Import all of the needed classes into code
In [28]:
from quantopian.research import run_pipeline, returns
from quantopian.research.experimental import get_factor_loadings, get_factor_returns
from quantopian.pipeline import Pipeline
from quantopian.pipeline.filters import QTradableStocksUS
from quantopian.pipeline.factors import ExponentialWeightedMovingAverage
from quantopian.pipeline.data.builtin import USEquityPricing
from quantopian.pipeline.classifiers.fundamentals import Sector
import alphalens as al
import numpy as np
import matplotlib as plt
import pyfolio as pf
import math
Set the stock universe we are wishing to work with
In [29]:
universe = QTradableStocksUS()
Create Factor
In [30]:
factor1 = ExponentialWeightedMovingAverage(inputs=[USEquityPricing.close], window_length=5, decay_rate=0.5)
Create the pipeline of dynamic stocks
In [31]:
pipe = Pipeline(
columns = {
'factor': factor1,
'sector': Sector(mask=universe),
},
screen = universe
)
Run the pipeline to get results
In [32]:
factors = run_pipeline(pipe, '2005-01-01', '2010-12-31')
Drop all factors that doesn't return a value
In [33]:
factors = factors.dropna()
Get the assets returned
In [34]:
assets = factors.index.levels[1]
my_returns = returns(assets, '2005-01-01', '2010-12-31')
Get prices for stocks
In [35]:
prices = get_pricing(assets, start_date='2004-12-01', end_date='2010-01-31', fields='close_price')
Set Sectors, Get Returns, and handle potential bad data
In [36]:
sector_labels = dict(Sector.SECTOR_NAMES)
sector_labels[-1] = "Unknown"
Create Alphalens Analysis
In [37]:
factor_data = al.utils.get_clean_factor_and_forward_returns(
factor=factors["factor"],
prices=prices,
groupby=factors["sector"],
quantiles=5,
periods = (1, 5, 10)
)
Create Pyfolio Analysis
In [38]:
factor_returns, factor_positions, factor_benchmark = al.performance.create_pyfolio_input(
factor_data,
period = '1D',
capital = 100000000,
long_short = True,
group_neutral = False,
equal_weight = True,
quantiles = [1, 5],
groups = None,
benchmark_period = '1D'
)
Plot the returns
In [39]:
factor_data.plot()
Out[39]:
Check correalation
In [40]:
np.corrcoef(factor_returns)
Out[40]:
Check risk factor of Quantopian risk model to our factor in pyfolio
In [41]:
start_date = factor_data.index.levels[0].min()
end_date = factor_data.index.levels[0].max()
factor_loadings = get_factor_loadings(assets, start_date, end_date)
factor_returns = get_factor_returns(start_date, end_date)
In [44]:
pf.tears.create_perf_attrib_tear_sheet(
returns = assets,
positions = factor_positions,
factor_returns = factor_returns,
factor_loadings = factor_loadings,
pos_in_dollars = False,
)