01 · Briefing · what happened

Sequence of returns risk: why the order of your good and bad years decides the outcome

Personal Money 6 min 22 sources

Two people can retire with the same pot, take the same withdrawals, live through the same returns, and finish with the same average - and one still runs out of money. The order of the years is the whole difference, and it only bites once money is coming out.

40 vs 25

years the money lasted

same four returns, different order, $1m pot

$96,318

left after ten years

a 2000 retiree; a 2010 retiree still held over $874,000

2 of 30

years near the average

US shares averaged 10.9% a year from 1995 to 2024

3.9%

safe starting withdrawal

Morningstar's estimate over a 30-year retirement

At a glance

Two retirees can start with the same pot, take the same withdrawals, and live through the same returns. One still runs out of money and the other does not.
The order of the years is the entire difference. It is arithmetic, not bad luck.
Order is irrelevant for money left alone. Multiplying by 1.2 then 0.8 gives the same answer as 0.8 then 1.2.
It starts to matter the moment money moves in or out - and when money is coming out, it matters enormously.
A fall early on forces you to sell more units to fund the same spending, so fewer are left when the recovery arrives.
In one US Bank illustration, the same four returns in a different order meant lasting 40 years or running out after 25.
The danger zone is roughly the ten years either side of the retirement date, not just the day itself.
The risk runs both ways: a strict 4% rule has been found to leave many retirees with far more money than they started with, unspent.

Forces in play

Early losses High→

A fall in the first years forces you to sell more units to fund the same spending, and those units are not there for the recovery.

Fixed spending Building→

A withdrawal that rises with inflation whatever the market does is what turns a dip into permanent damage.

Time to recover Easing

The same fall ten years in lands on a bigger pot with fewer years left to fund, so it does far less harm.

Overcaution Steady→

Fear of a bad run also makes people underspend - one study found a strict 4% rule often ends with more money than it started with.

In play Bill Bengen — devised the 4% rule in 1994; now argues 4.7% is a better starting point Morningstar — forward-looking research putting the safe starting rate at 3.9% Wade Pfau — named the retirement risk zone - the ten years either side of retiring The retiree — cannot choose which years come first, only how much comes out

Where this points

Watch whether the two camps converge: Bengen's historical work has moved up to 4.7% while Morningstar's forward-looking estimate has crept back to 3.9%. The gap between them is the honest size of what nobody knows.

Full briefing

The question nobody asks about an average

Two people retire on the same day. Same amount saved, same plan, same yearly withdrawal. They live through exactly the same set of yearly returns and finish with exactly the same average return. One dies with money left over. The other runs out.

The only difference is which years came first.

This is not a figure of speech. It is arithmetic, and the industry has a name for it: sequence of returns risk [1]. Two retirees with identical portfolios and identical withdrawal rates can post the same average annual return and land in completely different places [2]. It is one of the two hardest problems in turning savings into an income, alongside not knowing how long you will live [3].

Why order suddenly starts to matter

Here is the part that surprises people. For money that just sits there, the order of returns genuinely does not matter at all. Wade Pfau, who researches retirement income, puts it plainly. Invest once, leave it alone for thirty years, and there is no sequence risk [4]. You always end with the same balance. Multiplying a pot by 1.2 and then 0.8 gives the same answer as 0.8 then 1.2.

Order only starts to matter the moment money moves in or out [4].

And when money is coming out, it matters enormously. A fall early in retirement forces you to sell more units to fund the same withdrawal. Fewer units are left when the recovery arrives, so the loss is not merely deferred - it is locked in [1]. That is the whole mechanism. Once assets are spent they can no longer generate the growth the rest of the plan was counting on [1]. Take the same fall ten years later and it lands on a pot that has already had a decade to grow, with fewer years left to fund [1]. Same drop, a fraction of the damage.

It is also why the switch from saving to spending is the dangerous moment. While you are still contributing, a loss is something you ride out. Once the paycheck stops and the withdrawals start, the same loss is much harder to recover from [18].

The numbers, worked

US Bank ran two retirees, each with $1,000,000, each drawing $45,000 a year rising with inflation. Both saw the same four returns in their first four years - just in a different order. The one who got +25%, +10%, +5% and then -15% still had money after 40 years. The one who got -15% first, then +5%, +10% and +25%, ran out after 25 years [1].

Fifteen years of retirement, decided by a shuffle.

A second illustration, from the real record. Two savers each started with $500,000 and each drew $30,000 a year. One began in 2010, one in 2000. A decade in, the 2010 retiree had taken out $300,000 and still held more than $874,000. The 2000 retiree had $96,318 left [5].

The scale of the sensitivity is easy to miss. On a $1,000,000 pot drawing 4%, a 10% fall in year one leaves about $860,000 rather than the $960,000 a flat year would have left [1]. That missing $100,000 never gets to compound.

The common mistakes

The first is trusting the average. US shares returned an annualised 10.9% a year from 1995 to 2024. In only two of those thirty years did the actual return land within two points of that average [2]. An average is a summary of a bumpy road, not a description of any year on it.

The second is assuming this is only a retirement problem. Pfau names a “retirement risk zone” of roughly the ten years either side of the retirement date [4]. In the final years of saving, a fall hits the full accumulated balance rather than a small early pot, so those years carry unusual weight too [4].

The third is confusing this with an ordinary bear market. A bear market is usually defined as shares falling at least 20% from a high, and that threshold is an arbitrary line rather than a special event [22]. Falls of that size are common. Only four times since 1929 has the broad US market fallen in two consecutive calendar years [1]. Rolling ten-year returns have been negative only about 4.6% of the time since 1926 [6]. Bill Bengen devised the 4% rule in 1994. He argues a bear market temporarily lifts a withdrawal rate from perhaps 5% to 11%, then usually resolves itself [7]. Sustained high inflation does not [7]. Sequence risk is not the only thing that erodes a pot; inflation, market swings and health costs all pull at it [19].

The fourth is treating it as a market problem alone. The other half is how long the money has to last. The number of Americans aged 65 or over is projected to reach 95 million by 2060, up from roughly 56 million in 2020 [20]. Running out is consistently near the top of what retirees say they fear, partly because a late care bill can arrive as one enormous payment at the end [21].

What is genuinely uncertain

There is no agreed number for how much to draw. Bengen’s original 4% has become 4.7% in his 2025 work, on a portfolio of up to 65% shares [8]. Morningstar’s forward-looking research puts the highest safe starting rate at 3.9%. That assumes a thirty-year horizon, a 90% success rate, and a stock weighting between 30% and 50% [9]. It is up from 3.7% the previous year, and 3.3% back in 2021 [9]. Those are estimates for someone starting out, not a dial to be reset every year [10].

The risk also runs the other way, which is rarely said. One analysis found something odd about a strict 4% rule. In a median market, it leaves a retiree after thirty years with half again as much wealth as they started with, in real terms [11]. That is money never spent. Morningstar’s own testing found that fixed inflation-adjusted withdrawals produced the lowest lifetime spending and the biggest leftover balances of the systems it compared [12]. A flexible constant-percentage rule supported a 5.7% starting rate [13].

The defences people describe all work the same way: they put something between a bad year and the need to sell. Cash and bond buckets, which hold a year or two of spending in cash and several more years in bonds [14][15]. A glide path that shifts the mix of shares and bonds as the date approaches [16]. Guaranteed income bought in, at the cost of inflation protection [17]. Or simply spending less in a bad year [9]. None removes the risk. They change who carries it, and each has a price.

The one thing to carry: an average return tells you almost nothing while money is coming out. The order is the story.

02 · Lesson · why it matters

The order of the years, not the average, decides who runs out

Order is completely irrelevant while money sits still. The moment it starts flowing out, order becomes the one thing that decides the outcome.

How it works

A pot left alone ends the same whatever the order of returns
Money moving in or out breaks that
A fall while withdrawing forces selling more units at a low price
Fewer units remain when the recovery arrives
The shortfall compounds against every later year
So a bad year early costs far more than the same year late

The twist

The order of returns is completely irrelevant right up until the moment you start taking money out - and then it becomes the single thing that decides the outcome.

Where you've seen this

A shop's opening year

a bad first season eats the float that would have funded the good years behind it

Paying down a debt

a rate spike early leaves a bigger balance compounding for the whole term

A reservoir in drought

a dry year on a full reservoir is survivable; the same dry year on a low one is not

Any tank being drained

when something is flowing out, when the shock lands matters as much as how big it is

The catch

It is a real and measurable effect, not a reason to do any particular thing. Fear of it makes people underspend at least as often as bad luck makes them run out.

Full lesson

The number that hides the road

An average return is a strange thing to trust. It is a single number standing in for thirty separate years, most of which looked nothing like it. US shares averaged around 10.9% a year across three decades, and the actual return landed near that figure in only two of them. The average was real. It just never happened.

For most of a saving life that gap does not cost you anything. The pot goes up, the pot goes down, and at the end the average has done its work. You can hold the summary and ignore the road.

Then the direction of the money reverses, and the summary stops being a summary of anything you can use.

Nothing, then everything

Here is the arithmetic, and it is genuinely simple. Multiply a pot by 1.2 and then by 0.8, and you get the same answer as 0.8 then 1.2. Multiplication does not care about order. A sum of money left completely alone for thirty years ends in exactly the same place no matter how the good and bad years are shuffled.

That is not a small point. It is why order feels irrelevant, because for money sitting still it genuinely is.

What breaks it is a flow. Once money is coming out each year, a fall does two things at once. It shrinks the pot, and it forces you to sell more units to raise the same amount of cash. Those units are gone. When the recovery arrives it arrives for a smaller pile, and the shortfall does not catch up. It compounds against every year that follows.

The same fall late in retirement lands on a pot that has already grown, with fewer years still to fund. Identical drop. A fraction of the harm. In one worked case the same four returns, shuffled, meant a portfolio lasting forty years or running dry after twenty-five.

The same fact, with the sign flipped

Now turn the flow around. Someone still paying money in every month faces the same arithmetic, and it works for them.

A bad year early in a saving life lands on almost nothing, and every payment made during it buys more units cheaply. A bad year in the final stretch lands on the whole accumulated balance, which is why the years just before stopping carry weight far out of proportion to their number.

So the order that would ruin a person drawing down is, near enough, the order a person paying in would choose. Same market. Same years. Opposite sign. Nothing about the world changed except which way the money was moving.

The rule that does not bend

Underneath the whole problem sits something that looks like plain prudence and is actually a design choice.

The standard way to plan a retirement income is a fixed percentage of the starting pot, raised each year with inflation, regardless of what the market did. It is clean. It is comparable between people. It makes a plan into a number you can check.

It also treats spending as the fixed thing and the market as the variable, which is precisely the arrangement that lets a bad early run do maximum damage. A rule that never bends transmits the whole shock to the pot.

That arrangement is not a trick. It genuinely helps: it stops people spending wildly in good years and it makes a plan legible enough to argue with. But it was built for clarity, and clarity is not the same as safety. Every defence anyone offers against a bad early run is really a way of letting something bend somewhere else, and each has a cost attached.

Everyone with something running out

This is not really about retirement. It is about anything with a level in it and a flow coming out.

A new business drawing on its float. A household living off a redundancy payment. A charity spending from an endowment. In each case the average tells you almost nothing, and when the bad stretch lands tells you almost everything.

And the ledger runs both ways, which is the part that rarely gets said. Fear of a bad early run also makes people spend far less than they safely could, for decades, and end with money they never used. Both failures come from the same blind spot: treating an average as a description of the road.

The hardest part is that nobody sees their own order until it has already happened. Two people can reason identically, decide identically, and be separated by nothing but which year came first. Whoever came through it well is holding evidence of a shuffle, not of a method. That is worth remembering before reading anyone’s outcome, including your own, as proof of anything.

03 · Lab · your turn

Shuffle the Years

Rearrange the same five years of returns and watch order do nothing while money sits still, then decide everything once it starts flowing.

04 · Hope · carry this

Nobody chooses which years come first. But people worked this out and kept refining it for thirty years, so the shape of the problem is known long before it arrives.