A lot of retirement planning, including a fair amount of FIRE math, quietly treats "average return" as the number that decides whether a portfolio lasts. The logic feels sound: if stocks return something like 7% a year on average over the long run, and a withdrawal rate is set below that, the portfolio should hold up. This is the assumption behind a lot of back-of-envelope FIRE spreadsheets — plug in an expected average return, run it forward, watch the balance grow.
It's also the assumption that sequence-of-returns risk breaks.
Here's the actual mechanism. During accumulation — the years you're still working and contributing — the order of returns barely matters, because you're adding money to the pot regardless of what the market is doing. A bad year early in your career, when your balance is small, works in your favor: you're buying shares at depressed prices with each paycheck, and when the market recovers, you own more of it. This is the whole logic behind dollar-cost averaging. Early losses, for an accumulator, are a discount.
Withdrawal changes the equation completely, because now money is leaving the portfolio instead of entering it. When a portfolio loses value and the retiree is also withdrawing a fixed dollar amount for living expenses, that withdrawal has to come from selling shares at depressed prices. Selling more shares to generate the same dollar amount permanently shrinks the number of shares left to participate in the eventual recovery. The portfolio doesn't just take a temporary hit; it permanently loses claim on a chunk of the eventual rebound. A retiree who experiences the identical average return but in reverse order — good years first, bad years later — never has to sell into the trough, and their ending balance can look entirely different, even though the arithmetic average of the returns is unchanged in both cases.
The clearest real-world illustration of this is the retiree cohort that began drawing down savings around 1966. This is the cohort that effectively defines the worst-case scenario underlying Bengen's original safe-withdrawal-rate research, which the syllabus piece "The 4% Rule Was Built for Thirty Years and One Country. You Might Need Fifty." describes and puts into contemporary context. The 1966 cohort walked into more than a decade of high inflation and stagnant-to-negative real stock returns, a stretch that didn't meaningfully turn around until the early 1980s. A retiree drawing a fixed, inflation-adjusted income through that stretch was selling shares into a market that wasn't just flat — it was losing real value year after year, while the cost of living kept rising underneath them.
By the time the 1982 recovery arrived, that portfolio had been drawn down for sixteen years through a market that kept losing ground in real terms, and what remained was a small fraction of the inflation-adjusted amount the retiree started with.
That's the gap between average return and actual outcome, stated in real terms instead of abstract ones. A portfolio with the same 30-year average return, but with the good years arranged up front instead of the bad ones, would have finished that period barely touched, still throwing off withdrawals from a base many times larger.
The genuine point of debate here isn't whether sequence risk is real — it is, and it's mechanical, not speculative. The debate is over how much it should be feared given that few real retirees actually behave like the fixed-withdrawal model assumes. The original safe-withdrawal-rate research holds spending constant in real terms regardless of market conditions, which is a useful worst-case stress test but not how most people actually live. Proponents of flexible, "guardrail" withdrawal strategies argue that a retiree who trims discretionary spending during a bad stretch — skipping the big trip, delaying a car purchase — recovers a meaningful amount of the safety a rigid model assumes away. Critics of that view counter that spending flexibility has limits, especially for early retirees who are years from Medicare eligibility: healthcare costs are a substantial and often non-discretionary line item, and current unsubsidized marketplace premiums have made that line item large enough that "just spend less" isn't always available the way it is for someone whose only discretionary spend was travel.
Standard mitigations work by addressing the same mechanism from different angles. A cash or short-term-bond buffer covering one to three years of expenses lets a retiree draw from that buffer instead of selling depressed equities during a downturn, buying time for a recovery before touching the portfolio itself — and with the 10-year Treasury yield at 4.57% (July 2026), that buffer isn't sitting entirely idle while it waits to be used. Guardrail withdrawal strategies build the spending flexibility in from the start rather than treating it as an emergency improvisation. And some early retirees keep a part-time income option in reserve specifically as a hedge against a bad sequence landing in year two or three, when the portfolio can least afford it.
The point is not that sequence risk is unsolvable. It's that a plan built entirely around an average return number has already skipped the variable most likely to determine whether that plan survives — and the fix isn't a better average, it's a plan that assumes the first decade could go badly and prepares for that specifically.