Skip to main content
Documentationv2.4
Open dashboard →
DocsMonte Carlo

Monte Carlo

A thousand resampled futures, and why the median is not the number that matters.

UPDATED 6 DAYS AGO·5 MIN READ·03 SECTIONS
01

What gets resampled

The engine resamples your realised trade sequence one thousand times, shuffling order and, where appropriate, bootstrapping returns. Each resample produces an alternate equity path that is consistent with your strategy's actual trade distribution but a different sequence of luck.

The output is a cone: a median path with a P5 to P95 band widening over time, plus a distribution of terminal outcomes.

02

Reading the cone

The median path is the least useful line on the chart. What matters is the lower edge — the P95 drawdown — because that is the scenario that determines whether you are still trading the strategy in six months.

A narrow cone means results were driven by a consistent edge. A wide cone means sequence luck dominated, and your realised backtest was one draw among many very different possibilities.

!
SIZE FOR THE BAD PATH

If the P95 drawdown would force you to stop trading, the strategy is too large regardless of its median return.

03

Limitations

Monte Carlo assumes trades are broadly exchangeable. Strategies with strong serial dependence — pyramiding, martingale sizing, or regime-conditional entries — violate that assumption and their cones understate true tail risk.

The engine flags such strategies. Grid and martingale systems in particular routinely produce P95 drawdowns beyond seventy percent and fail this gate outright.

NOT A CRYSTAL BALL

Monte Carlo describes the distribution of your historical edge. It cannot anticipate a regime that never appeared in your data.

Was this page helpful?Report an issue →