Does Academic Research Destroy Stock Return Predictability?
Read the paperopens doi.org in a new tab
What they found
The authors replicated 97 published stock return anomalies and measured how each performed in three windows: the original in-sample period, the gap between sample end and publication, and after publication. Returns fell by about 26% from the original sample to the pre-publication period (statistical bias, since researchers only publish what worked) and by about 58% after publication (investors learning about and trading away the anomaly). The decline was larger for anomalies that were cheap to arbitrage.
What you can use
- Expect any published strategy to deliver roughly half of its backtested return, at best, once you trade it.
- The decay is worse for liquid, easy-to-trade signals because they get arbitraged fastest.
- Some return usually remains, so publication does not fully kill an anomaly, but you should size for the post-publication number.
Caveats
Post-publication windows are short for recent anomalies. Decay may also reflect changing market conditions rather than arbitrage alone. Long-short portfolios, gross of costs.
Tags: anomalies, publication-effect, replication, alpha-decay
Summaries are our own reading of the paper, not the authors' words. Educational only, not advice. Discuss it in Book Club.