Will My Money Last? The Risk an Average Return Can't Show You.
A 30% market drop in your first year of retirement does far more damage than the identical drop in year fifteen. The average return doesn't change. The outcome does — completely.
- Portfolio survival depends on more than withdrawal rate, average return, and lifespan — it depends critically on the order good and bad years arrive. This is sequence-of-returns risk, and it's largest in the first several years of retirement.
- Two portfolios can post the identical average annual return over 30 years and produce wildly different ending balances, purely because of which years the losses landed in.
- You can't know your own sequence in advance. The real defense is testing a plan against thousands of possible orderings — which is what the calculator below does with your own numbers.
Why "it averages 7% a year" doesn't answer the question
Most retirement planning starts with an average return — "the market's done about 7% a year historically, so I'll use that." It's a reasonable input for a rough projection. It is a genuinely bad input for the question "will my money last," because an average erases the one piece of information that determines whether a retirement plan actually survives: the order the returns show up in.
While you're working and contributing, sequence barely matters. A bad year early in your career just means you buy more shares at a lower price with the next paycheck, and decades of compounding smooth the timing out. The moment you start withdrawing, that symmetry breaks. Every dollar you pull out during a down year is a dollar that can never participate in the recovery, because it's gone. The portfolio that has to fund withdrawals during a crash is structurally worse off than one that hits the identical crash after years of growth have built a larger cushion.
The mechanism: withdrawing from a shrinking, not a growing, pool
Picture two retirees, each starting with the same balance, each withdrawing the same inflation-adjusted amount every year, each experiencing the exact same set of annual returns over a 25-year retirement — just in a different order.
- Retiree A gets the bad years first: two down markets in years one and two, then a long recovery. Their withdrawals in those first two years come out of a portfolio that's already shrinking from losses. By the time the recovery arrives, the balance is too depleted for the good years to fully rebuild it — they're compounding growth on a smaller number, permanently.
- Retiree B gets the identical two down years, but at the end of the same 25-year run instead of the start. By then, two decades of growth and compounding have built a much larger balance. The same-size loss, in dollar terms, is a much smaller percentage hit — and there are few or no future withdrawals still relying on a full recovery.
Same annual returns. Same average. Same withdrawal amount. Completely different outcome — one plan is meaningfully more depleted, or has run out entirely, while the other ends with a healthy balance. That gap is sequence-of-returns risk, and it's invisible to any plan built on a single average-return assumption.
A spreadsheet with "7% average return" in one cell cannot show you this. It literally cannot — averaging away the order is the whole problem.
Both lines use the exact same 20 annual returns — literally the same numbers, just reversed. Same $1,000,000 start, same $55,000/year withdrawal, same 6.05% average return. The only difference is which years the down markets land in — and it's the difference between a portfolio at $535,602 and one at $1,336,478.
View the underlying numbers as a table
| Year | Bad years first | Bad years last |
|---|---|---|
| 0 | $1,000,000 | $1,000,000 |
| 5 | $679,647 | $1,142,901 |
| 10 | $636,919 | $1,325,555 |
| 15 | $584,438 | $1,578,842 |
| 20 | $535,602 | $1,336,478 |
Why this can't be predicted, only tested
There's no way to know in advance whether your own retirement date happens to land you at the start of a bad decade or a good one — nobody in 2007 knew what 2008 held, and nobody retiring today knows what the next five years hold either. That uncertainty is exactly why "will my money last" doesn't have a single confident answer, only a probability.
What a real simulation does — the kind behind the calculator on this page — is run your specific plan through many hundreds or thousands of different possible return sequences, drawn from a realistic distribution of market behavior, and report the fraction that survive. Some of those simulated paths front-load the bad years. Some back-load them. Some never see a serious downturn at all. The success rate that comes out the other end already has sequence risk baked in, because it was tested against many orderings rather than one smoothed average.
A worked example
Consider two people, each 50 years old, each with $1,000,000 saved, each contributing $25,000 a year, each planning to retire at 65 and spend $5,000 a month with $2,500/month in Social Security starting at full retirement age. On paper their inputs are identical — same savings, same spending, same age. The only thing that differs is which 15 years, out of all the possible market sequences between now and their retirement date, they happen to live through.
| Scenario | What happens | Effect on the plan |
|---|---|---|
| Favorable sequence | Strong markets in the years right after retirement | Portfolio grows even while drawing from it — the plan gains a cushion it never needed to spend down |
| Adverse sequence | A significant downturn lands in the first few years after retirement | Withdrawals compound the loss — the same dollar amount taken out represents a larger share of a smaller portfolio, and there's less time left for a recovery to matter |
Neither person did anything differently. Neither made a mistake. The entire difference in outcome comes from a variable no one controls: which years the market happens to be down. This is precisely why a single success-rate percentage is more useful than a single average-return projection — the success rate already reflects that this uncertainty exists and reports how much margin the plan has against it, rather than pretending the timing doesn't matter.
What actually helps
You can't choose your sequence. But a plan can be built with more room to absorb a bad one:
- A cash or short-term bond buffer. Holding a year or two of spending outside the market means a down year doesn't force selling depressed assets to fund withdrawals — you spend from the buffer and let the portfolio recover before touching it again.
- Flexible spending. A plan that can trim discretionary spending after a bad year, instead of holding a fixed inflation-adjusted withdrawal no matter what, takes real pressure off the portfolio exactly when it needs it most.
- A bond tent. Temporarily raising fixed-income allocation in the years immediately around the retirement date — then gradually easing back toward stocks — reduces how much of the vulnerable early-withdrawal period is exposed to full market volatility.
- Delaying retirement or Social Security. Working a bit longer, or claiming benefits later, shortens the number of years the portfolio has to survive an early bad sequence unassisted — you can't pick a good decade, but you can shrink the window that a bad one has to hurt you in.
None of these eliminate the risk. They widen the margin. The honest way to know whether that margin is wide enough for a specific plan isn't intuition — it's running the plan against many possible sequences and seeing how it holds up, which is a different question than "what's the average return."
What this doesn't capture
A simulation tests against historically-plausible ranges of return sequences — it doesn't know about a specific future crash, a specific policy change, or a market regime that's never happened before. It's a projection, not a prediction: strong evidence about how a plan holds up against a wide range of plausible futures, not a guarantee about the one future that actually arrives. Treat a solid result as a real signal of margin, not as proof the risk is gone.
How this is calculated
The calculator above runs a 1,000-scenario Monte Carlo simulation, testing your plan against many different randomized orderings of market returns rather than one smoothed average, and reports the percentage that survive. It's the same engine behind the full app — free to run, nothing you enter leaves your browser. The complete math, including the specific assumptions and data sources, is documented in the methodology.
Common questions
This article runs one calculator on default numbers. The full app models your complete plan — taxes, Social Security optimization, and stress-testing against real market history.
Run your full plan free →