TradingBarrioOpen platformTen Backtesting Mistakes Traders Should Avoid
Backtesting can create false confidence when the process quietly uses information or execution that would not have existed in real time. The most dangerous errors often make results look unusually smooth.
Hindsight and cherry-picking
Testing only memorable trends or skipping unclear setups selects the answer in advance. Use chronological data and include every occurrence that meets the written rule.
Overfitting parameters
Repeatedly changing indicators, thresholds, and targets to improve the same sample fits noise. Reserve unseen data for validation and prefer rules with a defensible market explanation.
Unrealistic execution
A candle touching a price does not guarantee a perfect fill, especially around news or gaps. Include spreads, commission, slippage assumptions, and order constraints where relevant.
Ignoring risk distribution
A positive average can hide rare large losses or an unacceptable drawdown. Examine losing streaks, adverse excursion, dependence on the best trades, and performance across regimes.
Preparation checklist
- Use chronological samples
- Freeze rules before testing
- Model realistic costs
- Keep unseen validation data
- Inspect tail losses
Frequently asked questions
What is overfitting?
It is tailoring a strategy so closely to historical noise that it may not generalize to new data.
Why can candle data exaggerate fills?
A bar shows a range but not always the exact intrabar sequence, liquidity, or execution available at each price.