monte carlo simulation meaning in trading

A Monte Carlo simulation in trading reshuffles the order of your own past trade results thousands of times and replays each ordering, so you end up with a distribution of final values and worst drawdowns instead of a single number. Your backtest showed one ordering, and that ordering was luck. The output answers a better question: across a thousand sequences of the same trades, what range of outcomes could this edge have produced?

the order in which a sequence of trade results arrives is called

Path dependency — the property that the sequence of outcomes changes the lived result even when the outcomes themselves are identical. Multiplying the same set of percentage gains and losses in any order gives the same arithmetic answer, but a real account has a drawdown rule, a margin limit and a human behind it, so a run of losses arriving first can end the account before any of the wins show up.

how many trades do I need before a monte carlo test on my strategy means anything

Thirty trade results is a bare minimum and a hundred is far better. Reshuffling twenty trades a thousand times still contains only twenty trades of information while presenting it as though it were a thousand, which is how small samples mislead confidently. Record the results as percentages rather than rupees, so position sizing does not contaminate the sample.

what does the 5th percentile of a monte carlo backtest tell you

It is the honest planning number: 95% of the reshuffled orderings finished better than that value, so it is the outcome to budget for rather than the average. Reading the mean of a Monte Carlo run just recreates the optimism of reading a single backtest. The other figure worth extracting is the share of runs that breached your own stop-trading level, which is your practical risk of ruin.

why is my backtest drawdown smaller than my monte carlo drawdown

Because the drawdown you actually experienced is one draw from a distribution, not the ceiling of it. A strategy whose realised history shows a 14% worst fall will commonly show falls half again as deep in the unluckiest few percent of reshuffles, simply because the losing trades did not happen to arrive back-to-back in your particular sample. The distribution, not the realised figure, is what a position size has to survive.

Monte Carlo testing a trading strategy

A backtest gives you one equity curve: the one produced by the order in which the trades actually happened. That order was luck. The same set of wins and losses, dealt in a different sequence, produces a completely different-looking chart — and sometimes a blown-up account.

Think of it like this

The same deck, shuffled again

A teen patti player has a good night and goes home convinced of something. The cards were the same fifty-two; the order was luck. Play the identical deck in a different order and the same player might have left after an hour, broke.

In the market

Your trade history is one shuffle. Monte Carlo deals the same cards again, thousands of times, and shows you the hands you might just as easily have been given.

Why the order matters so much

If you never risked a fixed rupee amount, the sequence changes the result. Losses early, while the account is small, cost proportionally more than the same losses later. And a long losing streak that arrives at the wrong moment can take an account below the level from which it can recover at all.

Worked example

Same trades, two orders

20 trades: 8 wins of +12%, 12 losses of −5%, risking the full account each time

Order A: wins firstThe eight wins compound on a growing base₹1,00,000 → ₹1,34,100

Order B: losses firstIdentical — multiplication commutes₹1,00,000 → ₹1,34,100

Now add a 25% drawdown ruleNever falls far enough to trigger itOrder A survives

Order BTwelve losses first breaches the limit before any win arrivesStopped out in trade 6

Final position AFull result realised₹1,34,100

Final position BSame edge, same trades, different life₹74,000, and out

The arithmetic result is order-independent. The lived result is not, because real accounts have limits — a drawdown rule, a margin call, a spouse, a nerve. Sequence is what decides whether you were still there for the good part.

Running one, without any software

The whole method

1
List every trade result as a percentage
Not rupees — percentages, so position size does not contaminate the sample. Thirty trades is a bare minimum and a hundred is far better.
2
Shuffle the list and replay it
Apply each result to the running balance in the new order. Record the final value and the worst drawdown reached along the way.
3
Do it a thousand times
A spreadsheet can do this. You now have a thousand final values and a thousand worst drawdowns rather than one of each.
4
Read the distribution, not the average
The 5th percentile is the honest planning number: 95% of orderings did better than that. And the proportion of runs that breached your stop-trading level is your real risk of ruin.

What it cannot tell you

It assumes trades are independent. They are not. Correlated positions lose together, and a market regime change makes several trades fail for one reason. Reshuffling treats each as a separate coin flip, which understates the true tail.
It cannot invent outcomes you never had. If your sample contains no 2008, no reshuffle will produce one. The distribution is bounded by what already happened to you.
It does not validate the edge. Reshuffling a losing strategy produces a thousand ways to lose. Monte Carlo measures variance around an edge; it cannot tell you whether the edge exists.
Small samples mislead confidently. Twenty trades reshuffled a thousand times still contains twenty trades of information, presented as though it were a thousand.

Check yourself

Your backtest shows a maximum drawdown of 14%. A Monte Carlo reshuffle of the same trades shows 8% of orderings exceeding 25%. What should you plan around?

Simple bhasha mein

Wahi patte, dobara phenta hua

Teen patti mein raat achhi gayi toh lagta hai hunar tha. Patte wahi 52 the — tarteeb kismat thi. Apne trades ko hazaar baar shuffle karke chalao: wahi jeet-haar, alag tarteeb, aur aksar aisa daur nikalta hai jahan aap beech mein hi chhod dete. Us daur ke liye size tay karo.

What to remember

A backtest gives one ordering of your trades; the ordering was luck.
Multiplication commutes, but drawdown limits and nerve do not — sequence decides survival.
Read the 5th percentile and the ruin frequency, not the average.
Realised drawdown is one draw from a distribution that reaches considerably further.
It measures variance around an edge; it cannot tell you whether the edge is real.

You reached the endMark it done and keep your streak going.

Up nextIchimoku: five lines that describe a whole trend Previous: When news arrives without a date

Finished this lesson?

Mark it done to track your progress through the curriculum.

Common questions

Short, direct answers to what people ask about this topic.

monte carlo simulation meaning in trading: A Monte Carlo simulation in trading reshuffles the order of your own past trade results thousands of times and replays each ordering, so you end up with a distribution of final values and worst drawdowns instead of a single number. Your backtest showed one ordering, and that ordering was luck. The output answers a better question: across a thousand sequences of the same trades, what range of outcomes could this edge have produced?
the order in which a sequence of trade results arrives is called: Path dependency — the property that the sequence of outcomes changes the lived result even when the outcomes themselves are identical. Multiplying the same set of percentage gains and losses in any order gives the same arithmetic answer, but a real account has a drawdown rule, a margin limit and a human behind it, so a run of losses arriving first can end the account before any of the wins show up.
how many trades do I need before a monte carlo test on my strategy means anything: Thirty trade results is a bare minimum and a hundred is far better. Reshuffling twenty trades a thousand times still contains only twenty trades of information while presenting it as though it were a thousand, which is how small samples mislead confidently. Record the results as percentages rather than rupees, so position sizing does not contaminate the sample.
what does the 5th percentile of a monte carlo backtest tell you: It is the honest planning number: 95% of the reshuffled orderings finished better than that value, so it is the outcome to budget for rather than the average. Reading the mean of a Monte Carlo run just recreates the optimism of reading a single backtest. The other figure worth extracting is the share of runs that breached your own stop-trading level, which is your practical risk of ruin.
why is my backtest drawdown smaller than my monte carlo drawdown: Because the drawdown you actually experienced is one draw from a distribution, not the ceiling of it. A strategy whose realised history shows a 14% worst fall will commonly show falls half again as deep in the unluckiest few percent of reshuffles, simply because the losing trades did not happen to arrive back-to-back in your particular sample. The distribution, not the realised figure, is what a position size has to survive.