Monte Carlo simulation for prop firm sizing, explained

Here is the problem with a single equity curve. It shows exactly one sequence of wins and losses. But the sequence is partly luck. If three of your worst losing trades happened to land back-to-back early in a challenge, you would have blown the account — even though the same trades spread out would have passed comfortably. One curve cannot tell you how fragile you are to that reshuffling.

Monte Carlo fixes this by taking the real trade results and replaying them in thousands of different randomized sequences. Each replay is a plausible alternate history. Run 1,500 of them and you stop having a single answer and start having a distribution of answers.

Reading a distribution instead of a number

Once you have a distribution, you describe it by percentiles. The three that matter most:

• P10 — the bad tail. Ten percent of simulated years came in at or below this. This is your realistic worst case, the number you should actually plan around.
• P50 — the median. The middle outcome. Half of years did better, half did worse. This is the honest “expected” figure — not the best one.
• P90 — the good tail. The top decile. The number marketing departments love to quote in isolation. Useful only when shown next to the other two.

Look at the width of each range, not just the dots. The 50K Balanced portfolio spans roughly $5.8k to $12.4k with a median near $10k — a tight, predictable band. The Forex Champion portfolio sits much higher and wider. Width is uncertainty. A wider bar means the same strategy can deliver very different years depending on sequencing, and you should size for the left edge, not the middle.

Why this drives position sizing

The whole point of running the simulation is to set position size against the bad tail, not the median. If you size so that the median year looks great but the P10 year breaches your drawdown limit, you have built a strategy that works until the first unlucky sequence ends it. Sizing against P10 means the account survives the bad tail and the median simply becomes upside.

This is also why blow rate and Monte Carlo are the same conversation from two angles. Blow rate is just the fraction of those 1,500 simulated paths that touched the drawdown floor. You cannot honestly state one without running the other.