quantopian / alphalens

Performance analysis of predictive (alpha) stock factors
http://quantopian.github.io/alphalens
Apache License 2.0
3.26k stars 1.13k forks source link

API: breaking apart factor_data into fwd returns and factor value/quantile #276

Open luca-s opened 6 years ago

luca-s commented 6 years ago

I am copying/pasting here the comment by @ssanderson (from here) which requires a dedicated discussion thread

As an aside, in the long term I'm not sure I'm convinced we need combining logic at all. I commented in #258 (comment) that the current factor_data layout forces a lot of duplication if you want to analyze multiple factors.

Depending on how important a use-case it is to analyze multiple factors (and it seems like the desire to efficiently analyze multiple factors is one of the motiviating ideas for this thread), I think we might want to consider breaking apart the monolithic factor_data format into returns + group columns (which can be shared across multiple analyses) and factor-specific factor value/quantile columns . That's a much larger API change than what we've been discussing so far though, so it probably deserves its own discussion thread.