Monte Carlo simulation
Monte Carlo simulation: see every way your strategy could have gone
Last updated: · Reviewed by the INDfolio AI research desk
In short
Monte Carlo simulation on INDfolio AI reshuffles and resamples a backtest’s trades across 1,000 alternate sequences to show the full distribution of equity curves the same edge could produce — maximum drawdown at 50th/95th/99th percentiles, longest losing streaks, probability of ruin thresholds and confidence bands on returns — instead of the single path history happened to deal. It runs in one click from any NSE backtest and is included from the Trader plan (₹999/month).
Your backtest shows one path — the exact sequence of trades history happened to deal. Monte Carlo simulation reshuffles and resamples those trades thousands of times to show the full range of paths the same edge could produce: the lucky versions, the unlucky versions, and how deep the drawdowns get in the bad ones.
It is the difference between "this strategy made 22% with a 12% drawdown" and "this strategy makes 14–28% in most runs, and one run in twenty sees a 19% drawdown — can my account and my nerves take that?"
Thousands of simulated equity curves
Trade sequences are resampled to generate the distribution of outcomes, not just the one history produced.
Drawdown at confidence levels
Know the drawdown you should plan for at 95% confidence — the number that should size your capital.
Streak analysis
See how long losing streaks plausibly run, so a normal rough patch does not get mistaken for a broken strategy.
One click from any backtest
Run Monte Carlo on any backtested strategy — no exports, no spreadsheets, no code.
Specification
Monte Carlo simulation on INDfolio AI — specification at a glance
| Dimension | Monte Carlo simulation on INDfolio AI |
|---|---|
| Method | Trade-sequence reshuffling and bootstrap resampling of the backtest’s trade list |
| Paths | 1,000 simulated equity curves per run |
| Outputs | Equity fan chart; max drawdown at 50th/75th/95th/99th percentiles; longest losing streak distribution; return confidence bands; probability of breaching a user-set loss threshold |
| Inputs | Any INDfolio AI backtest (equity, futures, options, indices); starting capital; optional position-sizing rule |
| Costs | Inherits the backtest’s Indian cost stack and slippage |
| Availability | Trader plan (₹999/month incl. GST) and above; one click from any backtest report |
| Coding required | None |
Why a single backtest overstates your confidence
Two strategies with identical trade lists can feel completely different depending on the order the trades arrive. History dealt one order; live trading will deal another. If your worst drawdown only looks survivable because the losses happened to be spread out, you are one unlucky sequence away from abandoning a profitable system at its low.
Monte Carlo simulation makes that risk visible before it costs money. By resampling your backtest’s trades thousands of times, it shows the drawdown distribution across alternative histories — and the right position size falls out of the pessimistic tail, not the average.
How to act on Monte Carlo results
Three practical readings. First, capital: size your account so the 95th-percentile drawdown is tolerable, not the backtest’s drawdown. Second, expectations: the median simulated return is a fairer forecast than the single backtest number. Third, kill criteria: if live losses ever exceed what the simulations called plausible, the market has probably changed and the strategy earns a review — that threshold is your objective off-switch, decided calmly in advance.
Methodology & assumptions
- Simulations reshuffle the realised trade sequence (and, optionally, bootstrap-resample with replacement); they do not generate new trades or alter the underlying edge.
- Results therefore describe the range of outcomes for the tested edge, not the probability that the edge persists — walk-forward analysis and forward testing address that.
- Percentile drawdowns are computed on the simulated equity curves at the chosen starting capital.