Join waitlist
The books... not free though
10 Strategier Som Slår BörsenSvenska10 Strategier Som Slår Börsen10 Strategies That Beat the MarketEnglish10 Strategies That Beat the MarketProRealTime Coding Made SimpleEnglishProRealTime Coding Made Simple
600+StatisticsMarket statistics and edges1,327StrategiesBacktested strategy library18Research papersThe study behind the algos301DocsProBuilder coding reference27Algo development programCourses and lessons24Everything here feeds the 24 algosSee the algos

Stats › Backtesting and evidence › Anomaly decay

Published anomalies lose 58% after publication, and data mined signals that were never published lose almost as much

In sample the original paper: 0.582%+0.582%In samplethe original paperOut of sample before publication: 0.432%+0.432%Out of samplebefore publicationPost publication: 0.264%+0.264%Post publication

Mean monthly long short returns for 97 cross-sectional predictors from 80 studies, by regime. The decline is minus 0.150 with a standard error of 0.077 out of sample and minus 0.337 with a standard error of 0.090 after publication. The middle bar is the in-sample mean plus the out-of-sample coefficient; the source prints the coefficients, not that level.

Returns to published predictors fall 26% out of sample and 58% after publication. But mechanically data mined signals, which nobody published and nobody traded, retain 51% against published predictors' 53%. Most of the decay is not about publication.

58%Fall in published predictor returns after publication
26%The same measured out of sample but before publication
About 53%Share of original returns retained by published predictors
About 51%The same for mechanically data mined signals

Members only

The rest of this page is for members

Below this point there are 5 sections and 5 named sources, roughly 800 words of it. Every figure carries the source it came from and the date the data is from.

You can keep browsing every statistic in the library for free. The intro and the summary are always open.