Anomalies Concentrate in a Small Set of Stocks?

Do a relatively few stocks drive the alphas of many anomalies? In the May 2025 revision of his paper entitled “The Intersection of Expected Returns”, Austin Sobotka explores stock overlap among the portfolios of 164 cross-sectional asset pricing anomalies. Specifically, he each month:

• Ranks stocks into tenths (deciles) by each anomaly characteristic, lagged by one month.
• Computes the number of times each stock falls into extreme deciles for each anomaly.
• Identifies stocks that appear in many extreme deciles (overlap) across anomalies.
• Forms for each anomaly three extreme-decile, long-short portfolios: (1) the conventional anomaly portfolio; (2) the anomaly excluding overlap stocks per some threshold (filtered); and, (3) the anomaly with only overlap stocks per some threshold (overlap). For example, 90th percentile filtered (overlap) portfolios exclude (include only) the 10% of stocks with the greatest long side overlap and the 10% of stocks with the greatest short side overlap.
• Holds these portfolios for one month.

He considers both equal-weighted and value-weighted versions of all portfolios. For robustness, he repeats the analysis ranking stocks into fifths (quintiles), applying various liquidity screens, rebalancing annually rather than monthly and using different sample periods. Using cleaned, winsorized monthly firm-level data for publicly traded non-financial stocks with non-zero market equity and priced over $1 as available during 1926 through 2023, he finds that: (more…)