Replicating Financial Anomalies in Portfolio Selection

Financial anomalies have been studied in the US for over 90 years. Recent evidence suggests that financial anomalies have diminished in the US and possibly in non-US portfolios. Have the anomalies changed and are they persistent? Have historical and earnings forecasting data been a consistent, and highly statistically significant, source of excess returns? The authors test many financial anomalies of the 1980�1990s and report that several models and strategies continue produce statistically significant excess returns. The authors test a large set in US and non-US markets over the past 30 years. The authors report that many of these fundamentals, earnings forecasts, revisions, and breadth and momentum strategies maintained their statistical significance during the 1995�2023 time period. Moreover, the earnings forecasting model and robust regression estimated composite model excess returns are greater in non-US and global markets than in the US markets. The authors further report that several weighting schemes for robust regression identify financial anomalies. Yes, Virginia, Martin, Guerard, and Xia (2024) report a more statistically motivated robust regression methodology, mOpt, that is highly statistically significant, but the authors report that alternative procedures offer effective models produce highly statistically significant stock selection models.