Backtest the challenge, not just your strategy

Start with the rules of the specific evaluation you intend to take. That normally means the profit target, maximum daily loss and maximum overall loss, including whether the overall limit is static or trailing. Then check for minimum trading days and any consistency rule limiting how much of your total profit can come from a single day. These are not administrative details to look at after backtesting. They define whether a particular equity curve passes or fails. A strategy can finish comfortably profitable and still be unsuitable for an evaluation if its normal losing sequences repeatedly collide with those limits.

Your test therefore needs to follow the account day by day rather than jumping straight to the final P&L. Track where each loss occurs, how much drawdown has accumulated and whether that day's trading would have breached a rule. If the firm uses trailing drawdown, model that rather than quietly substituting a static limit because it is easier. The point is to test the challenge you will actually take. Otherwise you have tested a different set of rules and given yourself reassuring numbers with limited practical value.

Treat the backtest like the evaluation, minus the fee and the elevated heart rate. Define the rules before you begin, replay historical markets without looking ahead and record every trade rather than quietly deleting the ugly ones. Then test whether the resulting sequence survives the firm's limits. Finally, use Monte Carlo to challenge the convenient assumption that your historical wins and losses will arrive in roughly the same order next time.

A handful of trades tells you very little about whether a strategy can survive a prop firm evaluation. You need enough observations to encounter losing streaks, different volatility regimes and periods where the strategy simply behaves badly. A few hundred trades across different market conditions tells you considerably more than 20 unusually convenient ones, but there is no scientifically settled magic number that makes a strategy "proven". Sample size also depends on trading frequency and strategy type. The useful question is whether you have tested enough varied conditions for drawdown and loss sequencing to become visible rather than exceptional.