Why the Order of Your Returns Matters More Than the Average

Two retirees. Same average return. Same starting balance. Same withdrawal amount. One runs out of money at 74. The other dies with twice what they started.

The difference? The order in which returns showed up.

This is sequence-of-returns risk - one of the biggest risks in retirement, and one a simple calculator can't show you.

The Problem With Averages

Suppose your portfolio earns an average of 7% per year over 30 years. A simple calculator multiplies that forward and tells you everything is fine.

But "average 7%" can look like a thousand different paths. Maybe you earn 25% in year one and lose 15% in year three. Or maybe you lose 15% in year one and earn 25% in year three. The arithmetic average is the same. The ending balance is not.

When you are adding money (during your working years), a bad early stretch is actually helpful - you buy more shares at lower prices. But when you are withdrawing money, a bad early stretch is devastating. You sell shares at depressed prices to fund your spending, and those shares are gone forever. They cannot participate in the recovery.

This asymmetry is what makes retirement fundamentally different from accumulation.

A Concrete Example

Take 1,000,000 of savings and a withdrawal of 40,000 at the start of each year (a classic 4% withdrawal rate). Two scenarios over 5 years:

Scenario A (bad years early): -15%, -10%, +8%, +18%, +22%

Scenario B (bad years late): +22%, +18%, +8%, -10%, -15%

Both scenarios have the same average annual return of 4.6%. But after 5 years of withdrawals:

• Scenario A balance: about 917,000
• Scenario B balance: about 1,005,000

That is a gap of almost 90,000 from exactly the same returns, after only five years. Over a 25-30 year retirement, the gap can become the difference between money left over and money running out.

Try it yourself - change the amounts, or switch to real US market history and pick a retirement year:

Why the First 5-10 Years Matter Most

The returns in the first 5-10 years of retirement have an outsized effect on whether the money lasts. By the time a decade has passed, the portfolio has either built enough of a buffer to withstand later downturns, or it has been permanently impaired by early losses combined with withdrawals.

This is why someone who retired in 1995 (ahead of a roaring bull market) had a completely different experience than someone who retired in 2000 (right before the dot-com crash), even though their 20-year average returns might look similar.

The retirement date itself becomes a source of risk. Not something you can control, not something you can predict.

How Monte Carlo Captures This

A Monte Carlo retirement calculator runs thousands of scenarios, each with a different randomly generated sequence of returns. Some sequences start with a crash. Some start with a boom. Some have the crash in the middle.

The result is not a single "you will be fine" or "you will not." It is a probability distribution. If 85% of the sequences lead to success, you know that 15% of plausible return orderings would deplete your portfolio. That 15% is sequence risk made visible.

A fixed-return calculator, by definition, tests exactly one sequence: the same return every single year. It has zero ability to reveal sequence risk.

How to Quantify Sequence Risk in Your Own Plan

Knowing that sequence risk exists is not enough. You need to measure how exposed your specific plan is. A Monte Carlo simulation is the only practical way to do this, because it generates thousands of different return orderings and shows you the full range of outcomes.

Here is what to look for in the results:

The gap between the median and the 5th percentile. The median outcome is the "typical" future. The 5th percentile is the outcome where the return sequence was bad - usually because large losses hit early. If your median portfolio at age 85 is 800,000 but the 5th percentile is 0, sequence risk is doing most of the damage. That gap is your vulnerability window made visible.

The iteration count matters here. Sequence risk lives in the tails of the distribution. At 1,000 iterations, you might only generate a handful of truly devastating early-loss sequences. At 10,000 or 50,000, you get a more complete picture of how bad the timing can get. If you care about the 5th percentile (and you should), you need enough iterations to populate it reliably.

Fat-tailed distributions make it worse. Under a normal distribution, a -30% first year is rare. Under a Student's t-distribution, it is merely uncommon. Since sequence risk is driven by extreme early losses, the distribution you choose for your simulation directly affects how much sequence risk it reveals. A Gaussian model will underestimate the probability of the worst sequences.

Historical Retirement Cohorts: Timing as Destiny

The data makes this concrete. Someone who retired in January 2000 with 1,000,000 walked straight into the dot-com crash and the 2008 financial crisis. Withdrawing 40,000 a year from a 60/40 stock and bond mix, they were down to about 847,000 by the end of 2009 - before counting inflation - after selling investments at low prices in the worst years.

Compare that to someone who retired in January 1982 - right at the start of one of the longest bull runs in U.S. history. Same savings, same withdrawals, completely different outcome: the same 60/40 mix grew to about 2.1 million within five years. The early gains built a buffer that could absorb later downturns.

The arithmetic average return across both periods was positive. The sequence of those returns determined whether the plan survived.

What You Can Actually Do About It

Sequence risk cannot be eliminated, but it can be managed:

Build a cash buffer. Keeping 1-2 years of spending in cash or short-term bonds means you do not have to sell equities during a downturn. You draw from the buffer instead and give your portfolio time to recover.

Use a flexible spending strategy. Rigid fixed withdrawals are the worst case for sequence risk. Strategies like Guyton-Klinger guardrails or floor-and-ceiling approaches reduce withdrawals in bad years and increase them in good years. Spending a little less after bad years makes the money last noticeably longer - the price is a less predictable income.

Consider your asset allocation glide path. A slightly more conservative allocation in the first 5 years of retirement (the vulnerability window), shifting back toward equities later, can reduce the impact of early losses. This is sometimes called a "bond tent" or "rising equity glide path."

Test with realistic distributions. A Monte Carlo simulation that uses a normal distribution will underestimate the frequency of extreme early losses. One that uses fat-tailed distributions gives a more honest assessment of sequence risk - because extreme years are exactly the ones that do the most damage.

The Takeaway

Sequence-of-returns risk is not an edge case or a theoretical curiosity. It is one of the main reasons retirement plans fail. Two people with identical savings, identical spending, and identical long-run average returns can have opposite outcomes based purely on timing.

Any retirement projection that does not account for this - that gives you one number based on one assumed return - is not telling you what you need to know.