A single projection uses one smooth average return every year — useful, but no real market behaves like that. Monte Carlo runs your plan thousands of times, each with a different random sequence of yearly returns drawn from your expected return and its volatility. The share of runs where your money lasts to your planning age becomes the likelihood figure (e.g. '92% likely to last').
Why a probability, not a yes/no
Because markets are uncertain, the honest answer isn't 'yes' or 'no' — it's *how likely*. A plan at 95% is robust; one at 60% is a coin-flip you'd want to de-risk. The same machinery powers the safe-withdrawal-rate solver and the effect of turning on flexible spending, which typically lifts the likelihood by reducing draws in bad runs.
Monte Carlo draws returns randomly. For a complementary view that replays *actual* history in order — capturing crashes and recoveries as they really happened — see the sequence-risk stress test.