What portfolio diversification actually means

Diversification is one of the most cited principles in investing and one of the most misunderstood. The standard summary, "don't put all your eggs in one basket", is technically correct but misses what actually matters.

The insight is not just "own many things." It's "own things that don't move together."

Correlation is the concept

Two assets are perfectly correlated if they always move in the same direction by the same amount. Two assets are uncorrelated if their movements are independent. Two assets are negatively correlated if they tend to move in opposite directions.

A portfolio of 50 stocks that are all large-cap U.S. technology companies is not well-diversified, even though it holds 50 positions. They're all exposed to the same underlying factors: U.S. economic conditions, interest rate sensitivity of growth stocks, regulatory risk for tech, consumer spending on software. When one falls, the others tend to fall for the same reasons.

A portfolio of 50 positions across U.S. equities, international equities, bonds, real estate investment trusts, and commodities is more diversified, not because it has more holdings, but because those asset classes have historically lower correlations with each other.

The math of diversification

Here's the unintuitive part: you don't need hundreds of positions to capture most of the benefit of diversification. Research consistently shows that a portfolio of 20-30 well-selected securities captures roughly 90% of the diversification benefit available from owning the full market.

Beyond 30 positions, the marginal benefit of adding another uncorrelated asset is small. The benefit of adding highly correlated assets is essentially zero regardless of how many you add.

This is why total market index funds work so well from a diversification standpoint. A fund like VTSAX or FSKAX holds 3,500+ stocks, but those holdings span small, mid, and large cap across every sector. The sector and factor diversification is what matters, not the raw number of holdings.

The diversification most investors are missing

Most retail investors are better diversified within U.S. equities than they realize (broad index funds handle this) and less diversified across asset classes than they think.

International diversification: U.S. stocks represent roughly 60% of global market cap. Holding only U.S. equities means missing 40% of the investable world. U.S. and international markets are correlated but not perfectly so, especially in emerging markets, which have different economic drivers.

Factor diversification: Within equities, value stocks, small-cap stocks, and momentum stocks have distinct risk/return profiles. A portfolio that's 100% large-cap growth (effectively what you get if you hold only the S&P 500) misses these factor premiums.

Duration diversification: In fixed income, short-duration and long-duration bonds respond differently to interest rate changes. Holding only short-duration bonds in a bond portfolio is a directional bet on rates.

What diversification doesn't protect against

Diversification reduces idiosyncratic risk, the risk specific to individual companies or sectors. It does not eliminate systematic risk, the risk that affects all assets simultaneously.

In March 2020, correlations across nearly all asset classes spiked toward 1.0 as liquidity-driven selling hit everything. A diversified portfolio still fell, just less than a concentrated one. In a severe market stress event, diversification reduces losses; it rarely eliminates them.

Long-term bonds and gold have historically provided the most consistent negative correlation to equities during crises, which is why they remain components of many diversified portfolios despite lower expected returns.

How Prismfolio shows correlation

The Prismfolio web app calculates the correlation matrix of your holdings and shows where you have concentration risk you might not be aware of. Many investors who believe they're diversified are holding large-cap U.S. growth through multiple different funds, effectively the same exposure, just with different names on the wrapper.

Seeing the actual correlation data, rather than the number of positions, gives you a more accurate picture of your actual risk profile.