Back to Community
Help with Alphalens on Infrequent Data

Hi All,

I am having some issues with using alphalens to analyze alpha factors, particularly economic ones that are released infrequent i.e. Weekly or monthly. Can any please help me out into combining the adp data in the worksheet into the 'class' that would be able to fit into the pipeline & feed alphalens? Its fine for the universe to just be SPY for the example.

Thanks for any help.

Loading notebook preview...
Notebook previews are currently unavailable.
3 responses

I don't believe this is a good example of Alphalens usage because you don't have a factor value associated with each security in your universe, instead you only have one time series adp_empl_sec. Why not simply testing the correlation between adp_empl_sec and the market? What kind of analysis would you like to do with Alphalens?

Predominantly I am after the walk-forward returns (correlation/IC across multiple time frames). I would like to see if the factors are predictive overtime frames from 1-5 days to 1-3 months. Then I would like to test it over multiple sectors. Understand the quintile ranking & most other functions are not applicable here.

I would like to see if the factors are predictive overtime frames
from 1-5 days to 1-3 months. Then I would like to test it over
multiple sectors.

But you don't have a factor, you only have a single time series of information while you need a factor value computed for each stock if you like to run Alphalens. You might use adp_empl_sec to build an event study maybe but that's a different story.