A supervised algorithm that finds an optimal boundary (hyperplane) separating classes. SVMs can be used for classification of market regimes or regression of prices. The algorithm projects data into high-dimensional space and selects a hyperplane maximizing the margin between classes. For instance, an SVM could classify candlestick patterns. In trading, SVMs are valued for robustness in high-dimensional feature spaces; they also handle regression (SVR) for price forecasting.

Back to Robots Glossary Index