The process of choosing the most relevant features for model training. This reduces model complexity and overfitting by removing irrelevant or redundant indicators. Feature selection methods (filter, wrapper, embedded) evaluate features by correlation or predictive power. In forex, feature selection might drop highly correlated technical indicators or noise from high-frequency data. For example, a filter method might compute correlation with returns and keep only top indicators. Effective feature selection leads to simpler, faster, and more generalizable trading models.

Back to Robots Glossary Index