How to backtest a trading strategy step by step

1. Convert the idea into fixed rules

Name the instrument, timeframe, trading session, indicators, data lookback, entry and exit conditions, stop and target behavior, position size, pyramiding policy, and the conditions that prevent a trade. Every rule should use information available at the simulated decision time. If two implementers could make different choices, the specification is incomplete.

2. Run and inspect a baseline

Implement the simplest complete version and save its code and settings. Run it on a defined period with commission and slippage assumptions. Inspect the trade list for impossible fills, duplicate orders, missing exits, and session-boundary behavior before reading summary metrics. Fix implementation errors before changing parameters.

• Save code version, symbol, timeframe, and dates.
• Record every cost and order setting.
• Review individual trades before optimizing.

3. Test robustness, then stop tuning

Change one assumption at a time and look for broad stability rather than one peak parameter set. After choosing the final rules, run an untouched period without more tuning. Paper trading or platform simulation can then reveal alert timing, latency, rejected orders, and live-data differences that a historical report cannot prove.