Out-of-Sample Testing
Validating on data the strategy has never seen — the only test that resembles the future.
What it means
Split history: tune on the in-sample portion, then run once on the untouched out-of-sample portion. In-sample results measure how well you fit the past; out-of-sample results estimate how the system handles data it could not have memorized — which is the only thing the future is.
The discipline is in the "once": re-tuning after peeking at out-of-sample results quietly converts it into in-sample data. Walk-forward testing repeats the split across time to squeeze more honest evaluations from the same history.
Why traders care
- The strongest cheap defense against overfitting.
- A big in-sample/out-of-sample gap is itself the diagnosis: you fit noise.
Related terms
Educational content, not investment advice. Engine details describe how TraderWe computes this value; other platforms may define it differently.