Monte Carlo Retirement Calculator

Why one smooth line lies to you

Every simple retirement calculator draws a single curve: your balance compounding at 7% a year, forever, like clockwork. The market has never once done that. It returns +26%, then −18%, then +11% — and when you're withdrawing, the order matters enormously. A crash in year two of retirement forces you to sell depressed assets to eat, and the portfolio may never recover, even if the long-run average ends up exactly 7%. The same crash in year twenty is a shrug. This is sequence-of-returns risk, and averaging hides it completely.

A Monte Carlo simulation confronts it directly: run the same plan through 1,000 different market histories — each drawing yearly returns at random around your average — and count the survivors. The answer isn't "you'll have $2.1M"; it's "your plan works in 87% of markets," which is a more honest shape for the truth.

How to read the cone

The chart shows the middle of the distribution: the median run (the line) and the 10th-to-90th percentile band (the cone). Half of outcomes land above the line, half below; one run in ten finishes above the cone and one in ten below. Two habits worth forming:

Plan around the median, not the average. A handful of lucky runs compound into fortunes and drag the average far above what typically happens. The median is the honest "expect roughly this."

Watch the bottom edge. The 10th percentile is your bad-luck rehearsal — not a catastrophe scenario, just a normal unlucky market. If that line hits zero while you're alive, the plan is thinner than the headline odds suggest.

What the success rate means — and what it doesn't

90%+ is generally sturdy. 75–90% is workable if you can flex — real retirees cut spending in bad years, which a fixed-spending simulation doesn't credit — so the figure is a floor for someone who can genuinely cut back. Below 75% leans on luck. And 100% usually means you're over-saving: the simulation's worst-case markets are rare, and insisting on surviving all of them means working years longer than the typical outcome required.

Random runs versus the real record

The random engine draws each year's return from a bell curve around your average. That is a fair stand-in for "markets are unpredictable," but it has no memory: it doesn't know that inflation runs in decade-long regimes, or that crashes tend to cluster, or that a 1966 retiree faced sixteen years of flat stocks and rising prices at the same time. The real market history mode replays your plan against the actual record instead — Robert Shiller's annual data from 1871 on, stocks and 10-year Treasuries blended at the share you choose — one run per possible start year. A 50-year plan gets about 105 of them, and the odds are simply the share of start years where the money lasted. This is the method behind the 4% rule, the Trinity study, and tools like FIRECalc and cFIREsim.

Each mode has a blind spot. History can't show anything worse than what has already happened, and its start years overlap so heavily that they are far from independent trials. Random runs can produce histories worse than any on record but treat every year as a fresh coin flip. Read them together: when both agree you have a sturdy plan; when they disagree, the list of failed start years in history mode tells you exactly which kind of market breaks yours — a crash at the start, or a slow inflationary grind — which is what you would actually prepare for. Narrow the start years (since 1950, say) to ask a sharper question: "would this plan have survived any retirement in living memory?"

What do you actually do with an 85% success rate?

First: don't chase 100. A plan that survives every simulated catastrophe usually means saving too much or spending too little — years of unnecessary frugality purchased as insurance against histories worse than almost anything markets have produced. Many planners consider 80–90% strong, and the reason is what the failures look like: they cluster in the worst simulated markets, and real people adjust — spend a little less in a bad year — rather than marching blindly off a cliff. The simulation can't model that mid-course correction. You can.

Second: treat the number as a comparison tool more than a verdict. "Is plan A sturdier than plan B?" is a question these odds answer well. Retire a year later, spend $5,000 less, save $200 more a month — run each and watch which one moves the number most. The direction and size of those moves is more trustworthy than any single reading, because the model's biases apply equally to both sides of the comparison.

Third: know what the odds can't see. Simulations draw from a statistical bell curve, and real markets are wilder than a bell curve — extreme years, crashes and booms alike, happen more often than the math expects. The output is also only as good as the spending and saving assumptions fed in; a precise success rate on top of a rough spending guess is precision in the wrong layer. The odds are a stress test, not a promise.

One mechanical note about this page: the simulation is seeded, so the same inputs always give the same answer. That's deliberate. Odds that wobble on every recalculation invite reading noise as signal — a plan didn't get sturdier because the dice came up differently.

The honest summary: a good success rate says the plan survives most plausible histories, and the gap to 100% is mostly histories you'd respond to anyway.

Common questions