Back to Community
Is it wise to combine related factors to get a higher alpha?


I'm looking at two factors right now.

  1. fcf_per_share
  2. free_cf_fcfe_qf

Both are related to free cash flow. When researched separately, the former gives me a 1-D IC Mean of 0.018 and the latter gives 0.017.

When combined, however, I get 0.019.

The combined alpha is larger, but the factors are related. Does this matter?

2 responses

They are related in that they both use fcf, but the first one on it's own I'm not sure makes much sense. You'd get a size tilt and possibly a tilt to companies with fewer shares outstanding (per marketcap), which I'm not sure is desired. Comparing fcf_per_share to price per share or book value per share (or assets per share) etc would make more sense. The second one might be ok to use on it's own as it compares total fcf to total equity, though you might be getting some seasonality tilt as it's quarterly fcf being used. Personally I would try to remove these tilts from both fields and see what their mean IC is on their own as well as when combined, as well as checking if the mean IC is statistically significant.

Thanks for the answer.

Regarding the size tilt, I feel it can go either way. I'd think that many small businesses have less fcf because they are more heavily investing their cash in future growth, at those earlier stages. If you have more fcf because you are a larger company (and with lots of shares), at least your "fcf score" is balanced out by dividing it by num of shares. But using simply fcf may tilt toward larger companies. I guess the solution is to incorporate both to get rid of tilts?

As for the seasonality tilt, I agree that using quarterly figures could contribute to this tilt. I'll look into removing this tilt, thanks.