How we test strategies
A historical test is a controlled experiment, not a forecast. It is useful only when the inputs are known, the rules can be repeated, and costs and risk are represented honestly.
Set the conditions before the test
Record the exchange and market type, trading pairs, candle interval, date range, starting balance, position size, entry and exit rules, and risk limits. Change one material condition at a time when comparing variants.
Fees belong in the result
Trading fees and perpetual-futures funding can materially change the outcome, especially when turnover or holding time is high. Review the result after costs rather than relying on gross return.
Profit is not the only metric
Review maximum drawdown, losing periods, trade count, time in the market, and whether a few exceptional trades explain most of the outcome. Run the test again after material market or configuration changes.
History cannot reproduce the live market
Candles cannot fully capture order-book depth, latency, slippage, rejected orders, connection failures, or future funding payments. Validate execution in demo or an exchange test environment before trading with real funds.