Top 10 Backtesting Mistakes & How to Avoid Them
A strategy that looks like a money-printing machine in simulations can quickly burn your capital in live markets if your backtest was flawed.
01. Overcoming Hindsight Bias
The absolute most common mistake is looking at a completed historical chart and deciding where you would have entered. Your eyes naturally gravitate to the bottoms and tops of moves. This introduces extreme bias, making the strategy appear far more successful than it is.
The Fix: Always use candle replay. Hide the future and step forward one candle at a time, forcing yourself to make entry and exit decisions without knowing what happens next.
02. The Danger of Curve-Fitting (Over-Optimization)
Curve-fitting occurs when you adjust your strategy rules and indicators to match a specific historical dataset perfectly. For example, finding that a 13.4 period EMA works perfectly on last year's EURUSD chart.
While this produces stellar historical returns, it is statistically useless. The market is dynamic, and a system built on hyper-specific past parameters will quickly fail in live conditions. Keep your rules simple, objective, and consistent across different years and markets.
03. Ignoring Transaction Costs and Spreads
A high-frequency scalping strategy might show 90% profitability in a frictionless simulation. But in reality, broker spreads, overnight swaps, and execution commissions will eat all your profits.
The Fix: Always configure realistic spreads and fixed commissions inside your backtesting software. If a strategy cannot survive typical market costs, it is not a viable business model.