As-traded results from the log, then what the same 876 entries would have produced under different exit rules. Every simulated number below comes from the recorded Max RR (max favourable excursion) — no new price data is assumed.
Cumulative net P&L (after brokerage) in trade sequence.
Same setup, three underlyings. BANKNIFTY is the only one that made money as traded.
The single largest effect in the dataset. Expiry-day behaviour dominates.
Net P&L per cell. Green = profitable, red = loss. Cell subtext is trade count / win rate. Use this to kill specific combinations, not whole days.
How far each trade went in your favour before it resolved. This is what makes exit optimisation possible without re-simulating price.
Two independent levers. BE@xR = move stop to breakeven once the trade reaches x R (turns a would-be loser into a scratch). T xR = take profit at x R. Best row highlighted.
A rule that only works on the sample it was fitted to is worthless. These are the tests that decide whether the optimisation is real.
Each row adds one rule on top of the previous. The ex-Friday column is the honesty check: if a rule only survives because of Friday, it is not a rule.
10,000 bootstrap resamples of the trade sequence on ₹2,00,000 starting capital. Your backtest is one ordering of these trades; this shows the other orderings you could just as easily have lived through.
Fixed-fractional risk per trade on the optimised exit rule. Net P&L scales linearly; drawdown scales linearly too — so this is purely a question of what drawdown you can sit through.