Improving factor momentum: Statistical significance matters

摘要

Factor selection in the crowded "factor zoo"presents a significant challenge. This study introduces the statistical factor momentum (SFMOM), a novel approach employing pairwise t-test procedures to adeptly balance Type I and Type II errors, thereby enhancing factor momentum. Through empirical analysis of 207 factors, we demonstrate SFMOM's superior performance, particularly in long-short portfolios. SFMOM prefers low-volatility factors and its effectiveness is most pronounced during periods of substantial dispersion in factors' risk-adjusted performance. Our study offers a new perspective on factor selection and a practical tool for portfolio managers, and the methodology can be applied to other markets.

关键词