How long will my money last?

Why the order of returns matters

These three scenarios use the exact same average return and the same average volatility. The only difference is when the bad years hit. Watch how many years separate them:

"Bad years first" models retiring just before a sustained downturn (think a 2000 or 2008 retiree); "good years first" models retiring into a bull run. Same average return, same volatility — only the timing differs. This is a deterministic illustration of the mechanism, not a Monte Carlo probability, and the weak years are front-loaded but interleaved to stay realistic.

How the drawdown works

Each year the calculator applies that year's return to your balance, then subtracts your spending, which rises with inflation. Money "runs out" the first year the balance can't cover a full year of spending. The formula each year is simply:

The "steady average" mode applies your average return every year. The sequence modes keep the same average but reorder the good and bad years — front-loading losses ("bad years first") is the dangerous case for a retiree, because you're selling assets while they're down and the recovery happens on a smaller base.

Frequently asked questions

What number should go in the spending box?

Every answer this calculator gives is downstream of one guess: annual spending. Getting the portfolio right and the spending wrong produces a confident, useless result. These are official survey benchmarks you can start from — Bureau of Labor Statistics means for 2024, with the personal insurance and pensions line removed, because that category is mostly retirement contributions rather than spending.

These are population means, not targets, and more than half of consumer units spend less. Full breakdown by age and category.

How long does each portfolio size last?

Spending held at the $60,775 retiree benchmark, rising 2.5% a year with inflation, using the steady-average mode. The two return columns are deliberately far apart: the gap between them is the honest width of the uncertainty.

Nominal returns. With 2.5% inflation, a 4% nominal return is roughly 1.5% real and a 6% nominal return roughly 3.4% real, by the Fisher relation rather than simple subtraction.

The same portfolio at different spending levels

$1.5 million, 5% nominal return, 2.5% inflation. Spending is the variable with by far the most leverage — more than asset allocation, and more than fees.

Read down that last column. Cutting spending from $75,000 to $60,775 — about 19% — does not buy 19% more time. The relationship is sharply non-linear near the point where returns stop covering withdrawals, which is why small, permanent spending changes matter so much more than they feel like they should.

How long a fixed withdrawal rate lasts

A different lens on the same question. This grid strips out sequence risk and asks: if you withdraw a fixed percentage of your starting balance every year (raised with inflation) and earn a steady real return, how many years does the money last? "Never" means the return covers the withdrawal, so the balance survives indefinitely. This is the idealised backdrop the sequence-of-returns point above pushes against — real markets do not deliver a steady return, which is why the safe rate sits below the arithmetic breakeven.

The 4% row is the origin of the "4% rule": at a 2% real return it funds 35 years, and from about 4% real upward it never depletes. Note how fast the safe zone shrinks above 4% withdrawal — and remember these are steady-return figures, so a real portfolio should sit a notch more conservative than the grid suggests.

If you retire before 59½

This page models one pot of money. An early retiree does not have one pot — they have several, and the rules about which can be touched before 59½ decide the answer as much as the arithmetic does. Tax-deferred money drawn early without a documented exception carries a 10% additional tax on top of income tax, and health cover before Medicare at 65 is a separate line that can run five figures a year. The early retirement bridge calculator models those years account by account instead.