Sequence-of-Returns Risk, Up Close
Why the first five years of retirement matter more than the next twenty-five, and what actually protects you.
Every retirement calculator asks you for an expected return. Ten percent, nine, eight. You type a number, it draws a smooth line, and the line tells you whether your money lasts.
Markets never deliver that smooth line. They deliver good years and bad years in some order, and for a retiree the order matters almost as much as the average. In the bucket strategy post I showed two retirees with identical returns, one of whom ran out of money in year 19 while the other finished with ₹1.95 crore. This post looks at that effect up close: when it bites hardest, how often it decides the outcome, and which defences actually work.
The setup
All the numbers below use one retiree, so the results are comparable:
- Retires at 60 with ₹1.5 crore.
- Withdraws ₹4.5 lakh in the first year (3% of the corpus), raised by 6% every year for inflation.
- Needs the money to last 30 years, to age 90.
- The portfolio earns 9% a year on average.
I'll come back to why I chose 3% rather than the more famous 4%.
Experiment 1: one bad year, moved around
Start with a world where the portfolio earns exactly 9% every single year, except for one year in which it falls 30%. Same crash, same size. The only thing that changes is which year of retirement it lands in.
| The −30% year arrives in | Money left at 90 |
|---|---|
| No crash at all | ₹7.60 Cr |
| Year 1 | ₹0.69 Cr |
| Year 5 | ₹1.49 Cr |
| Year 10 | ₹2.37 Cr |
| Year 15 | ₹3.13 Cr |
| Year 20 | ₹3.80 Cr |
| Year 25 | ₹4.38 Cr |
A crash in the first year leaves this retiree with less than a tenth of what the no-crash retiree has at 90. The same crash in year 25 still hurts, but leaves more than six times as much.
The reason is simple arithmetic. In year one you are withdrawing from a pot that just fell 30%, so each rupee you take out is a bigger slice of what is left, and those sold units never get to ride the recovery. Twenty-five years in, the pot has already done most of its work and you have fewer years left to fund.
The red zone cuts both ways
Sequence risk isn't only a retirement problem. It starts in the last years before you stop working, because that is when your portfolio is largest.
Take a saver investing ₹50,000 a month, raising the SIP by 7% a year, for 25 years at 10%. Again, one −30% year, moved around:
| The −30% year arrives in | Corpus after 25 years | Change |
|---|---|---|
| No crash | ₹11.90 Cr | |
| Year 1 of saving | ₹11.66 Cr | −2% |
| Year 10 | ₹9.80 Cr | −18% |
| Year 20 | ₹8.21 Cr | −31% |
| Year 25, just before retiring | ₹7.57 Cr | −36% |
A crash at the start of a career is almost a gift: you keep buying cheap for decades. The same crash in the final year wipes out more than a third of the corpus, at exactly the moment you need it.
Put the two tables together and you get what planners call the retirement red zone: roughly the five years before and the five years after you stop working. Returns in that window shape the outcome far more than returns at any other time.
Experiment 2: ten thousand retirements
One crash at a time is a clean illustration, but real markets mix good and bad years at random. So I ran the same retiree through 10,000 simulated 30-year retirements. Each year's return was drawn at random, averaging 9% with a standard deviation of 13%, roughly what a balanced equity and debt portfolio might show.
84% of the retirements lasted the full 30 years. That headline hides the interesting part.
| What the first five years looked like | Money lasted 30 years |
|---|---|
| Above-average returns in years 1–5 | 97% |
| Below-average returns in years 1–5 | 72% |
| The worst 10% of starts | 39% |
| The best 10% of starts | 99.7% |
And looking at it from the other side: 89% of the retirements that failed had below-average returns in their first five years.
That is the whole story in one number. Retirements rarely fail because of what happens in their third decade. They fail because a poor start forces the retiree to sell too much, too cheaply, too early, and the portfolio never catches up.
What actually helps
Here is what I tested, using the same 10,000 market histories so the comparisons are fair.
1. Start with a lower withdrawal
The single most powerful lever is how much you take out in year one.
| First-year withdrawal | Share of corpus | Money lasted 30 years |
|---|---|---|
| ₹3.75 L | 2.5% | 93% |
| ₹4.5 L | 3% | 84% |
| ₹6 L | 4% | 59% |
This is why I used 3% throughout. The famous "4% rule" comes from US data with US inflation of around 3%. With Indian inflation of 6% and withdrawals that rise with it, 4% fails in four out of ten of these simulations. It is not a safe number here.
2. Build an income floor
This is the bucket strategy. Set aside exactly enough in fixed income (assumed at 7%) to pay the first few years of withdrawals, and leave the rest invested and untouched until it is needed.
| Income floor | Set aside | All retirements | Worst 10% of starts |
|---|---|---|---|
| None | ₹0 | 84% | 39% |
| 5 years | ₹22.1 L | 85% | 43% |
| 10 years | ₹43.2 L | 87% | 54% |
| 15 years | ₹63.3 L | 90% | 64% |
Look at the two columns separately. On average, the floor helps a little. In the bad-luck cases, which are the ones you are actually insuring against, a 15-year floor lifts the chance of the money lasting from 39% to 64%.
The floor works because a crash in the early years no longer forces you to sell anything. Your bills are paid from the safe bucket while the growth bucket has time to recover. That is exactly the mechanism that failed in Experiment 1.
3. Spend a little less after a bad year
A popular piece of advice is to tighten your belt after the market falls. I tested a simple version: in any year after a negative return, withdraw 10% less than planned.
It helped, but modestly: overall success went from 84% to 86%, and in the worst 10% of starts from 39% to 43%. Flexibility is worth having, but a 10% trim for a year at a time can't fix a bad start on its own. It works best on top of a sensible withdrawal rate and an income floor, not instead of them.
4. Reduce equity on the way into the red zone
The accumulation table is the argument for a glide path: shifting gradually from equity to debt in the years before retirement, so a late crash hits a smaller equity slice. It costs some expected growth in exchange for a much narrower range of outcomes right when the range matters most. Both Nivesh Planner and Invest Planner build this into their year-by-year allocation.
What to take from this
- The first five years of retirement, and the five before it, matter more than any others. Plan for them specifically.
- Average returns are not a plan. A plan has to survive a bad start, not just a typical one.
- Your starting withdrawal rate is the biggest lever you control. In Indian conditions, 3% is a much more defensible starting point than 4%.
- An income floor protects you where it counts. Its value shows up in the bad-luck cases, not in the average.
- Flexibility helps, but don't count on it alone.
If you want to see these ideas with your own numbers, Nivesh Planner sizes a 15-year income floor and two growth buckets for your expenses, and shows the equity glide path both before and after retirement. Invest Planner does the same in dollars.
How these numbers were made: a small simulation of a single retiree, with withdrawals taken at the start of each year and returns drawn from a normal distribution (mean 9%, standard deviation 13%, 10,000 runs, fixed random seed). It ignores taxes, assumes inflation is a steady 6%, and real markets have fatter tails than a normal distribution, so the bad cases may be worse than shown. Treat the results as directional, not as precise probabilities. The idea of testing withdrawal strategies against random return sequences is inspired by M. Pattabiraman's retirement bucket strategy simulator at freefincal. This post is for education and is not investment advice.