Sensitivities of Multi-factor Stock Portfolio Performance

Why do portfolios formed from the principal components of many long-short stock return factors from two recent studies, one covering 207 factors and the other 153 factors (with overlap 97), have such different out-of-sample gross Sharpe ratios? In their November 2022 paper entitled “Factor Returns and Out-of-Sample Alphas: Factor Construction Matters”, Hendrik Bessembinder, Aaron Burt and Christopher Hrdlicka explore reasons for this divergence, such as differences in weighting methods used for individual factor portfolios, number of quantiles employed to construct these portfolios and differences in the total number of factors considered. Using descriptions/data from the prior studies spanning a total of 263 distinct factors with at least some factors in every month during 1926 through 2020, they find that:

(more…)