A backtest gives you exactly one equity curve — the one produced by the specific sequence in which your trades happened to occur. But that sequence is just one path out of an enormous number of possible orderings your same trades could have taken. Monte Carlo simulation answers the question a single backtest can't: if your winners and losers had landed in a different order, or a different sample from the same underlying edge, how much worse could things have looked?
It matters because sequence risk is real. A strategy with a genuinely positive edge can still produce a brutal losing streak early on, purely by chance, that wipes out an undercapitalized account before the edge ever gets a chance to play out. Monte Carlo simulation is how you find out whether your position sizing and account size can actually survive that kind of unlucky draw — before it happens with real money.
What Monte Carlo Simulation Actually Does
The method is simple to describe even though the underlying math traces back to nuclear physics research in the 1940s. You take your historical trade results — the actual list of wins and losses, in dollars or R-multiples — and you randomly reshuffle the order thousands of times, recalculating the resulting equity curve on each pass. Some approaches also resample with replacement, effectively drawing a fresh sequence of trades from the same statistical distribution rather than just reordering the exact same list.
- 1Take your closed trades (ideally 100+) as a list of P&L outcomes
- 2Randomly shuffle that list into a new order
- 3Recalculate the equity curve and max drawdown for that shuffled order
- 4Repeat 1,000–10,000 times, storing the result of each run
- 5Look at the distribution of outcomes across all runs, not just your one original backtest
The output isn't a single number — it's a distribution. Instead of "my max drawdown was 18%," Monte Carlo simulation tells you "in 5% of possible orderings of these same trades, max drawdown exceeded 34%." That's a materially different — and more honest — picture of your risk.
A Worked Example
Say your journal shows 200 closed trades with a 55% win rate and a 1.5 average R-multiple win against a 1R average loss — a real, positive edge. Your actual equity curve, in the order the trades happened, shows a 12% max drawdown. Run that same 200-trade list through 5,000 Monte Carlo shuffles and a different picture emerges:
| Percentile | Max Drawdown | What It Means |
|---|---|---|
| Best 5% of runs | 6% | Lucky ordering — losers were spread out |
| Median run | 15% | Typical outcome from this same edge |
| Worst 5% of runs | 29% | Same edge, unlucky clustering of losses |
| Worst 1% of runs | 38% | Tail-risk scenario worth planning for |
Your actual backtest, at 12%, sits near the lucky end of that distribution — not the median. A trader who sized their account around "my strategy draws down 12%, max" and didn't run the simulation would be badly under-prepared for the 29–38% drawdown that same edge is fully capable of producing on a different draw of the same trades.
Why One Backtest Isn't Enough
- A single equity curve confuses "what happened" with "what the edge is capable of producing" — they're not the same thing
- Early-sequence losing streaks are the most dangerous, since they hit an account before profits have built a cushion — Monte Carlo surfaces how bad that specific risk can get
- It separates strategy risk (does the edge exist) from sequence risk (did you get a favorable or unfavorable ordering of the same edge)
- It gives you a realistic worst-case drawdown to size your account and position risk against, instead of anchoring to whatever number your specific backtest happened to produce
Using It to Size Your Account and Risk Per Trade
The practical output most traders want from a Monte Carlo run is a defensible answer to "how much capital do I need, and how much can I risk per trade, to survive a bad draw of my own edge?" If the simulation shows a 5th-percentile max drawdown of 30%, sizing your account so that a 30% drawdown is merely uncomfortable — not account-ending or prop firm rule-breaking — is a very different decision than sizing around your single backtest's 12%.
This connects directly to risk of ruin: Monte Carlo simulation is effectively how risk-of-ruin figures get calculated in practice, since ruin probability depends on the full distribution of possible drawdown paths, not just the average one. Traders who've calculated position size off a single backtest's statistics are often sizing against a best-case sequence without realizing it.
TradeEdge logs every closed trade with full R-multiple and P&L detail, giving you the clean historical dataset a Monte Carlo simulation actually needs — no manually rebuilding your trade history from broker statements first.
What Monte Carlo Simulation Can't Tell You
It reshuffles the trades you already have — it can't invent a better or worse edge than the one your history reflects, and it assumes future trades will behave statistically like past ones, which breaks down if your strategy, market regime, or execution changes. It also needs a real sample to work with: reshuffling 20 trades thousands of times just produces 20-trade noise dressed up as a distribution, the same small-sample trap that undermines Sharpe ratio and other statistics calculated too early — see how many trades you actually need before trusting any of them.
Why This Needs a Journal, Not a Spreadsheet Rebuild
Running a proper Monte Carlo simulation means having a clean, complete list of every trade's outcome — something most traders don't have sitting ready, since it usually lives scattered across broker statements and partial notes. A trading journal that already logs every closed trade's P&L and R-multiple has the input data on hand at all times, which is what turns Monte Carlo simulation from an occasional research project into something you can rerun every time your trading edge picks up another 50 or 100 trades of history.