Why one backtest is one path
A single equity curve — one live account, one backtest run — is only one draw from a much wider range of outcomes your win rate and reward multiple can actually produce. Trade the exact same edge with the exact same fixed-fractional risk per trade, and the only thing that changes between runs is the order wins and losses arrive in — yet that order alone moves the final result and how deep the drawdowns get along the way. This tool plays the same rule out thousands of times (equity multiplies by 1 + risk% × reward R on a win, by 1 − risk% on a loss, exactly as it would in a real fixed-fractional account) with a seeded random number generator deciding each outcome, then reports the shape of every path it ran — not just the one you'd get if you happened to get lucky, or unlucky, in what order things landed.
What the percentile bands mean, honestly
Hold win rate at 50%, reward at 1.5R, and risk at 2% per trade over 100 trades (seed 7, 2,000 simulations, so this is exactly reproducible from the numbers alone): the 5th percentile final equity is 1.072×, the median is 1.596×, and the 95th percentile is 2.377× — meaning 1 in 20 simulated runs finished at 1.072× or worse, half finished below 1.596×, and 1 in 20 finished at 2.377× or better. The same runs show a median max drawdown of 13.6% and a 95th-percentile max drawdown of 23.9% — drawdown measured from each path's own running peak, not from the starting balance, since that's what actually triggers a firm's drawdown rule or a trader's own limit. 3.0% of runs ended below breakeven entirely. None of this assumes a fat-tailed or regime-shifting market: win rate and reward multiple are held fixed for the whole run, which real trading never does — so treat these bands as a lower bound on the surprise a live account can produce, never a forecast of what yours will do.
What to do with the numbers
Sweep risk per trade alone (same 50% win rate, 1.5R, 100 trades, seed 7) and the shape of the change is the real lesson: at 1% risk the band runs 1.043× to 1.555× with 12.7% worst-typical drawdown; at 2% it's 1.072× to 2.377× with 23.9% drawdown; at 3%, 1.086× to 3.574× with 34.3% drawdown; at 5%, 1.064× to 7.693× with 51.6% drawdown. The probability of ending below breakeven barely moves — 3.0% at 1%, 2%, and 3% risk, only ticking to 4.5% at 5% — while the drawdown percentiles more than triple and the band's own width explodes. That's the actual finding: a single "probability of loss" number stays deceptively calm long after the drawdown you'd have to sit through has become unbearable. Pick a risk per trade by looking at whether the 95th-percentile drawdown is something you could actually live with mid-run, not by looking at how attractive the median outcome looks — the median is who you might become, but the drawdown band is what you'd have to survive to get there.