Why Backtests Use Historical Prices Alongside Simulated Prices

Summary

The document raises a core backtesting question: if a fitted price model represents the underlying process, why test a trading algorithm on historical prices rather than simulated paths? It suggests that simulation could expose a strategy to scenarios absent from the observed record. The discussion frames the tradeoff between testing on realized market history and exploring hypothetical outcomes generated by a model.

No answer, model, empirical comparison, or validation method is provided, so the idea remains an open question rather than a recommended approach. A key limitation is that simulations depend on the assumptions and fit of the chosen model; the document does not discuss how to test those assumptions or represent market features the model omits. It therefore offers a useful prompt about scenario coverage, but no evidence that simulated prices alone make a backtest more reliable.

Key ideas

• Historical prices are one realized path through market conditions.
• Simulated paths could represent scenarios that did not occur in the historical sample.
• A fitted model’s usefulness for backtesting depends on whether it captures relevant market behavior.
• The document poses the issue but gives no answer or validation approach.

Tags

# Backtesting trading algos using simulated price instead of historical prices


# Backtesting trading algos using simulated price instead of historical prices












I don't have any real experience in trading. I have a question in my mind for which I can't convince myself.

Why do we need real history price data to do back-testing?

In my mind, real data is only a realization of 'the' process. If there's a good fitting model, isn't it better to use it for back-tests? By this means, we will be able to see how it is going on in other scenarios because real data may not include those bad scenarios.

Thanks

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.