#MARKETS & INVESTING
McLean and Pontiff replayed 97 signals said to predict stock returns. Once the study documenting them is published, their return falls by 58%.
Published on August 25, 2026
In the Journal of Finance (2016), R. David McLean and Jeffrey Pontiff rebuilt 97 variables that academic studies had shown to predict US stock returns. They compare three returns: the one measured in the original sample, the one obtained out of sample but before publication, and the one obtained after publication. The result: −26% out of sample, −58% after publication. The 32-point gap between the two only appears once the study is public; the authors attribute it to investors who read the study and arbitrage the signal away. Heiko Jacobs and Sebastian Müller extend the exercise to 241 anomalies across 39 stock markets, more than two million anomaly-country-months (Journal of Financial Economics, 2020): the United States is the only country with a reliable post-publication decline.
Does a stock market strategy that worked keep working once it is published?
In the United States it loses 58% of its return on average after publication (McLean-Pontiff, 2016). Across the 38 other markets studied by Jacobs and Müller, no reliable decline is measured.
In other words: publishing a recipe for outperformance is enough to cut its effect by more than half.
Four caveats, because they matter
- −58% is not −100%: the average return does not vanish, it is divided by a little more than two.
- The 26% out-of-sample decline is an upper bound on data mining: part of the original signal was statistical noise.
- McLean and Pontiff's sample is American. Jacobs and Müller find no reliable post-publication decline in the 38 other markets they study.
- The decline is sharper for anomalies with high initial returns, and returns stay higher where arbitrage is expensive: illiquid stocks, high idiosyncratic risk.
Why it still matters: a published backtest should be read at a discount. The past return was not only measured, it was made public, and therefore partly arbitraged away.
For you
- Can a market inefficiency survive its own publication?
- How many "proven" strategies have you read about without knowing they lose half their effect once known?
Source: R. David McLean and Jeffrey Pontiff, "Does Academic Research Destroy Stock Return Predictability?", The Journal of Finance, vol. 71 no. 1, 2016; Heiko Jacobs and Sebastian Müller, "Anomalies across the globe: Once public, no longer existent?", Journal of Financial Economics, vol. 135 no. 1, 2020.
View the source →