Sequence of returns risk: why the order of your gains and losses can matter more than their average
5 min read by Opthest
Two portfolios. Same 20-year average annual return. Same starting capital. Same withdrawals every year to live on. One runs out of money in year 14. The other still has more than half its capital left at year 20. The only difference between them is the order in which the good and bad years arrived. This isn’t a hypothetical — it’s the mechanism researchers call sequence-of-returns risk, and Morningstar’s 2026 State of Retirement Income report leans on it to explain why the “safe” withdrawal rate for new retirees this year sits at 3.9%, not the 4% rule of thumb everyone’s heard of.
If the average return is the same, how can the outcome be so different?
Because average return and withdrawal don’t commute — order matters the moment money leaves the portfolio. In a pure accumulation phase, with no withdrawals, the order of a −20% year and a +25% year is irrelevant: multiply them in either sequence and you land on the same number. Add a fixed euro withdrawal every year, though, and that symmetry breaks. A loss early on forces you to sell a bigger share of a smaller pot just to cover the same expenses — permanently shrinking the base that later gains compound on. A loss late, when the pot has already grown, eats into a cushion instead of the foundation. Same average, same losses, same gains — different arrival order, different fate.
Is this just a retirement problem?
Mostly, yes, and that’s exactly why it’s under-discussed — it only bites once you start pulling money out while the portfolio is still supposed to be working. Researcher Wade Pfau’s widely cited estimate is that the compounded return of just the first ten years of retirement explains roughly 77% of how the whole retirement plan ends up performing. Practitioners now talk about a “retirement risk zone” spanning the five years before and the five years after the withdrawal phase begins — get a bad sequence in that window and no amount of good years afterward fully undoes it, because the capital that would have compounded through them was already withdrawn or sold at a low.
Does this only apply to retirees drawing an income?
No — anywhere a portfolio has scheduled outflows during the horizon, not just at the end, the same mechanism applies: a house down payment funded partly from savings, a multi-year home-renovation drawdown, a business owner pulling a salary from an investment account, and yes, decumulation in retirement are all structurally the same problem. It doesn’t apply to a portfolio you’re purely accumulating — contributing to, never withdrawing from — because there the order of returns really doesn’t change the ending balance, only the smoothness of the ride.
So is a higher average return the fix?
No, and that’s the counterintuitive part — a higher average expected return doesn’t protect you from a bad sequence, and can even coexist with one. What actually helps is reducing exposure to being forced to sell low during the withdrawal window: a cash or short-duration bond reserve sized to a couple of years of expenses (so you’re not liquidating equities during a downturn), taking the withdrawal from whatever asset class is currently up rather than a fixed pro-rata slice, and — this is the part that connects back to methodology, not luck — actually looking at how a plan behaves across many possible return sequences instead of one assumed average. That’s precisely what a Monte Carlo simulation is built to do: it doesn’t just tell you the average outcome across thousands of paths, it shows you the ones where the bad years land early, so a “3.9%-looking” withdrawal plan can be checked against its worst plausible openings, not just its typical one.
How do you actually see this before it happens to you?
You don’t get a warning label on a bad sequence — by the time it’s visible in a single realized path, the withdrawals already happened at the wrong prices. The only way to see it in advance is to simulate many possible orderings of returns consistent with a portfolio’s expected return and volatility, and look specifically at the unlucky tail — the 10th or 25th percentile outcome, in years, not the median — the same discipline the math of recovery argues for on the loss side alone. A plan that only survives the median sequence isn’t a plan for the years that actually determine the outcome; it’s a plan for the years that don’t.
What this is NOT
This isn’t a suggestion to reduce your withdrawal rate, hold more cash, or change your allocation — those are decisions that depend on your own time horizon, other income sources, and risk tolerance, and nothing here is a personalized recommendation to do any of them. Morningstar’s 3.9% figure and Pfau’s 77% estimate are cited research findings, not universal rules that apply unmodified to any individual plan, and simulated outcomes describe a range of plausible futures, not a guarantee of what will actually happen to your portfolio.
Two plans can share the same average return and still land in opposite places — because a withdrawal doesn’t just take money out, it changes what the rest of the sequence can do. Plan for the order, not just the average.
Sources: Morningstar, “State of Retirement Income” 2026; CNBC, “Market volatility and retirement: Sequence of returns risk explained,” April 2026; Wade Pfau, retirement-research literature on the sequence-of-returns effect.