◈ Labs
Backtest check
v1.0Is this backtest Sharpe real, or the best of many lucky tries?Review: researchSign in to run Labs models
The free Explore plan includes 50 runs a month.
Sign in →How it works and its limits
- Inputs
- Periodic strategy returns, periods per year, number of variants tried, optionally the variants themselves.
- Outputs
- Sharpe with a bootstrap interval, Probabilistic and Deflated Sharpe, minimum track record, drawdown at 95%, probability of backtest overfitting.
- Evidence
- In 1,000 null simulations the best of 50 variants passed plain PSR 92% of the time and DSR 0%; a real edge among 20 variants was found 37% of the time.
- Limits
- Conservative by design: a pass is strong evidence, a fail is not proof of no edge. Costs must already be in the returns.
- Minimum data
- 30 returns; 64 rows per variant for overfitting.
- API
- POST /labs/backtest/check {returns, periods_per_year, trials, variants}