Two people retire on the same day with the same $1,000,000. Over the next 30 years, they experience the exact same annual returns, in a different order. One ends with $10.4 million. The other runs out of money.
That is not a hypothetical. It is what happens when you take the actual S&P 500 returns from 1966 to 1995 and reverse them. Same 30 numbers, same 10.62% annualized return, two completely different lives. This is sequence-of-returns risk, and it is the single most misunderstood risk in early retirement planning — not because people underestimate it, but because most of the advice about it is written for someone else.
I want to make a narrower argument than the usual one. Sequence risk is real, and it is brutal. But it is a withdrawal problem, and it only exists when money is moving out of your portfolio. For anyone on a Coast FIRE path, the longest and most consequential stretch of the whole plan — the coast itself — has literally zero sequence risk. Not "low." Zero. And the standard playbook that gets sold to coasters is designed to protect against a risk they will not face for another 20 years, at the cost of the growth they actually need.
What sequence-of-returns risk actually is
Start with the mechanism, because the mechanism explains everything downstream.
When you are saving, a bad year is a discount. Your monthly contribution buys more shares, and those shares participate in the recovery. When you are withdrawing, a bad year is a forced sale. You sell shares at a depressed price to pay rent, and those shares are gone before the recovery arrives. The same market decline that helps you at 35 hurts you at 65, and the only thing that changed is the direction of cash flow.
The asymmetry is easy to see in one year. Take a $1,000,000 portfolio withdrawing $40,000 a year. In a +10% year, you withdraw 3.6% of the portfolio after growth. In a −30% year, that same $40,000 is 5.7% of what is left — you sold 57% more shares for the same amount of money. Those extra shares never come back. Recovery from $660,000 to $1,000,000 requires a 51.5% gain, not a 30% one.
But the honest boundary condition matters here: sequence risk requires a withdrawal. If you have no cash flowing out and no cash flowing in, the ending balance is just principal × (1+r₁) × (1+r₂) × … × (1+r₂₀). Multiplication is commutative. I tested this directly — I took the actual 1966–1985 S&P 500 returns, shuffled them into 2,000 random orders, and ran $500,000 through each one with no deposits and no withdrawals. The spread between the best and worst outcome was $0.00. Not small. Zero, to the cent.
None of that is my discovery. The boundary condition is standard, and every careful treatment of the subject says it in plain language: sequence effects need cash flows. What I wanted was the size of the effect on real data rather than on a hypothetical, and the size turns out to be exactly zero.
So the question "how much sequence risk am I exposed to" is really the question "how much of my money is moving in or out, and when."
The real numbers are worse than the textbook example
Most explanations of this use invented returns — +10%, −10%, +10% — which makes it feel like a math puzzle. It is more useful to look at what actually happened to real people.
The 1966 retiree is the famous worst case, and for good reason. Someone who retired on January 1, 1966 with $1,000,000, took out 4% adjusted for inflation every year, and held 100% S&P 500 was out of money by 1985 — nineteen years, not thirty. Bumping the withdrawal rate to 3.5% pushed depletion to 2004. At 3.0%, the portfolio survived the full 40 years I tested and ended at $6.8 million. A single percentage point of withdrawal was the difference between running dry at 60 and dying rich.
Compare that to 1982. Same $1,000,000, same 4% rule, same 30 years, same asset. The 1982 retiree ended with $14.36 million and never dipped below $1.16 million. The 1973 retiree — also a rough start, but a different decade — ended with $4.18 million. Same country, same stock market, same rules. Outcomes spanning from zero to fourteen times your starting capital, decided by the year you happened to stop working.
Here is the uncomfortable part. The 1966 retiree did not do anything wrong. They saved, they diversified by holding a broad index, they withdrew at a rate that a mountain of research called safe. They got unlucky about a date. And no amount of discipline fixes a date.
Which is why the "just stay the course" advice, while directionally correct, is also incomplete. Staying the course works when your course has time to recover. The 1966 retiree's course ran out of time in year nineteen.
Same returns, two endings — and what that difference really measures
The 1966-to-1995 reversal is the cleanest demonstration I know, so let me put the full numbers on the table.
Take the 30 actual S&P 500 total returns from 1966 through 1995. A retiree withdrawing 4% inflation-adjusted from $1,000,000 in that order is wiped out. Now hand the same 30 returns to an identical retiree in reverse order — 1995's return first, 1966's last. That person ends with $10,381,874. The geometric mean of the return set is 10.62% in both cases. Nothing about the market changed except the order of arrival.
That gap — roughly $10.4 million, from $0 — is the largest number in this article, and it is also the number most likely to be misused. Because if you assume sequence risk is the dominant risk, you will spend your planning energy on the wrong thing.
It is not the dominant risk. Javier Estrada at IESE Business School decomposed retirement outcomes into two pieces: cohort risk, which is the return environment you happen to be born into, and pure sequence risk, which is the ordering of returns within that environment. His finding, as summarized by Michael Kitces, is that roughly three-quarters of the variation in retirement outcomes comes from the cohort, and only about one-quarter from the ordering.
I did not want to take that on faith, so I ran my own version. Using 69 rolling 30-year windows from 1928 to 1995, with 2,000 random orderings per window, I got 72.7% cohort risk and 27.3% pure sequence risk. Essentially the same split, from an independent implementation. That is about as close to a replication as this kind of exercise gets.
The practical translation: the scary $10.4 million figure above is real, but it is the smaller half of the story. What actually determines whether you make it is which 30-year environment you get — and the spread there is enormous. The 1966–1985 stretch delivered 2.08% real annualized on the S&P 500. The 1976–1995 stretch delivered 8.80% real. That is a 6.7-point gap in real return, sustained for two decades, and no withdrawal strategy manufactures 6.7 points out of nothing.
Why Coast FIRE sits in the one phase with zero sequence risk
Now the part that gets dropped when this research is translated into FIRE advice.
Coast FIRE works like this: you front-load saving early, then stop contributing once your existing balance can grow into a full retirement number by traditional retirement age without any further deposits. You keep working, you cover your expenses, but the portfolio is left alone to compound.
Read that description again with the mechanism from earlier in mind. During the coast, money is not flowing out. Money is not flowing in. The portfolio is a closed system. And a closed system has no sequence risk — that is exactly what my $0.00 shuffle test showed.
This is not a semantic trick. It has a hard consequence: the entire bond tent and rising-glidepath literature, which is the best-known answer to sequence risk, is solving a problem that a coasting 45-year-old does not have. The Pfau–Kitces research that established the rising equity glidepath is explicitly about portfolios in drawdown — the whole argument is that a conservative start protects you from selling into a decline, and a rising equity allocation keeps you exposed when the good years finally arrive. If you are not selling, the first half of that logic does not apply to you, and the second half — staying exposed for the recovery — is the only half that matters. Which means the correct allocation during the coast is the aggressive one, not the defensive one.
Put differently: a 45-year-old with $600,000 who is coasting and has no intention of touching the money for 20 years is in a structurally different position than a 65-year-old with $600,000 drawing 4%. They share a portfolio size and nothing else. Giving them the same advice because both are "retirement-adjacent" is a category error.
And here is where I part company with a lot of FIRE content. The common framing — "watch out, a crash early in retirement can ruin you, so de-risk as you approach your number" — is correct for the retiree and actively harmful for the coaster. A crash at 45 while you are still earning and still not withdrawing is a buying opportunity in everything you already own. De-risking into it converts a discount into a permanent haircut.
The boundary condition is important, and it is where the real risk lives: the day you stop contributing is the day this logic flips. Not the day you retire. The day you stop adding.
The window that actually matters: the year you stop contributing
Everyone talks about the first five years of retirement. Morningstar's 2025 research found that nearly 70% of simulated failures involved portfolios that had lost value by the end of year five. That is a well-earned warning — for people in drawdown.
For a Coast FIRE plan, the equivalent window is not retirement. It is the five years straddling your stop-saving date, and it is dangerous for a reason that has nothing to do with withdrawals: it is the moment your portfolio is large and your future contributions are gone.
Consider the arithmetic. At 45, your remaining human capital — the future paychecks you could still save from — is your real shock absorber. A 30% decline on $500,000 costs $150,000, and you can plausibly save your way back from that at $15,000 a year over a decade. A 30% decline on $2,000,000 costs $600,000, and you cannot. Once you stop contributing, the absorber is gone, and the portfolio has to heal itself using only its own returns.
I ran the comparison to see how big this effect is, using real returns. Start at 35 with $150,000 and save $15,000 a year for ten years. If the bad years (1966–1975) come first, you end at $1,812,208 after the following good decade. If the good years come first and the bad decade arrives second, you end at $1,457,835. The bad-early path wins by $354,373 — because you were buying through the decline the whole time.
Now change one thing: add a 4% withdrawal. Same 20 years, same returns, same starting balance. Bad years first: $0. Good years first: $1,062,779. The order effect goes from a $354,000 advantage to a million-dollar wipeout, and the only variable that changed is the direction of cash flow.
That is the whole thesis in two lines of output. While you are adding, bad years early help. While you are doing neither, order is irrelevant. While you are subtracting, order decides everything. Coast FIRE lives in the middle state, which means the risk is not spread across your plan — it is concentrated at the two transitions: the moment you stop adding, and the moment you start subtracting.
What actually determines a coasting portfolio's fate
If order does not matter during the coast, what does? The average return over the whole coast period — the cohort effect that carried 72.7% of the variance in my test.
The magnitude is not subtle. Take $500,000 at 45 and let it coast 20 years with no contributions:
| Real annualized return | Value at 65 |
|---|---|
| 3% | $903,056 |
| 5% | $1,326,649 |
| 7% | $1,934,842 |
| 9% | $2,802,205 |
A 4-point difference in real return — 5% versus 9%, roughly the gap between a mediocre and a good two decades — is a $1.48 million difference on a $500,000 base. That is larger than anything the sequence-risk conversation offers you, and it is not something you can hedge with a bond allocation, because bonds do not generate the missing return; they only reduce the variance around it.
Which reframes the planning question. For a coaster, the thing worth optimizing is not "how do I protect against a bad first decade." It is "how confident am I in the real return assumption my coast number is built on, and what happens to my plan if I am wrong by two points."
The usual objection: doesn't that mean coasters should just take maximum equity risk and hope? No — and this is where I would push back on my own argument. There is a real cost to being wrong in the other direction. If you coast at 45 on a 7% real assumption and the next 20 years deliver 3% real, you arrive at 65 with $903,000 instead of $1.93 million. That is not a market-timing problem; it is a plan built on a number that did not hold. The mitigation is not a defensive allocation, it is a margin of safety in the coast target itself — coast to a number that works at 5% real, not at 7%.
De-risk at the withdrawal transition, not the contribution one
Three things, and they are not the standard list.
Stop calling the coast a retirement phase. The vocabulary is doing damage. If you think of yourself as "retired" at 45, you will import retirement risk management — cash buckets, bond tents, 3.9% safe withdrawal rates — into a phase where you have a paycheck and no withdrawals. Keep the portfolio aggressive while you are earning and not spending from it. The de-risking belongs at the withdrawal transition, not the contribution transition.
Set the stop-saving trigger on a real-return stress test, not on a point estimate. If your plan only works because 7% real shows up for 20 straight years, you have not found a coast number — you have found a hope. Recompute your coast target at 5% real. If you can still coast on that, the plan has a margin. If you cannot, you are not ready to stop saving, regardless of what the 7% version of the spreadsheet says.
Hold one year of spending in cash at the moment you start withdrawing, and not a day earlier. This is the one piece of the bond-tent logic that survives translation to Coast FIRE. Its purpose is not to reduce volatility; it is to make sure that when you begin taking money out, you are never forced to sell into a decline in the first twelve months. Building that buffer five years early costs you real growth for no benefit. Building it the year you need it costs you almost nothing.
If you want to check where you actually stand, run the numbers rather than reasoning about them: the Coast FIRE calculator will tell you the balance you need to stop saving, and the 4% rule calculator will tell you what that balance supports once you start drawing on it. The gap between those two numbers is your margin of safety — and if the gap is thin, the answer is more saving or a lower target, not a cleverer withdrawal rule.
Related articles
- What Is Coast FIRE? The "Stop Saving Early" Retirement Hack Explained
- When Can You Stop Contributing to Your 401(k)? The Coast FIRE Line
- The 25x Rule: How Much Money You Actually Need to Retire
- Coast FIRE Numbers by Age: From 25 to 50
- Lean FIRE vs Fat FIRE vs Coast FIRE vs Barista FIRE
Sources
- Damodaran, A. — Historical Returns on Stocks, Bonds and Bills: 1928–2025, NYU Stern. S&P 500 total returns and 10-year Treasury returns used in every simulation above; CPI series (FRED CPIAUCNS) as distributed in the same workbook.
- Morningstar — The State of Retirement Income: 2025 (Dec. 3, 2025). 3.9% base-case safe withdrawal rate, 5.7% for flexible spending, 30–50% equity optimum over a 30-year horizon.
- Morningstar — The Biggest Risk for New Retirees (Dec. 9, 2025). Nearly 70% of failed trials still showed a loss at the end of year five; the 2000 retiree on a 60/40 mix bottomed near $729,000 after 2008 and recovered to about $1.2 million by age 84.
- Kitces, M. and Pfau, W. — Reducing Retirement Risk with a Rising Equity Glidepath, Journal of Financial Planning, January 2014.
- Kitces, M. — Creating A Flexible Retirement Date "Window" To Mitigate Sequence And Cohort Risk (Apr. 8, 2026). Cites Estrada's variance decomposition: roughly three-quarters cohort risk, one-quarter pure sequence risk.
Disclaimer
All simulations use beginning-of-year withdrawals, annual rebalancing for the 60/40 series, and CPI-adjusted withdrawal amounts. "Real" figures are nominal returns divided by that year's CPI change. Historical results are not a forecast. Nothing here is investment advice; the withdrawal rates cited are research outputs, not recommendations for your situation. Data current through 2025.