Backtesting guide
Probabilistic Sharpe Ratio
The Probabilistic Sharpe Ratio estimates the probability that a strategy Sharpe ratio exceeds a chosen benchmark, given the observed estimate, sample length, skewness, and kurtosis. It expresses sampling uncertainty around Sharpe instead of treating the estimate as exact.
Read the practical guideKey Takeaways
- PSR answers a benchmark question, so the chosen Sharpe threshold must be stated.
- Sample length, skewness, and kurtosis change the uncertainty estimate.
- Return frequency and Sharpe convention must remain consistent.
- PSR does not correct a winner for an undisclosed search across many trials.
Choose the benchmark and return series
Define the minimum Sharpe ratio relevant to the decision. Zero may answer whether estimated risk-adjusted excess return is positive, while a higher benchmark may represent an alternative strategy or required hurdle. Use one return interval and one annualization convention for both the observed estimate and benchmark.
- State the benchmark before calculating the probability.
- Use returns aligned to the intended decision horizon.
- Document the risk-free or reference return used in the Sharpe estimate.
Calculate from the observed moments
PSR uses the estimated Sharpe ratio, the number of observations, skewness, and kurtosis to approximate the uncertainty around Sharpe. More observations generally narrow uncertainty, while asymmetric or heavy-tailed returns can change it. Serial dependence and overlapping observations require care because the raw row count may overstate independent evidence.
- Check missing values and return construction before estimating moments.
- Use a consistent definition of kurtosis in the calculation.
- Adjust the analysis when dependence reduces the effective sample size.
Read PSR as conditional evidence
A high probability supports the claim that the estimated Sharpe exceeds the declared benchmark under the model assumptions. It does not mean the strategy has the same probability of making money in live trading. If the strategy was chosen from many variants, DSR or another multiple-testing method is needed to address the selection step.
- Publish the benchmark and inputs beside the probability.
- Check cost realism and later-period performance separately.
- Use a selection adjustment when the reported strategy is a research winner.
Related Pineify workflow
Inspect the inputs before calculating PSR elsewhere
Pineify Backtest Deep Report provides a standard Sharpe ratio, distribution views, rolling results, and bootstrap scenarios from a TradingView trade export. These outputs help audit the return evidence used in a separate PSR calculation.
Review the selected backtest metrics and distribution before applying a PSR method.
Analyze the backtestBoundary: Backtest Deep Report does not calculate Probabilistic Sharpe Ratio or choose the benchmark and dependence assumptions for the user.
Primary sources
- Bailey and Lopez de Prado, The Sharpe Ratio Efficient FrontierVerified 2026-08-16
- SEC Investor.gov, Performance ClaimsVerified 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.