Monte Carlo

A Monte Carlo simulation replays a trading system many times over to reveal the range of paths it can take. You describe the system with three numbers · win rate, reward-to-risk ratio and risk per trade · and the tool draws every trade at random according to that win rate, across sequences of N trades repeated dozens or hundreds of times. Each sequence starts at 100% of capital and risk is applied to the current balance, so gains and losses compound.

The point is not to predict the future but to see the dispersion. Two sequences run with exactly the same settings can end far apart, simply because wins do not arrive at the same pace. The chart overlays the sample paths, and the results show the median, the best case, the worst case and the ruin rate · the share of sequences whose balance was wiped out entirely.

The formula

Each trade is drawn at random: it wins with a probability equal to the win rate. On a win, the balance grows by balance × risk % × ratio · on a loss, it drops by balance × risk %. This repeats through the whole sequence, then across every simulation. The final balances are then sorted: the median is the middle value, the worst and best cases are the two extremes, and the ruin rate is the share of sequences whose balance fell to zero.

One example, walked through trade by trade

Take the default settings: 50% win rate, ratio 2, 1% risk, starting at 100% of capital. First trade drawn as a win: the balance rises by 100 × 1% × 2 = 2 points, to 102%. Second trade drawn as a loss: it drops by 102 × 1% = 1.02 points, back to 100.98%. The draw repeats on every trade, and the risk always applies to the current balance · that is why gains and losses compound instead of merely adding up. Repeated over 100 trades and 200 sequences, this page's draw gives a median of 162.8%, a worst case of 113.8% and a best case of 262.5%. Nearly 150 points between two sequences that followed exactly the same rules: that part belongs to chance, and no setting removes it.

The limits

It assumes a constant win rate and ratio from the first trade to the last · no real system keeps that promise, markets change and so does your trading. It draws every trade independently of the previous one, while real results can cluster: same market context, same mistake repeated. It ignores fees, slippage and the concrete floors of an account. Its 200 sequences are only a sample: the worst case displayed is the worst of this particular draw, not the worst possible. Above all, the numbers you type in are themselves uncertain · a win rate measured over 30 trades can sit far from your true probability, and the simulation passes that error through without warning.

The classic mistake

The median is not your expected result, it is the middle of a range: half of the sequences finish below it. The symmetric mistake is just as common · judging a system on a single sequence, your own. After a bad month, concluding the system is broken while its path still sits inside the normal range; after a great start, believing it better than it is. Look at the illustration: endings far apart come out of identical settings. One single path, yours included, proves almost nothing about the system that produced it.

Frequently asked questions