The idea
Take the list of trades a strategy produced, each expressed in R. Draw trades from that list at random, with replacement, to build a new sequence of the same length, and compute its equity curve. Repeat this thousands of times. You now have thousands of possible futures that all rest on the same real results but arrive in different orders.
ATLAS draws ten thousand such futures by default in its future simulation, using the real results of the strategy you choose.
What it answers
The final result of a sequence depends on luck. The same trades in a different order can produce a deep drawdown or a smooth climb. By looking at all the futures together, you see the spread of outcomes: a typical result, a poor but plausible one and a favourable one.
It also gives you a more honest idea of the worst drawdown you should prepare for. The drawdown of the original backtest is only one value among many, and the simulated worst cases are often deeper.
How to read the percentiles
The median is the result in the middle: half of the futures end above it and half below. A low percentile, such as the fifth, is a poor outcome that happens in about one future out of twenty. A high percentile is a favourable one. A wide gap between the two says that the result depends heavily on the order of the trades.
The share of futures in which the account falls below a level you define is an estimate of the risk of ruin for that level, under the assumptions of the simulation.
What it cannot do
Monte Carlo simulation only reshuffles what already happened. It cannot create a market regime the strategy has never met, it assumes that future trades resemble past ones, and it usually treats trades as independent when real trades can cluster. If the backtest ignored costs, the simulation does too.
A simulation built on 20 trades is built on very little. The quality of the output cannot exceed the quality of the data that went in, which is why the sample size matters here as much as anywhere.
A tool for risk, not a prediction
Use the simulation to size positions and to decide whether you could live with the swings of a strategy. Do not use it as a forecast. A simulated future is not a probability of what will happen to your account.
Frequently asked questions
How many simulations are needed?
Thousands are enough for the percentiles to be stable. The limit is not the number of simulations but the number of real trades they are drawn from.
Is Monte Carlo the same as a backtest?
No. A backtest replays history once. A Monte Carlo simulation reuses the trades of that history in many different orders to show how much luck was involved.
See it in ATLAS
Future Simulation
Thousands of possible futures for a strategy, so you see the drawdowns you would have to live with.
Strategy Lab
Describe a strategy in plain words or paste a Pine Script; ATLAS turns it into rules and measures it.
Trading Journal
Record your trades, review them on the chart and get an AI diagnosis of your habits.
Terms used in this guide
Monte Carlo simulation, Drawdown, Risk of ruin, Sample size, R multiple.
More guides
This guide explains methods and arithmetic. It is not financial advice, and past results do not guarantee future ones. Read the risk disclosure.
