Hands-On Machine Learning for Algorithmic Trading

Timing of trades

The simulation could make unrealistic assumptions about the timing of the evaluation of the alpha factor signals and the resulting trades. For instance, signals may be evaluated at close prices when the next trade is only available at the often-quite-different open prices. As a consequence, the backtest will be significantly biased when the close price is used to evaluate trading performance.

The solution involves careful orchestration of the sequence of signal arrival, trade execution, and performance evaluation.