Sorry if it is a newby question, im new to quantopian.
Im trying to replicate the alphales results to understand them, but I'm not beeing able to do it.
factor_data = run_pipeline(make_pipeline(), startDay, endDay) #build factors
dfAAPL = get_pricing(symbols('AAPL'), start_date=startDay, end_date=endDay) #get apple Pricing
aaplResults = factor_data.xs(symbols('AAPL'), level=1) #get just apple factors
df = dfAAPL.join(aaplResults) # join aaple factors + price
I asume the returns are calculated:
Future Open / Today Open - 1
So I make the calculation
days = [1,2,3,4,5,10,15,20]
for d in days:
returns ='{}DRetun'.format(d)
df[returns] = df['open_price'].shift(-d)/df['open_price']
res = df[df['Factor'] == 1]
print(returns,(res[returns].mean()-1)*100 ,(res[returns].mean()-1)*100/d) #maybe its was the mean return by days?
Foward, Return, Mean Days Return
('1DRetun', 0.24037338368763095, 0.24037338368763095)
('2DRetun', -0.18177306856032027, -0.09088653428016014)
('3DRetun', 0.28722005736183664, 0.09574001912061221)
('4DRetun', 0.5946041417401515, 0.14865103543503788)
('5DRetun', 0.8046724143346617, 0.16093448286693235)
('10DRetun', 1.1221817218125096, 0.11221817218125096)
('15DRetun', 1.2784627411053462, 0.08523084940702308)
('20DRetun', 2.4049766319204746, 0.12024883159602373)
But the pipeline Results are
pricing_data = get_pricing(
symbols=factor_data.index.levels[1], # Finds all assets that appear at least once in "factor_data"
start_date=startDay,
end_date=endDay, # must be after run_pipeline()'s end date. Explained more in lesson 4
fields='open_price' # Generally, you should use open pricing. Explained more in lesson 4
)
merged_data = get_clean_factor_and_forward_returns(
factor=factor_data,
prices=pricing_data,
quantiles=None,
bins=[-2,-1,0,1,2],
periods = [1,2,3,4,5,10,15,20]
)
create_full_tear_sheet(merged_data)
---------------------------------------------------- -1D-----2D-----3D------4D------5D----10D----15D----20D
Mean Period Wise Return Top Quantile (bps) -0.080 / -1.058 / 1.157 / 1.136 / 1.428 / 1.568 / 0.871 / 0.812
I dont get it.
Is there some place where I can see what the alphalens results mean and how are they calculated?
What is Ann. alpha? What is beta? I know beta is how much my performances is linked to the performance of the market