Monte Carlo and robustness: how to read the output

Monte Carlo analysis in trading is usually a family of simulations built from your own results. Two common versions shuffle the order of historical trades, and resample trades or returns with replacement to create many alternative histories. The output is a range of possible outcomes, not a new prediction.

What the distributions describe

If a strategy produced 400 trades, the sequence of wins and losses could have arrived in a different order. Shuffling shows how much the equity curve depends on that order. Typical outputs include the distribution of maximum drawdown, final equity, longest losing streak and time under water.

Read percentiles, not the best simulation. The median outcome is a reasonable central case. The 5th percentile drawdown is the kind of stress you should be prepared to survive. If a small change in assumptions moves that percentile dramatically, the strategy is fragile.

What Monte Carlo does not do

It cannot create an edge that is not present in the trades. It assumes the sampled distribution is a fair representation of the future, which is exactly the assumption you should question. It usually assumes independence between trades, while real markets cluster losses. It does not model a change of regime, a broker outage or a gap through the stop.

That is why Monte Carlo belongs next to other evidence, not instead of it.

Robustness checks worth doing

• Parameter neighbourhood: small changes in parameters should produce similar, not catastrophic, results.
• Cost sensitivity: raise spread and slippage; the idea should degrade gradually.
• Instrument and timeframe variation: the same logic on a related market is informative, though not mandatory.
• Trade count: treat results from fewer than a few dozen trades as anecdotal.
• Start date sensitivity: shift the beginning of the test period; a single lucky year should not carry everything.

A sensible reading routine

Look at the distribution of drawdown first, then at the spread of final outcomes, then at the streak lengths. Ask whether your risk limits would have survived the stressful simulations, not just the median one. Write the numbers down before you see a live account.

Keep expectations honest

A robustness study reduces the chance of being surprised, but it does not promise a return. It answers a narrower question: would I still be comfortable with this system if the uncomfortable simulations happened?

This article is educational and is not financial advice.