ML methods that infer patterns from unlabeled data. Unsupervised algorithms (e.g. clustering, principal component analysis) find structure in market data without explicit targets. In trading, unsupervised learning can detect market regimes or anomalies. For example, clustering can group similar days or price patterns together, helping a robot recognize “normal” vs “volatile” periods. Dimensionality reduction techniques (like PCA) fall under unsupervised learning and are used to compress feature sets. Overall, unsupervised learning helps extract insights (e.g. identify spurious outliers or underlying factors) from raw forex time series.

Back to Robots Glossary Index