Guide · Updated 2026-05-08

Sequence of returns risk: why average return is a lie

By Yi Liu · Updated May 8, 2026 · 11 min read

Sequence of returns risk is the simple, brutal fact that two portfolios earning the same long-run average return can produce wildly different outcomes for a retiree — purely because of whenthe bad years hit. A crash in year 1 of retirement is mathematically devastating. The same crash in year 20 is barely a blip. Below: the math, the famous 1966 vs 1982 cohort example, the "fragile decade," and the six mitigations that actually work.

Key takeaways

Same average return + same withdrawal rate can produce wildly different outcomes — when bad years hit is what matters.
The 1966 retiree (stagflation early) ran out at year 30; the 1982 retiree finished with $4M+. Same 4% rule.
~80% of variation in retirement outcomes is driven by the first decade's returns (Kitces & Pfau).
Mitigations that work: bond tent, cash buffer, dynamic withdrawal, part-time work, delaying SS, mortgage-free entry.

Answer box

Take two retirees. Both start with $1,000,000, withdraw 4% inflation-adjusted, and earn the same 30-year average return. Retiree A hits a crash in years 1–3. Retiree B hits the same crash in years 25–27. Retiree A goes broke around age 80. Retiree B dies with $3M+ on the books. Same average return. Same withdrawal rate. Wildly different outcomes. That gap is sequence of returns risk — and it is the single largest threat to a long-horizon retirement, especially FIRE.

The two-retirees example: 1966 vs 1982

This is the most cited example in retirement research, originally from William Bengen's 1994 work. Alice retires January 1, 1966 with $1,000,000 in a 60/40 portfolio, withdrawing 4% ($40,000) adjusted for inflation each year. Bob retires January 1, 1982 with the same $1,000,000, same 4% rule. The 30-year geometric average return for both periods is roughly comparable (~9% nominal stocks). Here is what actually happened to their portfolios:

Year of retirement	Alice (1966 cohort)	Bob (1982 cohort)
Year 0 (start)	$1,000,000	$1,000,000
Year 5	$760,000	$1,520,000
Year 10	$520,000	$2,180,000
Year 16 (Alice 1982)	$200,000	$3,400,000
Year 25	~$0 (depleted)	$3,900,000
Year 30	$0	$4,000,000+

Figures are illustrative based on Bengen (1994) and Trinity Study updates using S&P 500 + intermediate-term Treasuries with CPI-adjusted withdrawals.

Same starting balance. Same 4% rule. Same long-run averages. Alice hit the 1966–1982 stagflation grind right out of the gate (negative real returns for ~16 years). Bob walked into one of the greatest bull markets in history. The difference between them is only the sequence.

Why sequence dominates: the math

The intuition is simple once you see it. In accumulation, when you have no withdrawals, only the geometric average matters — the order of returns does not change your final balance one cent. But the moment you start withdrawing, every dollar you pull during a bear market is sold at a depressed price, permanently reducing the share count compounding for you in the recovery.

Year-1 crash: Withdraw $40K from a portfolio that just dropped 30% to $700K. You sold at a 30% discount. Even if the market doubles back, you permanently lost the future compound on those shares.
Year-25 crash: Same 30% drop, but the portfolio has compounded to $2.5M first. The $40K withdrawal is now 1.6% of the base, not 5.7%. The drawdown happens on a far larger base, with far less of it being sold off.
The asymmetry: Bear markets cluster early ⇒ catastrophic. Bear markets cluster late ⇒ irrelevant. Same average, opposite outcomes.

Wade Pfau and Michael Kitces have shown that ~80% of variation in retirement outcomescomes from the first decade's returns, even though that decade is only one-third of a typical retirement.

The "fragile decade"

Kitces coined the phrase fragile decadefor the 5 years before and 5 years after retirement — the 10-year window where sequence risk is maximally destructive. For a traditional retiree that's roughly ages 60 to 70. A bear market in your 40s barely matters (you're still a net saver, buying cheap). A bear market in your 80s barely matters (your base is already huge and your horizon is short). But a bear market right at the handoff from saver to spender can reshape the next 30 years.

Why 5 years before?You can no longer meaningfully "save your way out" of a 30% drawdown. Your human capital runway is short.
Why 5 years after?You're now a forced seller. Every monthly withdrawal during a drawdown locks in losses.
Implication: Your risk profile should be lowestat the start of retirement, not highest. Most savers do the opposite — they get more aggressive late because they "need the growth."

Six mitigations that actually work

There is no way to eliminate sequence risk — you cannot forecast which decade is a 1966 vs a 1982. But you can dampen it substantially. These six tools, in roughly decreasing order of evidence:

Bond tent / rising-equity glide path (Kitces & Pfau). Start retirement at 40–50% equities, then increase stock allocation over the first 10–15 years back to 70–80%. Counter-intuitive, but it directly addresses the fragile decade — lowest risk exposure happens when a crash would hurt most.
Cash buffer (2–3 years of expenses). Hold $100K–$200K in short-term Treasuries or a money market fund. During a drawdown, spend the buffer and let equities recover without being forced to sell at the bottom. Cheap peace of mind; modest opportunity cost.
Dynamic withdrawal (Guyton-Klinger guardrails). Instead of rigid 4% inflation-adjusted, cut withdrawals 10% when your withdrawal rate drifts 20% above target, raise them when it drifts 20% below. Adds 1–2% of effective safe-withdrawal headroom in historical Monte Carlo tests.
Part-time work in years 1–5.Even $15K/year of earned income during the fragile decade dramatically reduces withdrawal pressure. Research shows it effectively adds ~0.5% to sustainable withdrawal rates. Known in the community as "Barista FIRE."
Delay Social Security to 70. Each year of delay past Full Retirement Age adds 8% to the benefit for life — an inflation-indexed, government-backed annuity. Strongest hedge available against both longevity risk and sequence risk.
Enter retirement mortgage-free. Removes the single largest fixed expense, which in turn lets you cut discretionary spending (your lever during a crash) without impacting lifestyle. Also reduces the withdrawal-rate denominator you need to sustain.

Why FIRE folks face it harder

Sequence risk is the core reason the 4% rule was calibrated for 30 years — not 50. Early retirees face a strictly harder problem than traditional retirees:

35–50 year horizon vs 30. More years = more potential bad decades. Bengen-style historical success rates drop from 95%+ at 30 years to roughly 70–80% at 50 years using the same 4%.
No Social Security backstop. A traditional 65-year-old retiree has a partial inflation-indexed annuity waiting. A 45-year-old retiree is fully on their own for 20+ years before any SS kicks in — and those are the most sequence-sensitive years.
Re-entering the labor market is brutal.The plausible mitigation "just go back to work" gets weaker the longer you've been out. A 50-year-old returning after a 10-year gap in a tech field faces real wage decay.
Health insurance risk stacks on top.Pre-Medicare coverage adds a $1,500–$2,500/month fixed cost that can't be flexed during a drawdown — see our pre-Medicare gap calculator.

Watch out

The honest FIRE answer: plan for 3.25–3.5% withdrawal, not 4%, if your horizon is 40+ years. And build at least two of the six mitigations above into the plan before you pull the trigger.

FAQ

How do I know if I'm in the fragile decade?

If you're within 5 years of stopping work or within 5 years of having stopped, you're in it. For traditional retirees that's roughly ages 60–70. For early retirees the window slides earlier — a 45-year-old planning to retire at 50 is already in the fragile decade. The defining trait isn't age, it's the transition from net saver to net spender.

Does sequence risk affect 401(k) only, or also Roth IRAs and taxable accounts?

It affects every withdrawing portfolio identically — the math is account-agnostic. What changes between 401(k), Roth, and taxable is the tax cost of withdrawals, not the sequence dynamics. That said, having multiple account types lets you control which account you draw from in a down year (e.g., spend taxable basis during a crash, leave Roth growing tax-free), which is itself a mitigation.

What percentage of Trinity Study failures were sequence-driven?

Effectively all of them. Bengen's original 1994 paper and the Trinity follow-up identified the 1966 cohort, the 1929 cohort, and the 1937 cohort as the worst-case starts — and in every case the common factor was severe negative real returns in the first 10–15 years. No 30-year retiree starting in a year with positive first-decade real returns has ever failed at 4% in the historical record.

Can I just hold 100% bonds in years 1–5 to avoid sequence risk?

It dampens equity sequence risk but introduces inflation risk, which can be just as destructive over a 30+ year horizon. The 1966 cohort that motivated this whole concept was killed by inflation, not by stock crashes. Pfau's research shows a bond tent (40–50% equities rising back to 70–80%) outperforms a static all-bonds-early approach in nearly every historical scenario.

Does buying an annuity solve sequence risk?

A SPIA (single-premium immediate annuity) for part of your portfolio does effectively eliminate sequence risk on the annuitized portion — you've converted a stochastic withdrawal problem into a fixed cash flow. Trade-off: you give up upside, liquidity, and bequest. Wade Pfau's research suggests annuitizing 20–30% of the portfolio at retirement materially improves worst-case outcomes; annuitizing 100% is rarely optimal.

How is sequence risk different from market timing?

Market timing is an active bet on direction ("sell now, buy later"). Sequence risk management is structural — you adjust the withdrawal rule, the glide path, or the cash buffer in advance, regardless of forecast. You're not predicting that 2027 will crash; you're acknowledging that some 5-year window in your retirement will be ugly and pre-positioning for it.

Does sequence risk apply to passive index investing?

Yes — it's purely a function of withdrawals + return path, not strategy. A 100% passive S&P 500 index investor faces identical sequence risk to an active manager with the same returns. In fact, indexing makes sequence risk more visible, because you can't blame poor outcomes on stock-picking. The only risk-factors that matter are the withdrawal rate and what happens in the first decade.

What would a 1929 retiree have ended up with using the 4% rule?

They would have survived — barely. Bengen's analysis shows the 1929 retiree, despite enduring the Great Depression and a 25-year nominal stock recovery, finished a 30-year retirement with a small positive balance using 4% inflation-adjusted withdrawals on a 50/50 portfolio. The 1966 retiree was actually the worse case because of stagflation, ending year 30 close to zero. Both cohorts illustrate that 4% is a floor, not a ceiling.

Methodology & Sources

The 1966 vs 1982 cohort comparison uses annual S&P 500 total return data plus 5-year intermediate-term Treasury yields, with CPI-U used for inflation-adjusting the annual $40,000 withdrawal. Portfolio assumed 60/40 stock/bond split, rebalanced annually at year-end, with withdrawals taken at the start of each year (the standard Bengen convention).

Bengen, W. (1994)."Determining Withdrawal Rates Using Historical Data," Journal of Financial Planning. The original paper that identified the 1966 cohort as the worst-case historical start date for US retirees.
Cooley, Hubbard, Walz (1998) — the Trinity Study.Extended Bengen's analysis and confirmed that of all historical failure cases, the common thread was a poor sequence in the first decade, not a bad average return.
Kitces, M. & Pfau, W. (2014)."Reducing Retirement Risk with a Rising Equity Glide Path," Journal of Financial Planning. Source for the "fragile decade" framing and bond-tent strategy.
Guyton, J. & Klinger, W. (2006)."Decision Rules and Maximum Initial Withdrawal Rates." Source for dynamic-withdrawal guardrails.
Robert Shiller historical data.S&P 500 returns and CPI back to 1871, used to validate the 1966–1995 and 1982–2011 outcome ranges.