Here are two retirees. Each starts with exactly $1,000,000. Each earns an average return of 7% a year over a 25-year retirement, withdrawing the same amount annually. By the math of averages, they should end up in roughly the same place.

They don't. One of them runs out of money in year 19. The other dies with more than they started with.

The difference isn't luck in the usual sense, and it isn't because one of them picked better investments. It's because of the order the same average return arrived in. This is called sequence-of-returns risk, and it's one of the most important things a retirement calculator can get wrong — quietly, while still reporting a reassuring-looking number.

Averages lie about the journey

Imagine a simplified market that returns +20% one year and -10% the next, alternating back and forth. Averaged out, that's a positive return most years. If you're saving — adding money every year, not withdrawing — the order those returns show up in barely matters. Your ending balance comes out about the same either way.

Retirement flips this. Once you start withdrawing money every year, the order suddenly matters enormously. If the bad years hit first, you're pulling money out of a portfolio that's already shrunk, permanently locking in a smaller base before it has a chance to recover. If the good years hit first, your portfolio grows a cushion before the bad years arrive, and that same market, in reverse order, barely dents you.

Same average return. Same withdrawals. Wildly different outcomes, purely based on order.

Where Monte Carlo comes in — and where it can mislead

Monte Carlo simulation is actually the right tool for this problem in principle — it runs thousands of different possible sequences of returns and reports how many of them left you with money at the end. That's a real improvement over a simple average-return projection, which ignores sequencing entirely.

The trouble is in how the result gets communicated. A tool that says "90% success rate" is quietly compressing thousands of very different stories — the fortunate sequences and the unfortunate ones — into a single number. It doesn't tell you:

  • How bad the unlucky 10% actually looks (do they run out of money in year 24, or year 12?)
  • Whether your specific plan is unusually exposed to a bad sequence in the first five years of retirement — historically the highest-stakes window, since that's when your portfolio is at its largest and a downturn does the most damage
  • What could be adjusted — a slightly lower withdrawal in the first few years, a different account draw-down order — to make the plan less fragile against exactly this risk

A single success percentage answers "did this work most of the time?" without answering the much more useful question: "if it doesn't work, why not, and what would have helped?"

The years that matter most

If there's one thing worth taking away from this, it's that the first five to ten years of retirement carry disproportionate weight. A market downturn during your saving years is often not just survivable but eventually helpful — you're buying more shares at lower prices. A market downturn during your first few retirement years, while you're also withdrawing money, is a different animal entirely: you're selling into a decline, which permanently reduces the base that has to recover afterward.

This is why a plan that only reports a single overall probability is missing the most actionable part of the picture. What actually helps is understanding whether your plan holds up specifically against an early downturn — and if it doesn't, whether a different withdrawal order, a cash buffer, or a temporary spending adjustment in the early years would meaningfully change the outcome.

What to ask instead of "what's my success rate"

Next time you run a retirement projection, it's worth pushing past the headline number:

  • Does this show me what happens specifically if the first few years are bad, not just the overall average across all scenarios?
  • Can I see the worst-case sequences, not just the percentage that succeeded?
  • Does the plan adjust for sequence risk at all, or does it assume the same withdrawal regardless of how the market has actually behaved so far?

A single percentage is a summary, not an explanation. The sequence your returns arrive in is often more consequential than the average return itself — and a retirement plan worth trusting should be able to show you that, not just report a score and move on.

A different way to frame the question

There's also a quieter alternative to running thousands of random futures and reporting how many of them worked: build the plan itself around the specific, known-risky periods — the early retirement years, the stretch before Medicare, the years right before required withdrawals kick in — and show, year by year, exactly what's driving the numbers in each one. That's a different kind of answer than a success percentage. It doesn't tell you the odds you got lucky. It tells you, in plain terms, what has to go right and what would happen if it doesn't — which is usually the more useful thing to know before you actually retire.