Out-of-sample
Pronunciation: out-of-SAM-pul
Data a strategy was NOT built or tuned on — used to check whether its edge is real.
Definition
Out-of-sample refers to market data that was not used to develop or optimise a strategy, as opposed to in-sample data (the history the rules were fitted to). Splitting history into in-sample and out-of-sample — or validating on later, unseen data — is the standard defence against overfitting: a strategy that performs on data it was never tuned to is far more likely to hold an edge live. It is a validation concept, not a guarantee.
In plain English — Out-of-sample data is the part of history you hold back when designing a strategy, so you can test it on data it has never seen. If the rules only look good on the data they were built on (in-sample), you may have just memorised the past. Out-of-sample testing is how you catch that.
Example
You design rules on 2018–2021 data (in-sample), then test them untouched on 2022–2023 (out-of-sample). A strategy that shines in-sample but collapses out-of-sample was probably curve-fitted.
Related terms
Where you see this in the app
Educational content only. Map.Trade does not provide financial advice or trading signals.