Ten Backtesting Pitfalls That Inflate Trading Strategy Results
Summary
The document describes ten ways a trading strategy’s backtest can misrepresent performance: look-ahead pricing, future data leakage, omitted fees and slippage, unsuitable index contracts, overfitting, limited market capacity, too few observations or signals, and narrow market coverage. It uses moving-average and Turtle-style strategy examples on Chinese futures to show how changing signal timing, costs, or contract construction can produce very different equity curves. The examples illustrate why trades must be simulated only at prices available after a signal becomes known, with realistic transaction costs and tradable instruments.
The article also warns that selected time periods, discarded losing days, or heavily tuned parameters can make historical results look unusually strong without improving future performance. It recommends considering liquidity constraints and testing across longer periods, more signals, and varied market conditions. The discussion is conceptual and relies on charts whose underlying data and methodology are not supplied; it does not quantify how common each pitfall is or provide a formal validation protocol.
Key ideas
- A backtest must execute trades only at prices available after the signal is observable.
- Fees and slippage can turn an apparently profitable strategy into a losing one.
- Continuous and index futures series can differ from prices and costs encountered in live contract rolls.
- Overfitting and selectively chosen periods can make historical performance look more reliable than it is.
- Strategy evaluation should account for liquidity, signal count, sample length, and varied market regimes.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.