Monte Carlo retirement simulator
A Monte Carlo simulation runs your plan against a thousand different sequences of returns rather than one smooth average, and reports the distribution of outcomes instead of a single date. It answers a better question than an average does: not when you will retire if everything goes to plan, but how much of your plan depends on it going to plan.
The calculator below runs the deterministic projection. Vault's own engine runs the same plan across a thousand simulated paths and draws the spread as a fan — the median, the good quartile, and the one where you are still working.
Your annual spending divided by the withdrawal rate — not a fixed multiple of 25.
A projection is arithmetic on assumptions you choose. It is not a forecast, and it is not advice.
Why an average return misleads
Two portfolios can average the same 7% and end in completely different places, because the order of the returns matters. A run of bad years early, while the pot is small and you are still contributing, is survivable. The same run just after you stop working, while you are withdrawing, is the scenario that ends plans. An average hides that distinction entirely; this is called sequence-of-returns risk and it is the main thing a projection should be honest about.
What the fan chart is telling you
Each simulated path is one plausible future. The band you see is where most of them landed. A narrow band means your date is mostly determined by your savings rate, which is under your control. A wide band means it is mostly determined by markets, which is not — and the correct response to that is usually a longer horizon or a lower withdrawal rate, not a better forecast.
The limits worth stating
A simulation samples from assumptions about volatility and return that somebody chose. It cannot know a regime it has never been shown, it does not model your job, your health or your spending changing, and a thousand paths drawn from wrong assumptions are a thousand wrong answers drawn confidently. It is a way to see the shape of the uncertainty, not to remove it.
Questions
How many simulations is enough?
For a personal projection, a thousand paths is plenty — the distribution stops moving meaningfully well before that. More paths make the picture smoother, not more accurate; the assumptions are the limiting factor, not the sample size.
What success rate should I aim for?
There is no correct answer, but people commonly aim for the plan surviving in 85–95% of paths. Chasing 100% usually means working several extra years to insure against a scenario you could also handle by spending slightly less in a bad decade.
Is this the same simulation Vault runs?
The page above runs the deterministic projection. The full Monte Carlo, the coast-FIRE calculation and the bad-decade stress scenarios run inside the product against your real holdings and your real contribution history.