Backtesting guide
Deflated Sharpe Ratio
The Deflated Sharpe Ratio is a probability-based adjustment for an observed Sharpe ratio after strategy selection. It raises the benchmark to reflect multiple trials and also accounts for sample length, skewness, and kurtosis in the selected return series.
Read the practical guideKey Takeaways
- DSR evaluates a selected Sharpe ratio against a selection-adjusted benchmark.
- It needs the chosen return series and information about the broader trial set.
- The effective number and dependence of trials affect the adjustment.
- A high DSR is conditional on the recorded trials and statistical assumptions.
Understand what DSR corrects
The largest Sharpe ratio from a research search is biased upward because it was selected from alternatives. DSR combines the Probabilistic Sharpe Ratio framework with a benchmark informed by the number and dispersion of strategy trials. It also incorporates the selected series length, skewness, and kurtosis rather than assuming a simple normal return process.
- Selection bias grows when more variants are tested and only the winner is reported.
- Non-normal returns change the uncertainty around an estimated Sharpe ratio.
- A longer series can reduce uncertainty, but dependence can limit its information content.
Assemble the required inputs
Keep the selected strategy returns, observation count, estimated Sharpe ratio, skewness, and kurtosis. Preserve the Sharpe results from all related trials so their number and variance can be estimated. The trial family should include discarded parameters, markets, indicators, and sample choices that participated in the same selection decision.
- Use one return frequency and Sharpe convention across all candidates.
- Record failed and abandoned trials as part of the search.
- Document how correlated variants were treated when estimating independent trials.
Interpret the probability with its limits
DSR is commonly interpreted as evidence that the selected Sharpe exceeds a benchmark adjusted for the research search. It is not a forecast of live profitability. Incomplete trial records, dependent variants, changing return definitions, short samples, and unstable moments can make the reported probability too confident.
- Report the assumed trial count and benchmark with the result.
- Run the analysis before viewing final confirmation data.
- Pair DSR with cost, stability, leakage, and out-of-sample checks.
Related Pineify workflow
Prepare the backtest and trial record for a DSR calculation
Use Backtest Deep Report to inspect the selected return distribution and Strategy Optimizer to retain results across parameter trials. Export those records for a separate DSR calculation.
Inspect the selected strategy return and trade distribution.
Review the selected backtestBoundary: Pineify currently reports standard Sharpe and optimization results. It does not calculate Deflated Sharpe Ratio or estimate the effective number of independent trials.
Primary sources
- Bailey and Lopez de Prado, The Deflated Sharpe RatioVerified 2026-08-16
- Bailey and Lopez de Prado, The Sharpe Ratio Efficient FrontierVerified 2026-08-16
This page is educational and does not provide investment advice. Backtests are hypothetical, depend on their data and assumptions, and do not guarantee future results. Trading can result in substantial loss.