The practice of testing a trained machine-learning model on data it has never seen to ensure it generalizes well. Techniques like train-test split, cross-validation or walk-forward testing assess performance on unseen samples. In forex robots, rigorous validation is essential: a model that only works on its training set will fail in live markets. For example, cross-validation helps detect overfitting by revealing if error rates jump on held-out data. Proper validation (often using out-of-sample price data) ensures the robot’s strategy is robust and not just memorized past patterns.

Back to Robots Glossary Index