Expected Returns Ch. 4: The Same Bet, Bought Three Times
阅读中文版Diversification is usually measured by counting holdings. Ilmanen measures it by counting premia — and by that measure most portfolios are far less diversified than their owners believe.
🔊 Listen to Article (Chinese Audio)
Expected Returns Ch. 4: The Same Bet, Bought Three Times
Investment Background
This is the book's most diagnostically valuable chapter for an individual investor.
Its claim fits in one sentence:
Diversification should be measured by the number of risk premia you bear, not the number of products you hold.
By the first measure most portfolios are diversified. By the second, many are one bet purchased repeatedly.
The Wall Street Translation
A Concrete Diagnosis
Consider a real, apparently well-diversified retirement portfolio:
| Holding | Apparent category |
|---|---|
| US large-cap fund | Equity |
| US small-cap fund | Equity |
| International developed fund | Equity |
| Emerging markets fund | Equity |
| High-yield bond fund | Bonds |
| REIT fund | Real estate |
| Preferred share fund | Fixed income |
| Infrastructure fund | Alternatives |
Eight holdings, four asset classes. It looks well diversified.
Now count by premium:
| Holding | Dominant premium |
|---|---|
| US large-cap | Equity |
| US small-cap | Equity (plus a little size premium) |
| International developed | Equity |
| Emerging markets | Equity (plus country and currency risk) |
| High-yield bonds | Credit premium — materializes alongside equity in recessions |
| REITs | Equity plus leverage plus term |
| Preferred shares | Credit plus term |
| Infrastructure | Equity plus term |
Eight holdings. And in reality one dominant bet — economic growth — purchased eight times, plus some term risk.
In a recession all eight fall together.
This is the individual-portfolio version of the argument in Chapter 2 of When Genius Failed: LTCM believed it held a hundred independent bets and actually held one bet written a hundred times.
That book was about a hedge fund. This chapter is about your retirement account.
Why the Error Is So Common
Three reasons, none involving anyone's fault.
One, asset class labels classify by legal form, not by risk source.
The label "bonds" covers both Treasury bills (nearly riskless) and high-yield corporate debt (essentially a form of equity risk). They share a name and behave almost oppositely.
Two, sales are organized by product category.
An advisor presenting a portfolio "diversified across eight asset classes" looks more professional than one holding two funds. And the misalignment in Where Are the Customers' Yachts? explains why complex presentations sell better.
Three, correlations genuinely are lower in calm periods.
The subtlest point. Those eight holdings really do correlate below one in normal years. Their correlations converge only under stress — which is to say, at the only moment that matters.
Chapter 2 of When Genius Failed supplies the mechanism: correlation is not an intrinsic property of an asset but a function of the market's state.
Where Real Diversification Comes From
Counting by premium, genuine diversification means holding things sensitive to different bad times.
And here is a sobering finding: very little is genuinely independent of equity risk.
| Asset | Behavior in an equity crash |
|---|---|
| High-quality government bonds | Usually rise (flight to safety) |
| Cash / Treasury bills | Essentially unmoved |
| Inflation-linked bonds | Mild; depends whether the shock is inflationary or growth-driven |
| Gold | Unreliable — sometimes works, sometimes not |
| Most other things | Fall with equities |
The first three rows are the list.
The practical implication matters: you do not need eight holdings to diversify. You need two or three genuinely different premium sources.
Which is exactly the structure of a low-cost global index fund plus high-quality bonds plus cash:
- Global equity index → the equity premium across every country and sector (eliminating company- and country-specific risk while retaining the premium you actually want)
- High-quality intermediate bonds → moderate term premium, usually moving opposite equities in recessions
- Cash → zero duration, zero credit, deliberately forgoing the liquidity premium
Three holdings, three different premia, and their painful moments do not fully coincide.
That is far more diversified than the eight-holding portfolio above.
A Necessary Qualification
The framework has a real boundary that must be stated.
One, an inflation shock hits stocks and bonds together.
The table above assumes a growth shock (recession). In an inflation shock, bonds will not protect you — 2022 is the example, and Chapter 2 explained the mechanism (a shared discount rate).
So the complete list needs a fourth item: inflation-linked bonds or other real assets, as the hedge against inflation shocks.
Two, simplification should not be pushed to an extreme.
"You only need three holdings" is a statement about premia, not about optimization. Tax treatment, account types, and rebalancing mechanics can all justify holding more specific funds.
Three, this analysis is qualitative. Precisely quantifying how many independent premia a portfolio holds requires institutional data and tools. For an individual, this chapter's value is diagnostic direction, not precise computation.
Executable Trading Rules
-
Run a premium audit: list every holding and write the dominant premium source beside it, not the asset class label. The chapter's most important line. Then count how many distinct sources you actually have. Most people find the answer is one or two, not the five to eight they assumed.
-
Group by "what happens in a recession." Concretely: sort your holdings into those likely to fall in a recession and those unlikely to. The size of the second group is your real defensive capacity.
-
Be wary of products whose names imply diversification while the risk source is identical. Especially anything with "income" in the name — these usually package a credit premium or a volatility-selling premium (Chapter 3).
-
Do not solve a diversification problem by adding holdings. A ninth equity-type holding improves nothing. Diversification comes from different premia, not different products.
-
Confirm you have something addressing an inflation shock. The fourth item on the complete list. Inflation-linked bonds are the most direct tool, because they hedge that specific risk by construction.
Relevance to a Retirement Portfolio
This chapter matters more to retirees than to accumulators, for a specific reason.
A retiree's core risk is being forced to sell during a decline (sequence-of-returns risk).
And this chapter establishes something uncomfortable: if everything in your portfolio falls at the same moment, then at that moment you have nothing to sell.
That is the mechanical justification for the cash buffer, and it is far more precise than "cash helps you sleep":
The cash buffer's value lies not in its yield (which is low) but in its being the thing that can still be sold when everything else has fallen.
In this chapter's language: it is the one holding that does not share a painful moment with your dominant premium.
Which is why "one to three years of essential spending in cash" is among our core recommendations across this site.
It is not conservatism. It is the purest form of premium diversification.
And our standard allocation — global equity index, high-quality intermediate bonds, cash, plus inflation-linked bonds — audits under this chapter as four distinct premium sources whose painful moments fall in different economic scenarios.
An eight-fund portfolio usually cannot claim that.
Chapter 5 handles the natural next question: since these premia are visible, can you rotate between them?