Two retirees, same math, different fate
Run a thought experiment: two savers both retire with $1 million, both pull 4% ($40,000) a year adjusted for inflation, and both experience the exact same 30 years of stock returns — one in historical order, one reversed.
- Retiree A gets the bad years first: a 2008-sized 37% drop in year one.
- Retiree B gets that same drop in year thirty, after two decades of compounding.
Both see the identical average return. If you handed a spreadsheet the *mean* annual return, both plans would look identical. But their portfolios don't behave identically. In the ugliest historical sequences, Retiree A runs out of money with years of retirement still to fund. Retiree B ends year 30 with a balance several times their starting amount. Same returns, same rules, different order.
(This is a mirrored-sequence hypothetical, not a claim about actual 2008 retirees — the real 2008 cohort was rescued by the fastest recovery and longest bull market in modern history. That rescue is exactly the point: their plan needed one, and got lucky.)
That's sequence of returns risk. When your portfolio drops early in retirement and you're still pulling from it, you lock in losses. The dollars you withdraw don't get a chance to recover. And the portfolio that has to fund the *next* 25 years of retirement is compounding from a much smaller base.
Why an "average" return can lie
The 4% rule from William Bengen's original 1994 study is a historical *worst-case* — the highest constant real withdrawal rate that survived every rolling 30-year period in US market history, using his 50% stock / 50% intermediate-Treasury portfolio, including the retirees who retired into the 1929 crash, the 1966 stagflation, and the 1973–74 bear market. Those cohorts are why the rule isn't 5% or 6%.
But "the historical worst case *survived*" is not the same as "any market you retire into will be safe." Two things are true simultaneously:
- A 4% real withdrawal survived every rolling 30-year window in Bengen's balanced-portfolio backtest.
- The first decade of returns dictates whether you finish rich or broke.
Both facts follow from the same math. Losses at year 1 compound differently from losses at year 25. By year 25 your portfolio has (in most sequences) grown well past its starting balance — a 30% drop is uncomfortable but not fatal. At year 1 you're still at (or below) your starting balance, and every dollar you spend during that drop is a dollar that never gets to ride the recovery.
The mechanism, in one paragraph
You retire with $1M. Year 1 hands you −37%. You pull $40K. Your portfolio ends the year at approximately $590K. Year 2 the market rebounds +26%. You pull $41K (inflation-adjusted). Your portfolio ends at ≈$700K. On paper the market is back near where you started. Your portfolio is now 30% below where you started, and you're still pulling $40K/yr in real terms. Year 3, 4, 5 — even if the market runs at 8% real — your portfolio is compounding from ~$700K, not $1M. The math never quite catches up.
Now reverse the sequence: same 30 years of returns, worst year at the end instead of the beginning. Your portfolio grows through the good early years, absorbs the bad year from a much bigger base, and finishes year 30 with money to spare.
A back-of-envelope table
Rough numbers from a 30-year simulation, $1M start, $40K annual real withdrawal, actual S&P sequences (real returns, inflation-adjusted). Note this is 100% stocks — harsher than Bengen's balanced portfolio, which is why 1966 fails here while it (barely) survived his backtest:
| Retirement year | Balance at year 5 | Balance at year 15 | Balance at year 30 |
|---|---|---|---|
| 1966 (worst historical case) | ~$720K | ~$450K | ~$0 (year 24) |
| 1929 (Great Depression) | ~$600K | ~$700K | ~$1.1M |
| 2000 (dot-com + '08) | ~$800K | ~$650K | ~$500K |
| 1982 (best historical case) | ~$1.6M | ~$3.5M | ~$8M+ |
Same 4% rule. Same 30-year horizon. Radically different outcomes based entirely on which sequence of returns you got.
What actually mitigates it
Sequence risk isn't a reason to abandon FIRE — it's a reason to build a plan that doesn't require the market to cooperate in years 1–5.
- Cash and bond buffer. A common approach: hold 2–5 years of expenses in cash or short-term bonds. In a bear market, spend from that bucket instead of selling equities at a loss. This gives the stock portfolio time to recover.
- Guardrails (Guyton-Klinger). Dynamic withdrawal rules: raise spending in good years, cut it modestly in bad years. Trades a constant real income for portfolio survival.
- Delayed Social Security. Waiting until 70 to claim increases your check by ~76% vs claiming at 62. That's an inflation-adjusted, government-backed annuity that reduces the portfolio's job in later years — great insurance against a bad early sequence.
- Bridge accounts. Front-load withdrawals from taxable accounts (which don't have RMDs and are usually less growth-heavy) so tax-advantaged accounts keep compounding.
- Working longer or part-time. Not what the FIRE community wants to hear, but even 1–2 years of "barista FIRE" through a bad first year cuts sequence risk dramatically.
Each mitigation has trade-offs. A cash buffer earns roughly nothing. Guardrails require reducing spending in a downturn — the exact time it's psychologically hardest. Delayed Social Security means using more of the portfolio in the highest-sequence-risk years. There's no free lunch, only levers.
See it in your numbers
The Sequence Risk Simulator on STWLTH lets you plug in your portfolio, withdrawal rate, and horizon, then runs Monte Carlo trials plus the specific historical sequences that broke real retirees (1929, 1966, 2000, 2008). The visual is a fan of trajectories — most of them fine, a stubborn tail of them not. That tail is what a 4%-rule spreadsheet can't tell you. It's what "safe" actually means: high probability, not certainty.
If you're within five years of pulling the trigger, this is the tool that shows whether your plan survives a bad start, or only survives a good one.