Adding holdings to a portfolio reduces risk, but the reduction slows sharply after a modest number of positions. The reason lies in what kind of risk is being removed.
Two kinds of risk behave differently
Some risk is specific to a single company: a failed product, a lost contract, an accounting problem. These events are largely unrelated to one another across different firms.
Other risk is shared by everything at once, such as interest rates, recessions or a sudden repricing of credit. Every holding in the portfolio is exposed to it simultaneously.
Diversification works on the first kind and has no effect on the second. Once company-specific risk is spread thin, what is left is the market's own variability.
Correlation decides how much is actually removed
Combining two holdings reduces risk only to the extent that they do not move together. If they respond identically to the same forces, holding both changes very little.
Adding a second bank to a portfolio of banks adds a name without adding an exposure. Adding an asset driven by different forces does far more with a smaller allocation.
This is why counting positions is a poor measure of diversification. What matters is how many independent sources of return the portfolio is actually exposed to.
The benefit curve flattens quickly
The first few holdings remove a large share of company-specific risk, because each one dilutes the influence of every other. The improvement from the next few is smaller.
By the time a portfolio holds a few dozen genuinely different companies, most of what could be diversified away has been. Further names change the outcome only marginally.
Beyond that point, extra positions add administrative work and dilute conviction without a meaningful reduction in volatility. The cost of complexity begins to exceed the statistical benefit.
Correlations rise when they are least welcome
Relationships between assets are not fixed. In ordinary conditions many holdings drift somewhat independently, which is exactly what a diversified portfolio is built to exploit.
During severe stress, selling pressure tends to hit everything that can be sold, and previously unrelated assets fall together. Measured correlation rises precisely when diversification is being counted on.
The effect is temporary but it is real, and it explains why portfolios that looked well spread can still fall hard in a genuine crisis.
What still helps once the curve flattens
Once company-specific risk is handled, the remaining choices concern exposure to different economies, currencies, asset classes and time horizons rather than to more individual companies.
These are broader decisions with their own trade-offs, including currency risk and differing tax treatment. They shift the shape of the whole portfolio rather than smoothing individual names.
Understanding where the flattening occurs prevents two opposite mistakes: holding too few positions to be diversified at all, and holding hundreds under the impression that more is automatically safer.