Asset Allocation: What Your Stock and Bond Split Actually Does
The stock/bond split is the one decision that changes your results more than any fund pick. Here is the arithmetic of expected return, the arithmetic of a bad year, and a fully worked rebalancing example — with the assumptions stated at every step.
Ask ten investors which fund they own and you will get ten answers. Ask them what percentage of their portfolio is in stocks and you will get a number that matters far more. Asset allocation — the division of a portfolio between broad asset classes — sits upstream of every fund selection decision, because it determines the return you are aiming for and the loss you have agreed to accept along the way. It is also the decision most often made by accident, by buying things one at a time and never adding up the total.
This guide works through the arithmetic in three parts: what a split does to expected return, what it does during a bad year, and what rebalancing actually trades. Every number is reproducible with a calculator, and every assumption is stated where it is used.
The one well-known study, and what it really said
The most cited evidence on allocation comes from Brinson, Hood and Singer's 1986 paper in the Financial Analysts Journal, which examined 91 large US corporate pension plans between 1974 and 1983. They found that the plans' policy allocation — the long-run target weights they had committed to — explained roughly 93.6% of the variation in their quarterly returns over time.
That number is quoted constantly, and usually misquoted. The study measured how much of the variation in returns over time was attributable to the chosen policy weights, for that specific sample of pension plans. It did not say that allocation determines 93.6% of your return level, nor that fund selection is worth 6.4%. Later work — Ibbotson and Kaplan's 2000 paper, which asked whether the same statistic explained 40%, 90% or 100% of performance — showed the answer depends entirely on how the question is framed. The defensible summary is narrower and still useful: your long-run policy weights dominate your results over time; the choices inside each sleeve matter much less. Treat the 93.6% figure as a well-known finding with a specific scope, not as a law.
Part one: what a split does to expected return
Expected return of a portfolio is a weighted average of its parts — this part is arithmetic, not theory. Suppose, purely as a stated assumption for this exercise, that you expect a globally diversified equity portfolio to return 7.0% per year and a high-quality bond portfolio to return 3.0% per year. Three splits produce three expected returns:
| Portfolio | Arithmetic | Expected return |
|---|---|---|
| 80% stocks / 20% bonds | 0.8 × 7.0 + 0.2 × 3.0 | 6.2% |
| 60% stocks / 40% bonds | 0.6 × 7.0 + 0.4 × 3.0 | 5.4% |
| 40% stocks / 60% bonds | 0.4 × 7.0 + 0.6 × 3.0 | 4.6% |
Those differences look small. Compounded over a career they are not. Starting from $100,000 and contributing nothing further, with returns applied for exactly 30 years:
| Portfolio | Annual return | Value after 30 years |
|---|---|---|
| 80/20 | 6.2% | $607,765 |
| 60/40 | 5.4% | $484,416 |
| 40/60 | 4.6% | $385,434 |
Worked example: $100,000 × 1.06230 = $607,765 for the 80/20 portfolio. The same calculation at 1.054 and 1.046 gives the other two. The 80/20 split produces the 40/60 ending balance plus another $222,331.
Two caveats belong here rather than in a footnote. First, those 7% and 3% figures are assumptions, not forecasts — changing them changes the gap, and a decade of unusually poor equity returns would compress it. Second, the higher expected return is not free: it is compensation for variability, which is the subject of the next section, and the reason a portfolio's worst year is a more useful planning input than its average year.
Part two: the arithmetic of a bad year
Assume, again purely as a stated scenario, that global equities fall 35% in a bad year while high-quality intermediate bonds rise 5%. Applying the same weighted-average arithmetic:
| Portfolio | Arithmetic | One-year result | Gain needed to recover |
|---|---|---|---|
| 100% stocks | 1.0 × −35 | −35.0% | +53.8% |
| 80/20 | 0.8 × −35 + 0.2 × 5 | −27.0% | +37.0% |
| 60/40 | 0.6 × −35 + 0.4 × 5 | −19.0% | +23.5% |
| 40/60 | 0.4 × −35 + 0.6 × 5 | −14.0% | +12.4% |
| 20/80 | 0.2 × −35 + 0.8 × 5 | −7.0% | +7.5% |
Worked example: for the 60/40 portfolio, 0.6 × (−35) + 0.4 × (+5) = −21 + 2 = −19.0%. Recovery is calculated from the remaining value: after a 19% loss you hold 81% of your starting amount, so the required gain is 1 ÷ 0.81 − 1 = 23.5%. This recovery arithmetic is the reason losses are asymmetric, and it is worth committing to memory.
The important property is not that bonds rise in every crisis — over long history they have not always — but that the two columns move in opposite directions often enough to reduce the worst case. That is what you are buying with the lower expected return, and the size of the reduction is the only honest definition of how much risk you are taking.
Assumption check. The scenario above uses one specific bad year. Real downturns vary enormously in depth, duration and behaviour of bonds. The purpose of the table is to show the mechanics of the arithmetic, not to predict a particular market event.
Part three: rebalancing, with the full arithmetic
A target allocation drifts the moment markets move. Here is the complete sequence for the 60/40 portfolio through the bad year above, starting with $100,000.
| Step | Stocks | Bonds | Total |
|---|---|---|---|
| Start (60/40) | $60,000 | $40,000 | $100,000 |
| After the year: stocks −35%, bonds +5% | $39,000 | $42,000 | $81,000 |
| Drifted weights | 48.1% | 51.9% | 100.0% |
| Target 60/40 of $81,000 | $48,600 | $32,400 | $81,000 |
| Trade | +$9,600 | −$9,600 | $0 |
Worked example: $60,000 × 0.65 = $39,000, and $40,000 × 1.05 = $42,000, so the total is $81,000. Stocks are now 39,000 ÷ 81,000 = 48.1% of the portfolio. To restore the target, stocks should be 81,000 × 0.60 = $48,600, which requires moving $48,600 − $39,000 = $9,600 out of bonds and into stocks. After the trade, bonds hold $42,000 − $9,600 = $32,400, which is 40.0% of $81,000 — the target, restored exactly.
What rebalancing is actually doing is mechanical, and it is worth stating plainly: it sells whatever went up and buys whatever went down, without any view about which will do better next. That is uncomfortable in a falling market, which is precisely when it matters most.
On frequency, the research is unusually forgiving. Vanguard's published work on rebalancing practices found that over long horizons, the difference between checking annually, quarterly, or only when weights drift past a threshold was small — the discipline of having a rule mattered more than the rule's calendar. A common approach is a band: rebalance when an asset class drifts more than five percentage points from target, and otherwise leave it alone. The band reduces trading and, more importantly, reduces the number of decisions a person has to make under stress.
Choosing the split
There is no formula that produces the right number, because the right number depends on facts about you and not about markets: the time until you need the money, income that behaves like a bond (a stable salary, a pension), and the loss you would genuinely tolerate without changing course. The widely repeated heuristic of holding a bond percentage equal to your age, popularised by Vanguard founder John Bogle as a rough starting point, is a heuristic — it ignores pensions, human capital and individual tolerance, and should be treated as a default to adjust from rather than a finding to follow.
Three adjustments to that starting point are more evidence-driven:
- Horizon dominates. Money needed within a few years has no time to recover from a drawdown, whatever the investor's nerves say.
- Capacity to take risk is separate from willingness. A secure pension or a stable income supports a higher equity weight than the same person with variable income and no safety net.
- Write the target down. A number recorded somewhere is what makes drift measurable. An allocation held only in memory cannot be rebalanced, because there is nothing to rebalance to.
The takeaway
Allocation sets both the return you are aiming at and the loss you have agreed to accept, and the weighted-average arithmetic behind it is simple enough to do on a napkin. The historical literature on policy weights is genuinely supportive, provided its most famous statistic is quoted with its actual scope. Pick a split that reflects your horizon and your honest tolerance, write it down, rebalance on a rule rather than a feeling — and then treat the fund selection inside each sleeve as the minor decision it is. For how the sleeves themselves work, the index fund walkthrough covers the low-cost building blocks, and the compound growth calculator will run a fee-adjusted projection on your own numbers.
This guide is general educational information, not financial advice, and it recommends no specific security or allocation. The 7% and 3% return figures and the loss scenario are stated illustrative assumptions, not forecasts. Historical research findings describe specific samples and periods and may not generalise. Consider your full circumstances and consult a licensed professional before making investment decisions.
