_index(['timestamp'])
except Exception as e:
print('mongo data cursor error ', str(e))
success = False
if success:
try:
dataCursor = db.rawPrices.find(projection={'_id': False},sort=[("timestamp", 1)])
except Exception as e:
print('mongo data cursor error ', str(e))
success = False
if success:
try:
raw = pd.DataFrame(list(dataCursor)).set_index(['timestamp'])
except Exception as e:
print('mongo data cursor error ', str(e))
success = False
In [49]:
if success:
plt.figure()
plt.title('returns')
plt.plot(df['portfolio'])
plt.plot(df['spy'])
plt.plot(df['tlt'])
plt.plot(df['90dayTreasury'])
plt.legend()
plt.figure()
plt.title('filtered returns distribution')
sns.distplot(filteredDf['portfolio'],label="portfolio")
sns.distplot(filteredDf['spy'],label="spy")
sns.distplot(filteredDf['tlt'],label="tlt")
plt.legend()
plt.figure()
plt.title('cumulative returns')
plt.plot(cumulative['portfolio'])
plt.plot(cumulative['spy'])
plt.plot(cumulative['tlt'])
plt.plot(cumulative['90dayTreasury'])
plt.legend()
plt.figure()
plt.title('price vs returns error')
plt.plot(cumulative['portfolio']-(raw['portfolio']/raw['portfolio'].iloc[0])-1)
plt.plot(cumulative['spy']-(raw['spy']/raw['spy'].iloc[0])-1)
plt.plot(cumulative['tlt']-(raw['tlt']/raw['tlt'].iloc[0])-1)
plt.legend()
plt.figure()
plt.title('algorithm tracking')
plt.plot(tracking['spyTarget'])
plt.plot(tracking['tltTarget'])
plt.plot(tracking['spyActual'])
plt.plot(tracking['tltActual'])
plt.legend()
In [50]:
if success:
print('original data \n')
print(df.describe())
print('\n filtered data \n')
print(filteredDf.describe())
df['elapsed'] = df.index-df.index[0]
year = datetime.timedelta(days=365)
yearPeriods = divtd(year,(df['elapsed'])[-1])
df['spyExcess'] = df['spy']-df['90dayTreasury']
df['tltExcess'] = df['tlt']-df['90dayTreasury']
df['portfolioExcess'] = df['portfolio']-df['90dayTreasury']
df['spyInfoExcess'] = df['portfolio']-df['spy']
df['tltInfoExcess'] = df['portfolio']-df['tlt']
meanSpyExcess = df['spyExcess'].mean()
meanTltExcess = df['tltExcess'].mean()
meanPortfolioExcess = df['portfolioExcess'].mean()
meanSpyInfoExcess = df['spyInfoExcess'].mean()
meanTltInfoExcess = df['tltInfoExcess'].mean()
stdSpyExcess = df['spyExcess'].std()
stdTltExcess = df['tltExcess'].std()
stdPortfolioExcess = df['portfolioExcess'].std()
stdSpyInfoExcess = df['spyInfoExcess'].std()
stdTltInfoExcess = df['tltInfoExcess'].std()
rawSpySharpe = meanSpyExcess/stdSpyExcess
spySharpe = math.sqrt(yearPeriods)*rawSpySharpe
print('\nrawSpySharpe = ',rawSpySharpe)
print('spySharpe = ',spySharpe)
rawTltSharpe = meanTltExcess/stdTltExcess
tltSharpe = math.sqrt(yearPeriods)*rawTltSharpe
print('\nrawTltSharpe = ',rawTltSharpe)
print('tltSharpe = ',tltSharpe)
rawPortfolioSharpe = meanPortfolioExcess/stdPortfolioExcess
portfolioSharpe = math.sqrt(yearPeriods)*rawPortfolioSharpe
print('\nrawPortfolioSharpe = ',rawPortfolioSharpe)
print('portfolioSharpe = ',portfolioSharpe)
rawSpyInfo = meanSpyInfoExcess/stdSpyInfoExcess
spyInfo = math.sqrt(yearPeriods)*rawSpyInfo
print('\nrawSpyInfo = ',rawSpyInfo)
print('spyInfo = ',spyInfo)
rawTltInfo = meanTltInfoExcess/stdTltInfoExcess
tltInfo = math.sqrt(yearPeriods)*rawTltInfo
print('\nrawTltInfo = ',rawTltInfo)
print('tltInfo = ',tltInfo)