How to backtest a trading strategy properly: 7 checks before risking money

A backtest tells you how your rules would have behaved in the past. Done well, it is the best filter for bad ideas. Done badly, it makes any idea look brilliant. These seven checks separate the two.

1. Include fees and slippage. Short-term strategies live or die on costs. Use your exchange's real fees.
2. Avoid look-ahead bias. Your rules may only use data available at that moment. Libraries like backtesting.py help.
3. Use clean, long data. Several years and different market regimes. For crypto, Binance public data is free.
4. Keep some data unseen. Optimise on one period, test on another you never touched.
5. Walk-forward. Repeat that train/test cycle over time. PyBroker does it out of the box.
6. Distrust the best parameter. If only one exact setting works, it is luck. Look for stable regions (vectorbt sweeps thousands fast).
7. Paper trade before real money. Then start small with hard risk limits.

Want your strategy tested for you?

Our Strategy Lab backtests your idea with fees, walk-forward and a full statistics report, from €290. Your rules stay yours.