Reinforcement learning (RL)

Reinforcement learning – A paradigm where an “agent” learns to make decisions through trial-and-error interactions with the market (environment). The RL agent chooses actions (e.g. buy, sell, hold) in response to states (market features) to maximize cumulative reward. Unlike supervised learning, RL uses delayed feedback (profits/losses) rather than explicit labels. RL is used in some advanced trading bots that learn strategies by simulated trading. For example, a Q-learning agent might gradually learn which actions yield better returns over many simulated trades. A core focus in RL is balancing exploration (trying new actions) vs. exploitation (using known profitable actions).

⬅ Robots Glossary