A supervised task that assigns inputs into discrete categories. In forex ML, classification is often used for signal generation – for example classifying market movements as “up” (buy) or “down” (sell). Common classifiers include decision trees, SVMs, k-nearest neighbors, and neural networks. For instance, a trading model might use technical indicators to classify the next bar as positive or negative return. Classification models output class probabilities or labels (e.g. using predict_proba() to get buy/sell probabilities). Precision, recall and confusion matrix (see below) are key to evaluating classifier accuracy on historic trades.

Back to Robots Glossary Index