Monte Carlo Trading Simulator

How the simulation works

Each run builds one possible equity curve:

1. Start at your starting balance.
2. Draw a trade. With a pasted list, pick one of your trades at random, with replacement, so the same trade can come up more than once (a bootstrap). With win rate and averages, the trade wins the average win with a probability equal to the win rate and otherwise loses the average loss.
3. Add it to the balance, track the peak and the deepest fall from it, and repeat for the trades per run.
4. If the balance reaches the ruin threshold, the run stops there and counts as ruined.

After all runs, the tool sorts the results:

5th / 50th / 95th percentile = final balance that 5% / 50% / 95% of runs ended below
Max drawdown per run         = largest peak-to-trough fall (in money and % of the peak)
Probability of ruin          = runs that hit the ruin threshold / all runs

The chart draws 50 of the runs, the median path and the band between the 5th and 95th percentiles at each point. Results move a little each time because the draws are random; more simulations make them steadier.

What it can't tell you

• It assumes trades are independent. Resampling breaks up streaks. If your losses cluster, for example in one volatility regime, real drawdowns can be deeper than the simulation suggests.
• It assumes the future looks like the sample.A regime shift (a change in volatility, spreads or market structure) isn't in your trade list, so it isn't in the simulation either.
• Garbage in, garbage out. An overfit backtest gives an overconfident simulation. Check whether your backtest is overfit first.
• Small samples mislead. Resampling 30 trades can only recombine those 30 results. Our guide to backtest sample size covers how many you need.
• Fixed position size. P/Ls are added as they are, with no compounding. For risk as a percentage of the balance, use the risk of ruin calculator.

To test a specific change to the strategy, such as a different stop or a session filter, replay it on the real candles instead with Tessera's Simulations.