← All 25 termsQuant methodology
Out-of-sample (OOS)
QuantData the model has never seen during training or parameter selection. The closest a backtest gets to a real test of generalisation.
Full definition
Out-of-sample (OOS) data is the held-out portion of the historical record that the model was neither trained on nor used to select hyperparameters against. A backtest's OOS performance is the most honest signal of how the strategy would have behaved live. On this site, OOS is gated by G11 to G15 of the G1 to G31 evaluation stack: every published project reports its OOS window start, end, and embargo separately from its in-sample window.
Related terms
Each link opens the long-form definition of a closely related term.
See this term in action.
The /methodology page maps every term to the gate or artifact that implements it on this site.