Monte Carlo Trading Simulation: Stress-Testing Your Plan Before the Market Does in 2026
A Monte Carlo trading simulation takes the trades you already have and asks a question your backtest cannot: what if they had arrived in a different order? Same edge, same win rate, same average win and loss. Different sequence.
The answer is usually uncomfortable. The equity curve you looked at was one path out of an enormous number of equally valid ones, and some of those other paths blow through your drawdown limit before the edge has a chance to show up. Your backtest did not tell you that, because a backtest only shows the single ordering that happened.
In this guide we will cover what a Monte Carlo trading simulation actually does, the three numbers worth pulling out of one, how to run it against a funded account’s published rules, and the specific ways this technique quietly lies to people who trust it too much.
Key takeaways
- Test the sequence, not the edge. A Monte Carlo trading simulation reshuffles your own trades to reveal how much of your equity curve was ordering luck.
- Size for the fifth percentile, not the average. The plan has to survive the bad ordering, because you only get one run at it.
- Watch maximum drawdown above all. It is the output that moves most between a backtest and a resampled distribution, and it is the number that breaches accounts.
- Feed it real trades or expect nonsense. Thirty cherry-picked trades produce a confident-looking distribution that means nothing.
- Label the output as hypothetical. Simulated results carry known limitations, and regulators require that to be stated plainly when performance is presented.
What this guide covers
What a Monte Carlo trading simulation actually does
A Monte Carlo trading simulation repeatedly resamples your historical trade results in random order and rebuilds the equity curve each time, producing hundreds or thousands of alternative histories from the same underlying edge. You then read the distribution of outcomes instead of a single line.
The name comes from physics, where the technique is used to model systems too messy to solve directly. In trading the messy part is sequence risk: the fact that the same set of wins and losses can produce a smooth climb or a career-ending hole depending purely on the order in which they show up.
Resampling your own trades
The most useful version for a day trader is simple. Take your list of closed trades, expressed as percentage or R-multiple returns. Draw from that list at random, with replacement, until you have a sequence the same length as your sample. Rebuild the equity curve. Record the ending value and the maximum drawdown. Repeat a few thousand times.
You now have a distribution rather than a number. Instead of "my strategy returned 34 percent with a 9 percent drawdown", you get "across a thousand orderings of the same trades, the median drawdown was 12 percent and one run in twenty saw 21 percent or worse".
The second statement is far more useful, because the number you have to survive is the bad one, not the average one.
What it is not
A Monte Carlo trading simulation does not predict the market. It does not tell you whether your edge is real. It does not generate new information. It is a way of extracting the full range of outcomes that were always implied by data you already had, and nothing more.
It also does not fix a bad sample. If your trade history came from six weeks of one market regime, resampling it produces a beautifully detailed picture of six weeks of one market regime. Dressing thin data in statistics makes it look more credible, not more true, and that is one of the more dangerous things this tool does.
The three numbers worth extracting
Three outputs justify the effort: the maximum drawdown distribution, the risk of hitting a hard rule limit, and the spread between the fifth and ninety-fifth percentile outcomes. Everything else is decoration.
The drawdown distribution
Maximum drawdown is the output that moves most between a backtest and a resampled distribution, and it is the one that determines whether an account survives. In almost every case the worst drawdown across a thousand orderings is substantially deeper than the worst drawdown you actually observed, because you observed one ordering and the simulation searched many.
Traders often react to this by assuming something has gone wrong with the calculation. Nothing has. Your historical worst losing streak was not a limit. It was a sample.
Risk of ruin, stated as a rule breach
Academic risk of ruin asks how often an account goes to zero. That framing is useless to a funded trader, because you never get to zero. You get stopped at a published drawdown limit long before that.
The useful version replaces "ruin" with "breach". Set the threshold at your account’s actual maximum drawdown in dollars, run the simulation, and count the percentage of paths that touch it. That percentage is the honest probability that your current plan, at your current size, ends the account before it ends the month.
The percentile spread
The distance between your fifth and ninety-fifth percentile outcomes tells you how much of your result is skill expressing itself and how much is variance. A wide spread is not automatically bad, but it does mean that a single month of results tells you almost nothing about your edge, which is worth knowing before you draw conclusions from a good week.
One backtest, a thousand possible histories
A Monte Carlo run shuffles the order of your own trades and replays them many times. The edge never changes. The path does, and the path is what breaks accounts.
The outcome you should be sized for. If this path breaches your account rules, the plan is too big regardless of the average.
A more honest expectation than the backtest, because it is the center of many orderings rather than the one you happened to observe.
Useful only as a reminder that a great run is also luck. Never plan around it, and never quote it as a result.
The single number that changes most between a backtest and a Monte Carlo run is maximum drawdown. The observed worst losing streak is almost never the worst one available from the same trades.
tradefundrr.comIllustrative example only. Simulated trading environment. Curves and percentiles are drawn to demonstrate a method and are not results, projections or a record of any account.
Running one against a funded account’s rules
The version of this exercise that matters for a funded trader is not "will I make money" but "does my plan survive the rules I agreed to". That turns an abstract statistical exercise into a concrete sizing decision, and it is where the technique earns its keep.
Translating percentiles into a daily loss limit
Funded accounts publish two hard numbers: a daily loss limit and a maximum drawdown. Both are stated in dollars. Your simulation should be too.
Convert your trade returns into dollars at your intended position size, run the simulation, and read off two things. First, what fraction of simulated days exceed the daily loss limit. Second, what fraction of simulated equity paths touch maximum drawdown. If either number is uncomfortable, the fix is almost always size rather than strategy.
| Question | What a backtest tells you | What a Monte Carlo run tells you |
|---|---|---|
| Total return | One observed figure | A distribution with percentiles |
| Maximum drawdown | The deepest hole that happened | The deepest hole available from the same trades |
| Losing streaks | The longest one observed | How often longer ones occur |
| Probability of a rule breach | Not answerable | A percentage of paths that touch the limit |
| Whether the edge is real | Not answerable | Not answerable |
| Sensitivity to position size | Requires a rerun | Scales directly, size is a multiplier |
Neither tool validates an edge. Monte Carlo answers a narrower and more practical question: given this edge, how rough can the ride be?
The trading-days problem
Most funded programs require a minimum number of trading days before a payout is available. That constraint interacts with your simulation in a way people miss. You are not trying to reach a profit target as fast as possible. You are trying to reach it without breaching, across a required number of sessions.
Run the simulation over the number of trading days your program actually requires, not over an arbitrary thousand trades. The answer changes, usually in the direction of "size down".
Simulated results carry required language for a reason
If you ever publish or share simulated performance, know that this is regulated territory. Under 17 CFR 4.41, hypothetical or simulated performance results presented to the public must carry a prescribed disclaimer noting that they do not represent actual trading, may have under or over compensated for factors such as lack of liquidity, and are designed with the benefit of hindsight. The NFA sets out a comparable standard in Interpretive Notice 9025.
That language is not bureaucratic noise. It is a precise summary of what is wrong with every simulation, including yours, and it is worth rereading before you get attached to a chart.
Where a Monte Carlo trading simulation misleads you
The technique has three well-known failure modes, and all three make results look better than reality rather than worse. That asymmetry is the reason to be careful.
Independence is an assumption, not a fact
Standard resampling assumes each trade is independent of the last. Real trading is not independent. Losses cluster because market regimes cluster, and because traders behave differently after a loss than after a win.
The practical effect is that a naive Monte Carlo run understates how bad your streaks can get. Block resampling, which draws consecutive runs of trades rather than single ones, partially corrects this and is worth the extra effort if your tool supports it.
Garbage in, confident garbage out
The simulation inherits every flaw in the input. Survivorship in your trade log, a strategy that was quietly adjusted mid-sample, a period with unusual volatility, commissions or slippage left out: all of it flows straight through and comes back wearing percentile labels.
Include costs. Include the trades you would rather forget. A hundred honest trades beat five hundred flattering ones.
The future is not a resample of the past
The deepest limitation is the simplest. Every path the simulation produces is built from returns that already happened. If the market changes character, none of those paths describe what comes next, and the tidy distribution gives you no warning at all.
It cannot see the risk that has not happened yet
Related to the point above, but worth separating: resampling can only redistribute the kinds of losses already present in your sample. A strategy that has never met a gap, a halt or a liquidity vacuum will produce a distribution with none of those events in it, and the distribution will look reassuringly tidy as a result.
Short-volatility and mean-reversion approaches are the classic case. They post smooth curves for long stretches, and a Monte Carlo run on that stretch reports low drawdown risk with high confidence. The confidence is real. The safety is not. The tool is describing the sample faithfully; the sample is simply missing the event that defines the strategy’s true risk.
There is no clean statistical fix for this. The practical answer is judgment: know what your strategy’s bad day looks like structurally, and check whether anything resembling it appears in your trade history. If it does not, treat every percentile in the output as optimistic.
- The sample has at least a hundred trades, ideally spanning different market conditions
- Commissions, fees and realistic slippage are included in every trade result
- Returns are expressed in dollars at your intended size, not in abstract units
- The breach threshold is your account’s real maximum drawdown figure
- The run length matches the trading days your program actually requires
- You looked at the fifth percentile first and the median second
- The output is labeled hypothetical wherever anyone else might see it
A practical workflow you can repeat monthly
The whole exercise takes about twenty minutes once set up, and it is worth repeating whenever your trade sample grows meaningfully or your size changes.
Export your closed trades with entry, exit, size and costs. Convert each to a dollar result at your intended size. Load them into a spreadsheet or a short script and resample with replacement, a thousand iterations, over the number of sessions your program requires. Record ending equity and maximum drawdown for each path.
Then read three lines. The fifth percentile ending equity. The median maximum drawdown. The percentage of paths that touched your account’s drawdown limit. Write those three numbers in your journal with the date and your position size next to them.
Over a few months that record becomes genuinely useful, because you can see whether the distribution is tightening as your execution improves or widening as you take on more risk. That is a far better feedback signal than a monthly profit and loss figure, which is dominated by luck over any short window.
Related reading: drawdown recovery math, expectancy explained and backtesting a futures day-trading strategy.
Frequently asked questions
What is a Monte Carlo trading simulation?
A Monte Carlo trading simulation randomly reorders your historical trade results many times and rebuilds the equity curve for each ordering. The output is a distribution of possible outcomes from the same edge, which shows how much of your observed result came from sequence luck.
How many trades do I need for a Monte Carlo simulation?
Around a hundred closed trades is a reasonable working minimum, and more is better. Below that, the sample is too thin for the distribution to mean much, and the tool will still produce confident-looking percentiles from data that cannot support them.
How many iterations should I run?
A thousand iterations is enough for stable percentiles on a normal trade sample, and ten thousand costs nothing extra on modern hardware. Beyond that the numbers stop moving, so additional runs add precision without adding insight.
Can Monte Carlo tell me if my strategy is profitable?
No. It only redistributes returns you already recorded, so a losing sample produces a distribution of losing paths and a profitable sample produces profitable ones. It answers how rough the ride can be, not whether the edge exists.
What percentile should I size my funded account for?
Most traders use the fifth percentile, meaning the plan should survive the worst one run in twenty without breaching the account. That is a judgment call rather than a rule, but sizing to the median guarantees roughly half of all orderings are worse than what you planned for.
Does a Monte Carlo run help with a funded account drawdown limit?
Yes, and this is its most practical use for a funded trader. Set the breach threshold to your program’s published maximum drawdown in dollars, then read the percentage of simulated paths that touch it. That percentage is a direct measure of whether your current size fits the rules.
Why is my simulated drawdown worse than my backtest drawdown?
Because your backtest showed one ordering of trades and the simulation searched thousands. The worst losing streak you happened to experience is a sample, not a ceiling, and a deeper streak is almost always available from the same set of results.
Do I need special software to run one?
No. A spreadsheet with a random sampling formula, or twenty lines of Python, is sufficient for trade-level resampling. Dedicated tools add convenience and block resampling, but they do not change the underlying method or the quality of your input data.
The point is the bad path, not the average one
Most traders run a Monte Carlo simulation, look at the median, feel reassured and close the window. That is exactly backwards. The median is the outcome you do not need to prepare for. The fifth percentile is the one that decides whether you are still trading in three months.
Used properly, this is a sizing tool wearing statistical clothing. It answers one question well: is my position size small enough that an unlucky ordering of trades I have already made would not end the account? If the answer is no, nothing else about the strategy matters yet.
Stress-test your plan against real published limits
TradeFundrr runs a structured, simulated environment with clear daily loss limits, maximum drawdown and position rules, so the thresholds you test against are the ones you actually trade under.
Explore the programs →