def get_equity_returns(bundle, data_dates, run_dates):
"""Gets the close price for all assets over all trading days in run_dates."""
pipe = Pipeline(columns={'Close': USEquityPricing.close.latest}, )
# create the pipeline engine
spe = make_pipeline_engine(bundle, data_dates)
# stocks = spe.run_pipeline(pipe, run_dates[0], run_dates[1])
stocks = spe.run_pipeline(pipe, run_dates[0], run_dates[1])
unstacked_results = stocks.unstack() # what does this do?
prices = (unstacked_results['Close'].fillna(method='ffill').fillna(
method='bfill').dropna(axis=1, how='any').shift(periods=-1).dropna())
return prices.pct_change() #[1:]
from zipline.data.bundles.core import register
from zipline.pipeline import Pipeline
from zipline.pipeline.data import USEquityPricing
from alphacompiler.util.zipline_data_tools import make_pipeline_engine
def str2dt(datestr):
return pd.to_datetime(datestr, utc=True)
# constants
BUNDLE = 'crsp'
data_dates = ('2015-01-06', '2015-01-30')
backtest_dates = ('2015-01-06', '2015-01-30')
pipeline_data_dates = (pd.to_datetime(data_dates[0], utc=True), pd.to_datetime(data_dates[1], utc=True))
# Step 1. Run Pipeline
# 1.0 dummy bundle register
register(BUNDLE, int) # dummy register of a bundle
# 1.1 create the pipeline engine
spe = make_pipeline_engine(BUNDLE, pipeline_data_dates)
# 1.2 create your pipeline (this could be more elaborate)
pipe = Pipeline(columns={'Close': USEquityPricing.close.latest},)
# 1.3 run your pipeline with the pipeline engine
stocks = spe.run_pipeline(pipe, str2dt(backtest_dates[0]), str2dt(backtest_dates[1]))
print stocks
def get_style_cov(bundle, data_dates, run_dates):
"""Calculates the style covariance matrix."""
my_pipeline = make_pipeline(
Momentum) # Momentum is just placeholder to reuse code
spe = make_pipeline_engine(bundle, data_dates)
results = spe.run_pipeline(my_pipeline,
pd.to_datetime(run_dates[0], utc=True),
pd.to_datetime(run_dates[1], utc=True))
results.drop(['my_factor', 'my_longs', 'my_shorts'], axis=1)
# print results.head()
# mean returns of the biggest/smallest by market cap
R_biggest = results[results.biggest]['returns'].groupby(level=0).mean()
R_smallest = results[results.smallest]['returns'].groupby(level=0).mean()
R_highpb = results[results.highpb]['returns'].groupby(level=0).mean()
R_lowpb = results[results.lowpb]['returns'].groupby(level=0).mean()
R_low_momentum = results[results.low_momentum]['returns'].groupby(
level=0).mean()
R_high_momentum = results[results.high_momentum]['returns'].groupby(
level=0).mean()
R_lowvol = results[results.lowvol]['returns'].groupby(level=0).mean()
R_highvol = results[results.highvol]['returns'].groupby(level=0).mean()
R_low_streversal = results[results.low_streversal]['returns'].groupby(
level=0).mean()
R_high_streversal = results[results.high_streversal]['returns'].groupby(
level=0).mean()
SMB = R_smallest - R_biggest
HML = R_highpb - R_lowpb
MOMENTUM = R_high_momentum - R_low_momentum
VOL = R_highvol - R_lowvol
STREVERSAL = R_high_streversal - R_low_streversal
smb_n_hml = pd.DataFrame(
{
'SMB': SMB, # company size
'HML': HML, # company PB ratio value
'MOMENTUM': MOMENTUM,
'VOL': VOL,
'STREVERSAL': STREVERSAL # short term reversal
},
columns=["SMB", "HML", "MOMENTUM", "VOL",
"STREVERSAL"]).shift(periods=-1).dropna()
# get SPY data (not included in bundle)
spy_series = pd.read_csv(SPY_PATH,
index_col=0,
parse_dates=True,
usecols=[0, 4],
date_parser=date_utc)
assert "close" in spy_series.columns
MKT = spy_series.pct_change()[1:].rename(columns={"close":
"MKT"}) # market returns
F = pd.concat([MKT, smb_n_hml], axis=1).dropna()
print(F)
# calculate cov
return F.cov()
def get_factor_returns(alpha_str, bundle, data_dates, run_dates):
"""Calculates factor returns."""
# create the pipeline engine
spe = make_pipeline_engine(bundle, data_dates)
my_pipeline = make_pipeline()
results = spe.run_pipeline(my_pipeline, run_dates[0], run_dates[1])
print(results.head())
# mean returns of the biggest/smallest by market cap
R_biggest = results[results.biggest]['returns'].groupby(level=0).mean()
R_smallest = results[results.smallest]['returns'].groupby(level=0).mean()
R_highpb = results[results.highpb]['returns'].groupby(level=0).mean()
R_lowpb = results[results.lowpb]['returns'].groupby(level=0).mean()
R_low_momentum = results[results.low_momentum]['returns'].groupby(
level=0).mean()
R_high_momentum = results[results.high_momentum]['returns'].groupby(
level=0).mean()
R_lowvol = results[results.lowvol]['returns'].groupby(level=0).mean()
R_highvol = results[results.highvol]['returns'].groupby(level=0).mean()
R_low_streversal = results[results.low_streversal]['returns'].groupby(
level=0).mean()
R_high_streversal = results[results.high_streversal]['returns'].groupby(
level=0).mean()
R_longs = results[results.my_longs]['returns'].groupby(level=0).mean()
R_shorts = results[results.my_shorts]['returns'].groupby(level=0).mean()
SMB = R_smallest - R_biggest
HML = R_highpb - R_lowpb
MOMENTUM = R_high_momentum - R_low_momentum
VOL = R_highvol - R_lowvol
STREVERSAL = R_high_streversal - R_low_streversal
MyReturns = R_longs - R_shorts #
smb_n_hml = pd.DataFrame(
{
'SMB': SMB, # company size
'HML': HML, # company PB ratio value
'MOMENTUM': MOMENTUM,
'VOL': VOL,
'STREVERSAL': STREVERSAL, # short term reversal
'MyReturns': MyReturns
},
columns=["SMB", "HML", "MOMENTUM", "VOL", "STREVERSAL",
"MyReturns"]).shift(periods=-1).dropna()
# # get SPY data (not included in quantopian-quandl bundle)
# spy_series = pd.read_csv(SPY_PATH, index_col=0, parse_dates=True, usecols=[0, 4], date_parser=date_utc)
# spy_series = spy_series[spy_series.index >= run_dates[0]] # added to only use same dates as data from the Pipeline
#
# assert "close" in spy_series.columns
# MKT = spy_series.pct_change()[1:].rename(columns={"close": "MKT"}) # market returns
#
# return pd.concat([MKT, smb_n_hml], axis=1).dropna()
return smb_n_hml
