Back to Community
Factor Combination Theory, Tools, and Examples

The purpose of this post is to provide a simulated, educational example of analyzing combinations of factors. It also illustrates combining factors that may have "multiplicative interactions." In other words, the marginal effect of predicted forward returns for each factor will depend on the value of the other factor.

The other purpose of this thread is to allow for the community to brainstorm potential tools for a standardized "factor combination/interaction" tear sheet.

Loading notebook preview...
2 responses

Wow! So awesome! Thank you for sharing this! It will take me some time to digest and understand it all.

Do you know if one would need Alphalens available in the IDE in order to get the predictability (mean IC) of a factor in a strategy to dynamically weigh predictable factors more appropriately? Or is there some other way of getting the mean IC of a factor for a specified look-back window?

You might could do it, but it would be tricky. I guess one way would be to create a separate custom factor that calculates the factor of interest as of "x" days ago where "x" is your holding period. You could then calculate a spearman rank correlation coefficient of the factor values with the x day returns calculated today using the history method. Append this IC value to a list. Then after a certain number of days (determined by the user), you could calculate the mean IC.