What 60 trades at 80% win rate and 2% risk actually look like

The pitch most retail traders see for an "80% win rate" strategy is a single equity curve climbing smoothly from left to right, ending somewhere above the start by 20-40%. The pitch is misleading not because the math is wrong but because it shows you the median path and skips everything else. The real picture is a thousand different paths — some of which look like the marketing chart, some of which finish lower than they started, all of which represent the same strategy doing exactly what it's mathematically supposed to do.

This post runs the simulation honestly. Sixty trades at 80% wins, 1:0.4 reward-to-risk, 2% risked per trade, fees and slippage left out for clarity. One thousand independent paths. The interactive at the end lets you change every input and re-roll. The fixed picture below is what the default scenario produces.

The defaults and what they imply

Sixty trades is roughly four months of trading at a setup-frequency of about one signal every two days across a small basket of instruments. 80% wins is high but not unrealistic for mean-reversion strategies; the tradeoffs of that win rate are covered separately. 1:0.4 reward-to-risk means each winning trade returns 0.4× what each losing trade loses — small wins, full stops, the geometric shape that 80% wins forces. 2% risk per trade is the conventional retail floor that lets a moderate losing streak not threaten the account.

Per-trade expectancy in R-multiples: 0.8 × 0.4 − 0.2 × 1.0 = +0.12R. That's the average per-trade return in units of risk. Multiply by 60 trades and you get +7.2R as the simple-arithmetic projection. Compound at 2% per R: a deterministic projection ends near (1 + 0.12 × 0.02)^60 − 1 ≈ +14.6%. That's the number most strategy descriptions display.

The Monte-Carlo says something different. A thousand independent paths drawn from the same probability distribution produces a spread of outcomes around that median. The 5th percentile lands meaningfully below; the 95th well above. Some paths finish flat or negative even though the underlying expectancy is positive. None of those paths is the strategy "failing" — they're all the strategy working exactly as designed, drawn from the same distribution, just with different luck.

Why the spread matters more than the median

A trader looking at the deterministic +14.6% expects a year of running this setup to roughly double-and-a-bit that figure. A trader looking at the full distribution sees that, in any given 60-trade window, somewhere between 1-in-10 and 1-in-20 paths end below break-even. That's not a 1-in-100 catastrophe — it's a normal outcome that any honest 80%-win-rate strategy produces by construction when you don't get particularly lucky.

The implication: a single losing four-month run is not signal that the strategy stopped working. It's the bottom decile of an honest distribution. The signal that the strategy is failing only emerges over many more trades — long enough for the confidence interval on your win rate to narrow enough to discriminate between "true rate is still 80%, you got unlucky" and "true rate is now 70%, the edge is gone."

Most retail traders quit working strategies at exactly the wrong point: during the 5th-percentile path. The strategy was fine; the sample was too small to distinguish a normal cold streak from a broken edge. Reading the path-distribution chart instead of the deterministic line is the antidote.

The fixed-scenario distribution

The shape is the point. The gold band shows roughly where 90% of paths land at each trade number; the bold gold line is the median path. The narrow grey lines are individual sample paths. None of those is the path you'll actually walk through — yours will be a fresh draw from the same distribution. What's predictable is the shape of the band, not the position of any single line.

The interactive — change the inputs, re-roll

The simulator runs 1,000 paths server-side in the browser, sorts them per trade-step, and plots the median + the 5th-to-95th percentile band. The grey lines behind it are individual sample paths so you can see what one run looks like, not just the smoothed median.

Slide the win rate down and watch the band shift down. Slide the reward-to-risk up while keeping win rate fixed and watch the median climb plus the spread widen. Increase trades-per-run and watch the relative spread narrow — more samples cluster the per-path outcome closer to the true expectancy. Reduce trades-per-run to 20 and the spread overwhelms the median entirely; the strategy is the same, but the sample is too small for any single run to be informative about it.

Key takeaways from playing with the inputs

A few patterns emerge that aren't obvious from the deterministic projection:

The wider the band relative to the median, the less a single run tells you. At 60 trades and small per-trade edge, the 5th-to-95th band is wider than the median in most settings. This means a "typical" 4-month run can comfortably land at any point inside that band without any of those outcomes being signal that the strategy changed. Quitting after a bad run is statistical malpractice; over-confidence after a good run is the same mistake in the other direction.

Doubling trades doesn't double median return — it ~doubles it AND tightens the band. Compounding makes the median grow geometrically, but the band's width grows with √N (square root of trades). After 200 trades the band is roughly 1.8× wider in absolute terms but the median is roughly 4× higher, so the band-to-median ratio shrinks. This is why long-term trading judgements are more reliable than 60-trade ones.

Risk-per-trade matters more on the downside than the upside. Doubling risk-per-trade roughly doubles the median, but the 5th percentile worsens by a larger factor — compounding penalises drawdowns harder than equivalent gains help. The Monte-Carlo surfaces this asymmetry; the deterministic projection hides it.

80% wins is psychologically easy until it isn't. With 80% wins, a path with three losses in a row is unusual but happens. At 1:0.4 R:R, three consecutive losses cost 6% of the account at 2% risk; the strategy needs roughly 15 wins to recover. That's a month of trading for a typical retail setup. The strategy isn't broken — it's drawing the bottom of its honest distribution. Most retail traders quit during exactly this stretch.

For the deeper math of how these path stats relate to strategy evaluation generally, the Required R:R to break even and Annualized return projection tools cover the deterministic side, and the Win-rate confidence interval calculator shows when sample size is finally tight enough to discriminate between "you got unlucky" and "the edge degraded."

What this means for "annual return" claims

A strategy advertised as "+25% per year, 80% win rate, 2% risk" describes the median deterministic projection. The honest range in any given year is roughly +5% to +50% based on the same inputs, with a 5-10% probability of finishing the year flat or down. This is the strategy working as designed. None of those outcomes indicate a problem with the system.

When you're deciding whether to run a strategy, you should know not just the median but the distribution. A strategy with median +25% and 5th-percentile +5% is an entirely different product from a strategy with median +25% and 5th-percentile −15%. Both have the same expected value; the second is much harder to live with.

The Monte-Carlo above doesn't tell you which strategies are good — it tells you what range of outcomes you're actually signing up for given a strategy's stated stats. It's the missing context for almost every "the bot does X% per year" claim in retail trading.