The application of ML to forecast future currency prices or returns. Regression models (linear, tree-based, neural nets) are trained on historical price/time-series data to predict the next price or movement. Accurate price prediction models are the core of many forex robots. For example, an LSTM neural network can be trained on sequences of past prices and technical indicators to forecast the next period’s price. Successful price prediction typically requires careful feature engineering (lags, indicators) and regularization to avoid overfitting.

Back to Robots Glossary Index