Backtesting

Backtesting applies a trading or investment strategy's rules to historical market data to estimate how it would have performed had it been run in the past. It is the primary way quantitative trading and algorithmic trading strategies are evaluated before any real capital is committed.

A backtest simulates entries, exits, position sizing, and costs over the historical period, then reports performance statistics such as the sharpe ratio and maximum drawdown. This differs from paper trading or a live pilot, which test a strategy going forward on real-time data rather than on data the strategy's rules were, often, developed using — a distinction central to backtesting's main weakness.

Backtesting matters because it is far cheaper than live testing and lets a researcher screen many strategy variants quickly, but its results are only as good as the assumptions built into the simulation. The most common pitfalls are overfitting a strategy to quirks of the specific historical sample, look-ahead bias, where information not actually available at the time of a simulated trade leaks into the test, and underestimating real-world transaction costs and slippage, all of which tend to make backtested results look far better than live performance turns out to be.