The Deflated Sharpe Ratio: Correcting for Trying Too Many Things
Key takeaways
- The more strategies you test, the more likely the best one is luck rather than skill.
- A raw Sharpe ratio ignores how many attempts produced it — and how skewed or short the record is.
- The Deflated Sharpe Ratio (Bailey–López de Prado) corrects for the number of trials, skew, and sample length.
- Applied to single-name patterns, it rejected six of eight names that had raw Sharpe ratios above 0.7.
- We do not ship the pattern that happened to win the lottery.
Run enough backtests and one will look brilliant by chance alone. This is the multiple-testing problem, and it is the engine behind most “discovered” trading strategies that evaporate in live trading. The Deflated Sharpe Ratio is the correction — and applying it honestly killed most of the single-name patterns that looked best.
What does the DSR actually do?
It adjusts a strategy’s Sharpe ratio for three things a raw Sharpe ignores: how many strategy variants were tried to find it, how skewed the returns are, and how short the track record is. A Sharpe of 0.8 discovered after testing a single idea is far more credible than the same 0.8 discovered after testing fifty. The DSR translates “impressive number” into “impressive given how hard we looked,” and reports the probability that the true Sharpe is actually above zero.
What did it reject?
We ran a battery of common technical patterns — RSI thresholds, Bollinger reversion, gap-fades and the like — across a set of single names, and found several with raw Sharpe ratios above 0.7. Superficially attractive. After deflating for the number of patterns tried per name, six of eight were rejected as statistically indistinguishable from luck. Only the survivors earned any further attention. The same multiple-testing logic explains the subset mirage elsewhere in our work: test enough variations, on enough universes, and something will always look like an edge.
| Single-name patterns | Result |
|---|---|
| Names with raw Sharpe > 0.7 | 8 |
| Rejected by the Deflated Sharpe Ratio | 6 of 8 |
| Survived as plausibly real | 2 of 8 |
The standard we hold
Before a result counts, it has to survive deflation for the search that found it. If a Sharpe only looks good because we tried many things, it is luck wearing a lab coat — and we do not deploy it.
About this series: every figure comes from a leak-free research harness on US equities — point-in-time index membership, fundamentals keyed to filing date, expanding-window walk-forward, and transaction costs charged. Statistics are gross and in-sample unless noted, and describe published anomalies, not a Yayati product. Standing caveats: roughly a third of true historical index members are unpriced by the naive data source (survivorship); a 2 bps cost assumption is optimistic; fundamentals are post-2009 XBRL.
Keep reading
This article is for educational and informational purposes only and is not investment, tax, or legal advice. It describes findings from an internal research program about publicly documented market anomalies and research methodology; it is not a description of any Yayati product or its results. Research statistics are gross, in-sample illustrations subject to survivorship, data-coverage, transaction-cost, and modeling limitations described in the text, and do not represent actual trading or any client account. Past performance and backtested results are not indicative of future results. Yayati Asset Management is a Registered Investment Adviser. © Yayati Asset Management. VOLT™ and PLASMA™ are trademarks of Yayati.
See VOLT™ on a real position.
The tax-smart option overlay behind this paper, for concentrated stock.