Monte Carlo Simulation in Trading: What It Is and When to Use It

A Monte Carlo simulation in trading takes a strategy's return characteristics and replays them thousands of times in random order to build a range of possible outcomes rather than a single result. Instead of one equity curve, you get a distribution: a best case, a worst case, and everything between. It answers the question a single backtest cannot, which is how much your result depended on the exact order that trades happened to occur.

That distinction matters, because a strategy can look great on one historical path and fall apart if the same trades arrive in a different sequence. Monte Carlo stress-tests that fragility. Here is how it works, what it is good for, and where it misleads people.

How a Monte Carlo simulation works

The mechanics are simpler than the name suggests. You start with a set of outcomes, usually the individual trade results or the daily returns from a backtest, then resample them many times to see how the account could have evolved along different paths.

1. Collect the inputs. Take the trade-by-trade results or period returns from a tested strategy: the wins, the losses, and how often each occurs.
2. Shuffle or resample. Draw those returns in a new random order, or sample them with replacement, to build one alternate path for the account.
3. Repeat thousands of times. Each run produces a different ending balance and a different worst drawdown along the way.
4. Read the distribution. Sort the results into percentiles. The median is a typical outcome; the 5th percentile shows a bad-but-plausible one.

What Monte Carlo tells you that a backtest does not

A standard backtest reports one path: the actual historical sequence. Monte Carlo reports the range around it. The two are complementary, and confusing them is a common mistake.

What is Monte Carlo used for in trading?

Its most useful jobs are about risk, not return. Traders use it to estimate a realistic worst-case drawdown before it happens, to size positions so a bad sequence does not blow up the account, and to judge how much confidence a backtest deserves. Investors use the same math for retirement and withdrawal planning, projecting whether a portfolio survives decades of random market paths. In every case the point is the tail, not the average: you are looking at the ugly 5% of outcomes to decide whether you could actually live through them.

Where Monte Carlo simulations mislead people

The math is only as honest as its inputs, and that is where most Monte Carlo work goes wrong. If you feed it the trade results from an overfit backtest, you get a beautifully detailed distribution of a fantasy. Reshuffling returns also assumes each trade is independent, which understates real markets where losses cluster in crashes and volatility arrives in bursts. And a simulation that resamples only calm-market history will never show you a 2008. Treat the output as a stress test on a rule you already trust, not as proof that a weak rule is safe.

Backtest first, then simulate

Monte Carlo sits downstream of a good backtest. You cannot simulate paths for a rule you have not first tested on real history, so the honest order is always: define the rule, test it on decades of actual data, then stress the survivor. That first step is what Agenttrading is built for. It is not a broker and executes nothing: you type a thesis in plain English, it restates the rule, backtests it on 20+ years of split- and dividend-adjusted daily data with a 0.1% cost per trade assumed by default, explains the risks in words, and stamps an honest verdict, HELD UP, MIXED, or UNDERPERFORMED. If you want to run the same idea across a group of holdings, you can bundle several tickers into a single weighted basket and test the mix as one. The mechanics of a real historical test live on backtesting software, testing a whole portfolio is covered on portfolio backtesting, and a related risk metric is explained in what is risk of ruin. Drop a rule into the trading strategy tester to get the historical result a simulation would start from.

Past performance does not guarantee future results. For educational and informational purposes only. Not financial advice. Consult a licensed advisor.

The honest bottom line

A Monte Carlo simulation turns one backtest into a range of outcomes by replaying returns in random order, which is the best cheap way to estimate a plausible worst-case drawdown and judge how fragile a result is. It does not tell you what really happened, it does not fix a bad strategy, and it is only as trustworthy as the tested rule you feed it. Run the historical backtest first, then use Monte Carlo to pressure-test what survived.