A backtest shows one thing that actually happened: one specific sequence of trades, in one specific order, over one specific historical window. It is tempting to treat that single curve as a preview of the future. It is more accurate to treat it as one example of what was possible — and Monte Carlo simulation exists to explore the range around it.
One equity curve is only one path
The trades in a backtest occurred in a particular order because that’s the order the market happened to produce them in. If a handful of the largest winning trades had landed earlier or later in that sequence, or if a cluster of losing trades had occurred back-to-back rather than spread out, the same underlying strategy — with the exact same edge — could have produced a visibly different equity curve, a different maximum drawdown, and a different worst losing streak. The strategy’s edge didn’t change; the sequencing of outcomes did.
What Monte Carlo simulation actually does
Monte Carlo simulation takes a strategy’s or portfolio’s own historical trade results and reshuffles them — resampling the same set of outcomes into many alternative sequences — to build up a range of plausible equity paths, all consistent with the same underlying trade evidence but arranged differently in time. Running this many times produces a distribution: a spread of possible outcomes rather than a single number.
Crucially, this process uses the strategy’s own real historical trades as the raw material. It is not inventing new outcomes the strategy never produced — it is exploring how differently those same outcomes could plausibly have combined, had history unfolded them in a different order or a different specific sample.
What this reveals: a range, not a point estimate
Rather than a single maximum drawdown figure, Monte Carlo simulation can show a distribution of maximum drawdowns across many simulated paths — for instance, what a strategy’s drawdown looked like in the worst 5% of simulated sequences, versus a typical or median sequence. This percentile way of thinking — “in most simulated outcomes, drawdown stayed below roughly this level; in a small share of outcomes, it was considerably worse” — gives a more honest sense of uncertainty than a single historical number ever can.
The same idea applies to return: instead of one backtest return figure, a range of plausible returns across the simulated paths gives a sense of how much the single historical result might have varied under slightly different sequencing.
What Monte Carlo can tell you
- How sensitive a strategy’s headline drawdown and return figures are to the specific order its historical trades happened to occur in.
- A more realistic range of plausible drawdown outcomes than the single historical maximum alone.
- Whether a strategy’s backtest result looks unusually dependent on a small number of favourably-timed trades.
What Monte Carlo cannot tell you
It cannot tell you what will actually happen going forward — it works entirely from historical trade outcomes, and if market conditions genuinely change, the simulation’s range is still bounded by what the strategy has already demonstrated historically. It is not a forecast, and it does not turn an uncertain result into a certain one. Simulated results should never be read as guarantees, predictions, or a promise of future performance — they widen the picture of historical uncertainty, they don’t remove it.
Where this fits in the SIPS workflow
SIPS’s own Monte Carlo experience is currently a representative preview of this intended robustness-analysis workflow — the finished design can be reviewed today, using an illustrative dataset rather than a live simulation engine connected to your own strategies. If and when that changes, it will be described as a live, connected feature at that point, not before. Monte Carlo & Robustness covers exactly what’s currently shown and what isn’t yet connected.
The practical takeaway
A single backtest equity curve is real evidence, but it is only one path through a wider space of things that could plausibly have happened given the same underlying trade results. Monte Carlo simulation exists to make that wider space visible — as a range to understand, never as a forecast to rely on.