Asset Allocation: Correlation Does the Work
Diversification is not owning many things. It is owning things that do not move together — which is measurable.
Asset allocation is the decision about what kinds of things you hold and in what proportion. It gets called the most important decision in a portfolio, which is roughly true and usually explained badly: the reason it matters is not that stocks and bonds are different words, it is that they move for different reasons, and that is a number you can look up.
The mechanism
Combine two holdings and the portfolio's volatility is not the average of theirs. It depends on their correlation. Two assets that move in perfect lockstep (correlation 1.0) give you exactly the weighted average — no benefit at all. Anything below 1.0 gives a portfolio that swings less than the average of its parts, and the lower the correlation, the larger that gap.
That gap is the entire product of diversification. It is free in the sense that it does not cost expected return, which is what makes it unusual in finance.
What the correlations actually are
| Ticker | Volatility | Beta vs SPY | Correlation vs SPY |
|---|---|---|---|
| SPYUS large cap | 13.0% | 1.00 | 1.000 |
| IWMUS small cap | 18.7% | 1.18 | 0.823 |
| EFADeveloped international | 15.9% | 0.97 | 0.791 |
| EEMEmerging markets | 25.3% | 1.50 | 0.769 |
| TLTLong Treasuries | 9.3% | 0.19 | 0.263 |
| LQDInvestment-grade credit | 5.4% | 0.20 | 0.483 |
| GLDGold | 29.3% | 0.75 | 0.332 |
| VNQReal estate | 13.7% | 0.31 | 0.294 |
Correlation is measured against SPY from daily returns over the past year.
Two things stand out. Equity sleeves that sound diversifying — small caps, international, emerging — tend to sit high against US large cap, because they are all equities and equities respond to the same global risk appetite. The genuinely low readings come from different asset classes: 0.263 for long Treasuries, 0.332 for gold.
This is why "I hold thirty stocks" is weaker diversification than it sounds, and why an allocation decision is mostly a decision about how much of the portfolio is exposed to one factor.
Correlations move, and they move at the worst time
The uncomfortable part: correlations are not constants. In a genuine panic, most risk assets fall together as investors sell whatever can be sold. Diversification across equity types thins out exactly when you were relying on it.
Holdings whose low correlation is structural — different cash flows, different buyers, different drivers — hold up better than holdings that merely happened to diverge over the measured window. The correlation pages show a one-year and a three-year figure side by side for this reason: when they agree, the relationship is more likely to be real.
Rebalancing is what makes the allocation exist
An allocation only means something if it is maintained. Left alone, whatever rose fastest becomes the largest holding, and a portfolio that started at 60/40 spends years drifting toward the risk profile you originally rejected. Rebalancing is the mechanism that keeps the stated allocation and the actual one from parting company.
Checking your own
Two questions worth answering with data rather than intuition. First: are any two of your holdings really one position? Anything correlating above about 0.9 is a single bet wearing two names. Second: does anything you own have a genuinely low correlation to the rest — and if not, the portfolio is one factor deep regardless of how many tickers it contains. Both are answerable in the correlation finder.
Figures on this page are measured from daily returns over the year to 2026-09-22, across 5,262 liquid US securities, and are rebuilt nightly. This is reference material, not investment advice.
Related: rebalancing · Index funds & ETFs · Back to all articles