A machine learning paradigm where an agent learns to make decisions by interacting with an environment to maximize cumulative reward. The agent takes actions (e.g. buy, sell, hold) and receives feedback (reward, such as profit/loss), adapting its policy over time. In forex trading, RL agents can theoretically learn optimal trading policies by simulating trades and learning from the outcomes. For example, an RL-based forex robot would try trades in a simulated market and gradually improve by reinforcing strategies that yielded higher rewards. Compared to supervised models, RL is unique in directly optimizing long-term returns through trial-and-error learning.

Back to Robots Glossary Index