A common problem where a model learns the training data (including noise) too well and fails to generalize to new data. An overfitted trading model might show excellent results on historical backtests but perform poorly on live markets. This happens when the model is too complex or trained too long on limited data, causing it to memorize idiosyncrasies rather than underlying trends. Mitigation techniques include regularization, pruning, and validation. In forex AI, preventing overfitting is critical: models must capture robust patterns (like true price dynamics) and not transient artifacts, so that predictions remain valid in future market conditions.

Back to Robots Glossary Index