Every allocation framework is an attempt to answer one question: how much volatility can this portfolio absorb before its owner does something destructive? "100 minus your age" was a serviceable approximation when bonds paid 8%, life expectancy at 65 was fifteen years and most retirees had a pension underneath them. Two of those three conditions have changed materially.
What follows is not a prescription. It is a structure — the variables that matter, the decade-by-decade shape most portfolios should follow, and the specific mechanism that turns an ordinary bear market into a permanent loss of retirement income.
The three variables that beat age
- Time to first withdrawalNot retirement date — the date money leaves the portfolio. A 62-year-old who will not touch invested assets until 72 has a ten-year horizon, not a zero-year one.
- Income stabilityA tenured professor and a commission-based salesperson with the same age and balance need different equity exposure, because one of them may be forced to sell into weakness.
- Behavioural capacityThe maximum peak-to-trough decline you have personally endured without selling. If you have never been tested, assume your tolerance is lower than you think.
The honest testMultiply your portfolio by 0.65 and look at the number. That is a normal — not catastrophic — equity bear market. If that figure would change your behaviour, your equity allocation is too high regardless of your age.
A decade-by-decade shape
Treat the table below as a starting point to argue with, not a target to hit. The ranges assume a diversified equity sleeve (domestic plus international), investment-grade bonds of intermediate duration, and cash sized against near-term spending.
| Life stage | Equities | Bonds | Cash / short | Primary risk being managed |
|---|---|---|---|---|
| 20s | 90–100% | 0–10% | Buffer only | Not investing enough, and career risk |
| 30s | 85–95% | 5–15% | Buffer only | Interruption risk: property, children, career change |
| 40s | 75–90% | 10–25% | Buffer only | Concentration — employer stock, single property |
| 50s | 60–80% | 20–35% | 1 yr spending | Forced early retirement |
| Retirement −5 yrs | 50–70% | 25–40% | 2 yrs spending | Sequence-of-returns risk |
| Retirement +10 yrs | 40–60% | 30–45% | 2–3 yrs | Longevity and inflation |
| 80s+ | 30–50% | 40–55% | 2 yrs | Inflation, care costs, simplicity |
Notice the equity floor never drops to zero. A 30-year retirement is a long investment horizon, and a portfolio of nothing but nominal bonds is a slow-motion bet against inflation.
Sequence-of-returns risk, concretely
Two retirees each start with $1,000,000, withdraw $40,000 annually adjusted for inflation, and earn the same average return of 7% over thirty years. The only difference is order: one meets a 25% decline in years one and two, the other meets it in years twenty-nine and thirty.
Identical average, radically different outcomes. Withdrawing a fixed dollar amount from a shrunken portfolio converts a temporary decline into a permanent reduction in shares owned. The portfolio then has fewer units to participate in the recovery.
Practical defenceHold two to three years of planned withdrawals outside equities as you approach the fragile zone, and be willing to reduce discretionary spending by 10% in a year following a large decline. Flexibility is worth more than an extra bond fund.
What actually belongs in the bond sleeve
- Intermediate-term investment grade. The core holding. Duration of roughly 5–7 years balances rate sensitivity against yield.
- Treasuries for the crisis correlation. When equities fall hard, government bonds are the sleeve most likely to hold value — corporate credit is closer to equity risk than its label suggests.
- Inflation-linked bonds for retirees. A partial allocation directly hedges the risk that most damages fixed income.
- Short-term for known spending. Money needed inside 24 months belongs in cash equivalents or a short ladder, not in anything that can drop 12%.
- High-yield credit as a "bond" substitute — it falls with equities precisely when you need stability.
- Long-duration bonds for their yield, without understanding that a 1% rate rise can cut the price by 15% or more.
- Complex structured notes marketed as capital protection with equity upside. The fee is in the payoff formula.
Two-minute risk profile
Answer honestly rather than aspirationally. The output maps to an equity band, not a product recommendation.
Rebalancing without kidding yourself
Rebalancing sells what has done well to buy what has done badly. It reliably feels like a mistake, which is why most people abandon it exactly when it matters. Two workable disciplines:
- Calendar-basedOnce a year, same month, no exceptions. Simple, defensible and hard to rationalise your way out of.
- Band-basedRebalance only when an asset class drifts more than 5 percentage points from target. Fewer transactions, better tax outcomes, requires you to actually check.
Cheapest rebalancing toolDirect new contributions to whichever sleeve is underweight. In the accumulation phase this often eliminates the need to sell anything at all.
Where allocation frameworks break
Employer stock. A 20% position in your employer means your salary, your bonus and a fifth of your portfolio share a single point of failure. Treat anything above 10% as a concentration to unwind on a schedule.
Home equity. A house is not a bond. It is an illiquid, leveraged, undiversified real-asset position with carrying costs. Track it on your net-worth statement, but do not count it as ballast.
Multiple accounts. Allocation is a household-level concept. A conservative IRA next to an aggressive taxable account may already be balanced overall — look at the total before adjusting anything.
Limits, errors and borders of the standard advice
Critical reviewAllocation rules of thumb survive because they are memorable, not because they are accurate. Most were coined decades ago under different life expectancies, different bond yields and different retirement structures, and they are repeated by firms whose products they happen to favour. Here is where each breaks.
"Hold your age in bonds" / "110 minus your age in stocks."
The claimA single arithmetic rule tied to age produces a sensible stock/bond split.
Where it breaksAge is a proxy for time horizon, and it is a poor one. Two 60-year-olds with identical ages can have completely different risk capacity: one with a full inflation-linked pension covering all essential spending has an enormous ability to hold equities, while one relying entirely on portfolio withdrawals does not. The rule also ignores the size of the portfolio relative to the spending it must fund — someone with far more than they need can take risk they don't need to take, and someone underfunded cannot de-risk their way to sufficiency. The specific constants were also reverse-engineered from historical returns in one market and one era.
The borderThe rules are a reasonable starting point for someone with no pension, no unusual liabilities and a conventional retirement date — which is why they persist. Replace age with two better inputs: years until the money is spent, and what share of essential spending is already covered by guaranteed income.
"Stocks are always the right answer over long horizons."
The claimEquities beat bonds over any sufficiently long period, so a long horizon justifies maximum equity.
Where it breaksThe evidence is drawn overwhelmingly from a handful of markets that did unusually well, and the most cited series belongs to the most successful economy of the twentieth century — a survivorship problem, not a law. Multi-decade periods of flat or negative real equity returns have occurred in major developed markets within living memory. And "long horizon" is misdescribed: a retiree's horizon is not one thirty-year period but a sequence of annual withdrawals, which makes the order of returns matter as much as the average.
The borderA high equity weight is defensible while contributions continue and no withdrawals are needed for well over a decade. It is much weaker in the years immediately before and after withdrawals begin, where sequence risk — not average return — is the thing that ruins plans.
"Target-date funds are set-and-forget."
The claimA dated fund automates the glide path so no further decisions are needed.
Where it breaksTwo funds with the same year on the label can hold materially different equity weights at the same date, because "to retirement" and "through retirement" glide paths are different products sold under similar names. The date also encodes an assumption about your retirement year that becomes wrong the moment your plans change. And the automation is only complete if the fund holds your entire portfolio: held alongside other accounts, it silently distorts the total allocation while looking correct in isolation.
The borderGenuinely excellent as a single-holding default, and better than most self-managed alternatives. Check the fund's actual equity weight at your target date rather than trusting the label, and stop treating it as automatic once you hold assets elsewhere.
"Rebalance annually to stay on target."
The claimRegular rebalancing controls risk and harvests a return premium.
Where it breaksThe risk-control argument is sound; the "rebalancing bonus" is largely an artefact of specific historical periods and can reverse in a sustained trend. Calendar rebalancing also ignores that the cost is real and asymmetric — in a taxable account, selling to rebalance can realise gains whose tax exceeds the risk-management benefit of a small drift. And the annual cadence is arbitrary: it does nothing during a violent mid-year move, which is exactly when allocation drifts most.
The borderRebalance with new contributions first, which is free. Use tolerance bands rather than the calendar, keep the rebalancing trades in tax-sheltered accounts where possible, and accept small drift rather than paying tax to correct it.
The allocations discussed here are frameworks, not recommendations, and they inherit the same historical data whose limits we just criticised — we have no better dataset than anyone else. Our emphasis on risk capacity over age is more accurate and considerably harder to act on, because it requires you to estimate guaranteed income and essential spending decades ahead. And the whole article assumes a portfolio of diversified public securities; it says nothing useful to someone whose wealth is concentrated in a business, a property or a single employer's stock, which describes a great many people.
Frequently asked questions
Is a target-date fund good enough?
For most investors in a workplace plan, yes — and it is markedly better than a hand-built portfolio that gets abandoned. Check two things: the equity glide path at your retirement year, and the expense ratio. If the glide path is more conservative than you want, choose the fund dated five years later.
Should international equity be included?
Yes, in some proportion. Home-country concentration is a bet that one market will continue to outperform, and that bet has reversed repeatedly across history. Common allocations sit between 20% and 40% of the equity sleeve.
How do I handle a large cash windfall?
Lump-sum investing wins on average because markets rise more often than they fall. Phasing in over three to six months wins on regret minimisation. If the sum is large relative to your net worth, the behavioural argument for phasing usually deserves to win.
Do bonds still make sense at low yields?
Their role is volatility damping and liquidity provision, not yield maximisation. If you hold bonds for the coupon alone you will be disappointed; if you hold them so you never have to sell equities at a loss to eat, they are doing their job.
The bottom line
Pick an equity band using horizon, income stability and honest volatility tolerance. Keep near-term spending money out of markets entirely. Tighten the glide path in the five years before withdrawals begin, hold a genuine bond and cash reserve through the fragile zone, then rebalance on a rule rather than a feeling. Do that and you have solved the part of investing that is actually within your control.
Revision log. Jul 2026: added risk-profile assessment and rewrote the sequence-risk illustration with fresh figures. Reviewed quarterly.