Overfitting

Overfitting is tuning a strategy so closely to past data that it captures noise rather than anything likely to happen again. An overfitted strategy looks excellent on the data it was built on and disappoints on anything else.

It happens when many settings are tried and the best is reported, or when rules are added one by one to avoid each bad patch in the past. With enough attempts, something will look good by chance.

The main defences are out-of-sample testing (judging a strategy on data not used to choose it), testing on other tickers and periods, keeping rules simple, and statistical corrections such as the deflated Sharpe ratio that allow for how many variations were tried.

A simple test: ask whether the result would survive a small change to each setting, and whether it appears on other tickers and in other years.

In one sentence

Tuning a strategy so closely to past data that it captures noise rather than a lasting pattern.

This definition is research and education, not financial advice.