🤖 AI Summary
This paper addresses severe look-ahead bias and widespread overfitting in high-throughput asset pricing research. It proposes an unbiased screening framework based on empirical Bayes (EB) to systematically evaluate 136,000 long–short strategies spanning accounting ratios, historical returns, and textual code features. Methodologically, it is the first to apply EB to large-scale strategy mining, integrating cross-sectional t-statistic distribution analysis with rigorous out-of-sample testing to eliminate look-ahead bias and ensure reproducibility. The contributions are threefold: (1) It identifies statistically significant and robust excess return clustering in accounting-based strategies, small-cap stocks, and pre-2004 periods; (2) it produces high-quality pricing signals whose out-of-sample performance matches that of top-tier journal strategies; and (3) it provides novel evidence for the limited-attention hypothesis and establishes a rigorous, unbiased empirical benchmark for asset pricing research.
📝 Abstract
We use empirical Bayes (EB) to mine data on 140,000 long-short strategies constructed from accounting ratios, past returns, and ticker symbols. This"high-throughput asset pricing"produces out-of-sample performance comparable to strategies in top finance journals. But unlike the published strategies, the data-mined strategies are free of look-ahead bias. EB predicts that high returns are concentrated in accounting strategies, small stocks, and pre-2004 samples, consistent with limited attention theories. The intuition is seen in the cross-sectional distribution of t-stats, which is far from the null for equal-weighted accounting strategies. High-throughput methods provide a rigorous, unbiased method for documenting asset pricing facts.
