Asset allocation

Why the split dominates the picks

Brinson, Hood and Beebower's 1986 study of large pension plans found that a fund's policy asset allocation explained the overwhelming majority of the variation in its returns over time — security selection and market timing accounted for very little. That finding became the most repeated line in retail investing, usually in a distorted form.

The distortion is worth knowing. Later work, notably Ibbotson and Kaplan in 2000, showed the result answers a narrow question: how much of a single portfolio's return variability across time is explained by its policy mix. It does not mean allocation explains that share of the difference between your results and your neighbour's — on that question, allocation's share is considerably smaller, and fees and behaviour do real work. The defensible version is still strong enough: your stock/bond mix is the main lever on how much your portfolio swings, and swings are what people react to. Which fund you buy is an index fund question with a short answer; how much of it you buy is this page.

The age rules, side by side

The classic shortcut is "100 minus age in stocks." Longer lifespans and thirty-year retirements pushed the number up, so the versions in circulation now are 110 and 120 minus age. Target-date funds do something similar but hold a high stock weight much longer, then glide down sharply near the retirement year.

Stocks / bonds. The glide-path column is a rounded illustration of how mainstream target-date series behave assuming retirement at 65; individual fund families differ, and several keep gliding for years past the target date.

Age is a proxy, and a poor one

What the rules are actually trying to measure is time to withdrawal and risk capacity. Age correlates with both, badly. A 30-year-old contractor with lumpy income, no cash buffer, and a plan to buy a house in four years has less capacity to hold 90% stocks than a 55-year-old with a pension covering her fixed costs and no need to sell anything for a decade. The rule hands them the opposite answers.

Two better questions. When do you need the money? Not "when do you retire" — retirement is a thirty-year withdrawal schedule, not a date, so a 65-year-old still has money with a 25-year horizon. Cash needed within about five years does not belong in stocks; money you will not touch for twenty can be aggressive regardless of your age. If you are targeting financial independence early, the horizon is longer than the calendar suggests, not shorter.

How much volatility will you actually sit through without selling? The honest measurement is behavioural, not hypothetical. What did you do in March 2020, or in 2022, or in 2008 if you were investing then? Did you stop contributing? Move to cash? A risk questionnaire asks how you would feel about a 30% decline in the abstract, which is a question nobody answers accurately while their balance is at an all-time high.

Price the ride before you buy it

Volatility is easier to plan for as a dollar figure than a percentage. Here is roughly how far each mix has fallen from peak to trough in major US bear markets, applied to a $500,000 portfolio:

Approximate illustrative ranges keyed to major US bear markets (1973–74, 2000–02, 2007–09), assuming a broad US stock index paired with investment-grade bonds and periodic rebalancing. These are order-of-magnitude figures for planning, not precise sourced returns; bonds have not always cushioned stocks — in 2022 both fell together.

Read the bottom row of your chosen mix out loud with your own balance in it. If the number makes you want to change the plan, change the allocation now, while it is cheap. Adding bonds costs expected return; the millionaire calculator will show you what a lower assumed return does to the timeline. Pay that cost deliberately rather than discovering mid-crash that you bought a ride you cannot sit through.

Rebalancing

Markets drag your mix away from target. A 70/30 portfolio through a strong stock run becomes 80/20 — more risk than you chose, arriving without a decision. Rebalancing sells the winner and buys the laggard to restore the target.

Two triggers, both fine. Calendar: check once a year on a fixed date. Threshold: act only when a holding drifts more than about 5 percentage points from target, so a 70/30 gets rebalanced at 75/25. Threshold bands respond to actual drift; calendar checks are easier to remember. Doing both — look annually, act only outside the band — is a reasonable default. More frequent rebalancing does not improve results and generates costs.

Be clear about what it buys: rebalancing is risk control, not a return booster. Over long periods a portfolio that never rebalanced usually ends up with more money, because it drifted toward stocks. Rebalancing keeps you at the risk level you can hold. Do it inside a 401(k) or Roth IRA wherever possible, where selling triggers no capital gains tax. In a taxable account, rebalance with new contributions instead: direct incoming automatic investments to whichever side is underweight and let the drift correct itself without a sale.

The one you will actually keep

Stocks and bonds carry the argument, but the same logic covers anything you add — a slice of real estate or a tilt toward dividend payers changes the ride, not the rules. Two portfolios, one 80/20 and one 60/40, will not differ much over forty years compared with the difference between staying invested and going to cash in year twelve.

So the worst allocation is not the suboptimal one. It is the one you abandon at the bottom, which converts a temporary decline into a permanent loss and usually parks the proceeds in cash until the recovery is over. A single target-date fund, held without interference, beats an elegant seven-fund split you stop maintaining in year three. Pick the mix whose worst year you can describe without flinching, then leave it alone — the rest of the plan is in the investing pillar.

Educational content, not financial, tax, or legal advice. Figures are illustrations based on stated assumptions, not guarantees; markets involve risk, including loss of principal.