How to Think About an Allocation
Not what to hold — how to reason about a mix, price its risk honestly, and see which sleeve is actually driving the outcome.
The order that makes this tractable
- Allocation debates usually start with percentages. That is the last step, not the first.
- The order that works: horizon → what each sleeve is for → risk you can actually sit through → the mix → what the mix does in a bad year → what it costs. Percentages fall out of the first four.
- Nothing on this page is a recommended allocation. The numbers are illustrative inputs to arithmetic, and the arithmetic is the point.
1. What each sleeve is actually for
| Sleeve | Its job | When it fails at that job |
|---|---|---|
| Equities | Long-run real growth | Any decade you need the money in |
| Government bonds | Ballast, and cash-flow matching | Inflation shocks — see 2022 |
| Cash | Liquidity and optionality | Always, in real terms |
| Alternatives / real assets | A different return driver | When the driver turns out to be equity beta with a fee |
| Inflation-linked | Contractual real protection | Rising real yields, which is exactly the shock |
- If a sleeve has no job, it has no place. "Diversification" is not a job; absorbing a specific shock the rest of the portfolio cannot is.
- Two sleeves with the same job are one sleeve. Forty tech stocks and a growth fund are the same position twice.
2. Risk shares, not capital shares
The single most useful thing a multi-asset calculation shows is that the pie chart lies. A 60% weight in the highest-volatility, highest-correlation sleeve routinely carries 80% of the portfolio's risk.
Interactive: five-sleeve allocation
Weights, expected returns and volatilities are yours to change. The correlation between equities and bonds is the input that matters most — it was negative for two decades and positive in 2022.
- Weights
- —
- Expected return
- —
- Volatility
- —
- Diversification benefit
- —
- Sharpe ratio
- —
- Real return
- —
- A bad year
- —
- A 2008-style shock
- —
- What drives the risk
- —
Move the equity–bond correlation from 0.2 to 0.6 and watch the diversification benefit shrink without a single weight changing. That move is what 2022 was. The "bad year" line assumes a normal distribution and is therefore the optimistic case — compare it with the shock line beneath, which makes no distributional claim at all.
3. The correlation assumption is the whole model
- Equity–bond correlation is regime-dependent: negative in demand-shock decades, positive in inflation-shock ones. A portfolio built on the first behaves very differently in the second.
- Stress it deliberately. Re-run the mix with all correlations at 0.8. If the result is unacceptable, the portfolio depends on peacetime staying peaceful — see diversification.
- Alternatives are the most over-credited sleeve. Reported correlations are computed from smoothed, appraisal-based valuations. The measurement is calmer; the asset is not.
- Correlation rises toward one in a crisis because everything is priced by the same variables — liquidity, margin calls, risk limits — rather than by fundamentals.
4. What a bad year looks like, two ways
- The statistical way — a percentile of the modelled distribution. Fast, comparable, and systematically too small, because tails are fatter than any calibration on recent history admits.
- The scenario way — state the moves and add up the damage. It makes no probability claim, which is why it survives regime changes. The stress test does this properly across equities, rates and spreads.
- The honest test is behavioural: could you hold this through the shock number without selling? A mix that is optimal on paper and unholdable in practice is worse than a duller one you keep.
5. Then the parts most people skip
- Costs come off the real return, not the nominal one. At 2.5% inflation and 6% nominal, a 1% fee is nearly a third of the real return — the cost calculator compounds it out.
- Rebalancing is the enforcement arm. Without it the mix drifts into whatever performed best, and the risk profile changes with nobody deciding. Bands beat calendars — see rebalancing.
- The withdrawal phase is a different problem. While saving, the order of returns is irrelevant; while withdrawing, it is decisive.
- Currency is a position taken by omission. Decide it explicitly rather than inheriting it from where the funds happen to be listed.
The checklist
- What is the horizon, and does any sleeve have a job that does not fit it?
- Which sleeve carries the most risk, and does that match what I thought I owned?
- What happens at correlation 0.8 across the board?
- What does the shock scenario cost, and could I hold through it?
- What is the real return after costs, not the nominal one before them?
- What rule brings it back when it drifts, and did I write it down before I needed it?
Information and education only. This page describes a way of reasoning and a simplified model. It contains no recommended allocation, no suggestion that any mix suits you, and every default is an illustration. Expected returns are assumptions you supply, not forecasts this site makes.