Monte Carlo

A Monte Carlo simulation is a way of testing "what could happen" by running a huge number of randomized versions of the same situation and looking at the spread of results. The name comes from the casino city, because the method relies on chance and repetition the way games of roulette do.

In trading, it's most often used to stress-test a strategy's track record. Instead of trusting a single backtest (one fixed sequence of trades in one fixed order), a Monte Carlo simulation takes that same set of trades and shuffles the order thousands of times, or randomly resamples wins and losses based on their historical frequency and size. Each shuffled sequence produces a slightly different equity curve, drawdown, and final result. Plotting all of them together shows a range of plausible outcomes rather than one lucky or unlucky path.

The nuance that trips people up is that Monte Carlo doesn't create new information about the future; it only reorganizes what already happened in the past (or samples from an assumed distribution). If the original data is too short, unrepresentative, or based on a curve-fitted strategy, the simulation will confidently produce a wide, official-looking range of outcomes built on a shaky foundation. It shows the range of results consistent with the input assumptions, not a guarantee of what will actually occur going forward.

People also confuse a single backtest's smooth-looking equity curve with the truth about a strategy's risk. Monte Carlo is a corrective: it reveals that the same win rate and average trade size could just as easily have produced a much rougher ride, including deeper drawdowns, if the trades had landed in a different order.