Skip to content
GetProfitable
Search
Dictionary

Minimum backtest length

The shortest history over which a claimed Sharpe ratio could be distinguished from the best of your search, roughly scaling with the number of trials.

The intuition: the more configurations you try, the longer a track record you need before the winner means anything. A rough published relation says the years required grow with the logarithm of the number of trials divided by the square of the Sharpe you claim.

Plugging in numbers: claiming a Sharpe of 1.0 after 50 trials needs on the order of 5 years of data; after 1,000 trials it needs closer to 8; after a million, more than 12. Intraday strategies escape some of this by having far more observations per year.

The practical reading is that a two-year backtest of a heavily searched daily strategy carries almost no evidence, regardless of how good the numbers are.

Related: deflated-sharpe-ratio, sample-size, multiple-testing, data-snooping

Educational only, not advice. Spotted an error? Post in Site Feedback.