Double-Adjusted Mutual Fund Alpha

Article BigQuant

Summary

The article reviews a method for evaluating active mutual funds by adjusting performance first for factor exposures and then for the characteristics of stocks held. It begins with Carhart four-factor alpha, then uses either a cross-sectional regression on weighted size, value, and momentum characteristics or a matched-portfolio approach to remove the characteristics-driven component. The remaining double-adjusted alpha is intended to better isolate manager performance. The article reports that holdings characteristics explain a meaningful share of traditional alpha and can materially change relative fund rankings.

Evidence summarized from U.S. fund data suggests that double-adjusted rankings predict future four-factor alpha more strongly than standard alpha rankings, including in some long-horizon tests, while the holdings-characteristic component itself shows little positive predictive relation. Long-short portfolios ranked by double-adjusted performance also show higher reported Sharpe and information ratios than those ranked by standard alpha. These findings are based on historical U.S. mutual fund evidence and a cited overseas study. They do not guarantee persistence or outperformance in other markets, and the article notes that its conclusions are not investment advice.

Key ideas

Traditional factor alpha can remain related to the size, value, and momentum characteristics of a fund’s holdings.
Double adjustment first estimates factor-model alpha and then removes a cross-sectional holdings-characteristic component.
The article describes regression-based and matched-portfolio methods for computing the second adjustment.
In the summarized historical evidence, double-adjusted performance rankings better predict later factor alpha than standard rankings.
The findings rely on historical U.S. mutual fund data and may not generalize to other markets or periods.

Double-Adjusting Mutual Fund Alpha for Factors and Holdings

Summary

Key ideas

Tags