What Monte Carlo Simulation Shows About a Trading Track Record
A track record is one path through a set of trades, in the one order they happened to occur in. Monte Carlo simulation takes that same set of trades, shuffles the order thousands of times, and shows the whole range of equity curves — some smoother, some far rougher — that the identical set of wins and losses could just as easily have produced. It doesn't change the trades. It changes what you're allowed to conclude from the one path you actually got.
What does "running the Monte Carlo" on a track record actually mean?
It means taking a real, already-graded list of trade outcomes and asking a narrow question: if these exact outcomes had landed in a different order, what would the equity curve and the drawdown have looked like?
The simulation doesn't invent new trades or change the win rate. It takes the same multiset of results — the same wins, the same losses, the same sizes if sizing is included — and deals them out in a new random sequence, thousands of times over. Each reshuffled sequence produces its own equity curve: its own path, its own peak, its own worst drawdown along the way. Line all of those simulated curves up together and you get a range, not a single number — a picture of what was plausible, not just what happened to occur.
Why does the order of the exact same trades matter at all?
Because a drawdown is a property of the sequence, not of the total. The same ten trades summed up the same way can produce a mild dip or a brutal one, purely depending on which losses happen to cluster together.
Put three losing trades back to back near the start of a run and the equity curve dives early and recovers later — a rough-looking stretch that might have scared a live trader out before the recovery arrived. Spread those same three losses evenly across fifty trades instead and the curve barely wobbles. The final number is identical either way. What changed was only the order, and the order is exactly the one thing a backtest report's single reported path can't tell you anything about, because it only ever shows the one order that actually occurred.
| What stays fixed across every reshuffle | What changes with the order |
|---|---|
| Total number of wins and losses | The depth of the worst drawdown |
| The overall win rate | How early or late the worst stretch lands |
| The final cumulative result | How smooth or jagged the ride looks along the way |
What does a Monte Carlo analysis tell you that a bare track record doesn't?
It tells you how much of the comfortable equity curve you're looking at was luck of the draw, and what a materially worse — but equally real — version of the same trades could have looked like.
A track record with a shallow, orderly-looking drawdown can be reassuring for the wrong reason: it may just be the order that happened to occur, not evidence that a rough stretch is unlikely. Run the same trades through a few thousand reshuffles and a cluster of plausible worse outcomes usually appears — a deeper peak-to-trough decline than the one actually lived through, sitting well inside the range of what that same set of trades could produce. That range is the honest answer to "how bad could this realistically have gone," which a single historical path can't answer on its own no matter how good it looks.
What doesn't Monte Carlo simulation fix?
It doesn't check whether the trades themselves were real, honestly counted, or drawn from a fair sample — it only reorders what's already there.
A few things it leaves completely untouched:
- A cherry-picked or incomplete set of trades stays cherry-picked after reshuffling. Shuffling the order of a curated list of winners produces thousands of curves that are all still built from a curated list of winners — see how to spot a cherry-picked trading track record.
- A small sample stays a small sample. Reshuffling eight trades thousands of times gives you thousands of orderings of the same eight trades — it tells you about sequence risk, not about whether eight trades is enough to trust the rate itself; that's a separate question covered in how many calls before a track record means anything.
- It assumes the future looks statistically like the past it's reshuffling. The simulation only ever draws from the outcomes already on record. A strategy that was quietly tuned to one historical stretch carries that same risk into every reshuffled version — see what curve-fitting actually is.
Does a smoother simulated range mean a strategy is safe to trade?
No — it means the sequence risk inside that specific set of trades is better understood, not that the edge itself is real or will continue. Those are two different questions answered by two different checks.
Monte Carlo simulation is a check on the shape of the ride, not on the existence of the edge. A strategy can pass a clean-looking reshuffle test and still be built on an overfit rule, or on a sample too small to mean much, or on a cherry-picked set of results to begin with. Treat it as one honest input among several, not a stamp of approval on its own.
Questions traders ask
Does Monte Carlo simulation make a track record look worse?
It doesn't make anything look worse or better — it widens the picture. The actual reported result stays exactly what it was; the simulation adds the range of equally-real alternatives around it, which usually includes some rougher paths the single reported curve didn't happen to show.
Is this the same thing as backtesting on more historical data?
No. Pulling in more historical data tests whether the edge holds up outside the original sample — a different question, covered in what curve-fitting actually is and why a backtest rarely matches live results. Monte Carlo reshuffling never adds new data; it only reorders the data already on hand to show the range of paths that one set of outcomes could produce.
Does a bigger sample make the reshuffle range narrower?
Generally, yes — more trades in the set gives any one random reordering less room to cluster an unusually long losing stretch together, so the simulated range tends to tighten as the sample grows, the same underlying mechanism sample size describes for a measured rate's confidence interval.
Should every published track record include a Monte Carlo range?
It's a meaningfully more honest disclosure than a bare equity curve alone, because it shows a worse plausible path instead of only the one path that happened to occur. It's still not a substitute for a large, honestly-built, out-of-sample-tested sample — it's one more layer of honesty on top of that work, not a replacement for it.
For the two foundations this sits on top of, see how many calls before a track record means anything and what maximum drawdown actually measures. For the sampling problems reshuffling can't fix, see cherry-picked track records and survivorship bias. The method behind any real trades referenced in this lane is documented at the methodology hub.
Keep reading
Get them free