Asset Allocation
The split between asset classes is the decision that determines most of a portfolio's variability. It is also the one made least often and reviewed least carefully.
MadStockAlerts Research · Updated August 29, 2026
What to take away
- Allocation across classes explains most of the variability of a portfolio's returns.
- The classes behave differently because they respond to different things.
- Correlations are estimates from a window and are unstable in stress.
- An allocation drifts on its own as the components perform differently.
- There is no allocation that is correct in general, only one that matches a set of constraints.
MAD Academy Training Video · 0:45
The Decision That Outweighs the Rest
How much sits in each asset class explains most of a portfolio's variability — far more than which specific securities were chosen.
This lesson is part of a Stock Alerts + Tools plan.
Why the split dominates
A widely cited body of research finds that the allocation across asset classes explains the large majority of the variability of a portfolio's returns over time, with security selection and timing accounting for far less. The finding is frequently overstated and the direction of it is robust.
The precise claim matters. The research addresses variability over time within a portfolio, not the difference in return between two portfolios, and those are different questions. The commonly repeated version conflates them.
What survives the caveats is still substantial: a portfolio that is eighty percent equities and one that is twenty percent equities will have different experiences in any given year, and the difference between them will swamp the effect of which particular equities were chosen.
What the classes are for
| Class | Responds to | Role usually assigned |
|---|---|---|
| Equities | Growth, earnings, discount rates | The return engine, and the source of most variability |
| Government bonds | Rates and inflation expectations | Income, and a diversifier in growth shocks |
| Credit | Rates plus default risk | Higher income, and equity-like in stress |
| Cash | The policy rate | Liquidity and optionality |
| Real assets | Inflation and supply | An inflation hedge with substantial variance |
The third row is the one that behaves differently from how it is often categorised. Corporate credit is frequently grouped with bonds and behaves substantially like equities in a downturn, which is when the grouping matters.
Correlation is an estimate
Diversification depends on components not moving together, and how much they move together is measured over a historical window. The measurement is an estimate with error, and it is not stable.
- Correlations rise in crises, which is exactly when the diversification was being relied upon.
- The equity-bond correlation has changed sign across decades, and the sign depends on whether inflation or growth is the dominant shock.
- A long window includes regimes that no longer apply; a short one is dominated by noise.
- Assets can share an exposure without a historical correlation showing it, until the exposure is tested.
The second item is the most consequential in practice. A portfolio built on the assumption that bonds rise when equities fall is relying on a relationship that held in some decades and reversed in others.
Why there is no general answer
An allocation is a function of constraints rather than a function of markets. Two people facing identical markets can hold entirely different portfolios without either being wrong.
- The horizon over which the money is needed, which determines how much variability is survivable.
- Whether contributions continue, since a portfolio being added to behaves differently from one being drawn down.
- Other income and its stability, since a secure income is itself a bond-like asset.
- The account structure, which changes what is efficient to hold where.
- The level of decline that would cause the plan to be abandoned, which is the binding constraint for most people.
The last is the one that is hardest to estimate and easiest to overestimate. An allocation that will be abandoned in a drawdown is worse than a more modest one that will be held, because abandonment happens at the bottom.
The allocation you can hold
Every allocation framework produces a recommendation, and the recommendation is only as good as the probability it is still in place after a bad year. That constraint is rarely modelled and frequently binding.
| Equity share | Approximate decline in a severe equity bear market | What it requires |
|---|---|---|
| 100% | Around half the portfolio | Holding through a loss of that size |
| 80% | Around 40 percent | The same, slightly reduced |
| 60% | Around 30 percent | Historically the most commonly held mix |
| 40% | Around 20 percent | A materially different experience |
The figures assume the fixed income portion holds up, which it did in several historical bear markets and did not in 2022. That is the correlation caveat made concrete rather than theoretical.
The question the table poses is not which row produces the best long-run return. It is which row would still be in place at the bottom, because an allocation abandoned in a decline realises the loss and misses the recovery.