A supervised learning task where the model predicts continuous outputs (e.g. future price or return). Common regression algorithms (linear regression, random forests, neural networks) map input features to a numeric target. In forex, regression might predict the next hour’s exchange rate, or expected return. Regression models learn from historical labeled data (past prices) to minimize error on continuous outcomes. For example, a model could use previous price and volume indicators to forecast tomorrow’s EUR/USD rate. It’s fundamental to predictive trading models that aim for a numerical forecast rather than a discrete classification.

Back to Robots Glossary Index