Diversification is about correlation, not about the number of holdings

This works through diversification in the order the parts actually depend on each other.

The mechanism

Combining assets whose returns are not perfectly correlated produces a portfolio that fluctuates less than the weighted average of its parts. The reduction comes entirely from the parts not moving in step, so the degree of independence is what does the work. Adding a holding that behaves identically to something you already own adds nothing except an extra line and possibly an extra charge.

This is why the count of holdings is a poor measure of how diversified a portfolio is.

The most common false diversification

Holding several funds from different providers that all track large companies in the same market is a single exposure held four times. Overlap is easy to check by looking at each fund's largest holdings, which are usually published monthly. People frequently discover the same handful of very large companies occupying a substantial share of every fund they own.

Concentration at the index level is a real phenomenon, and tracking a market-weighted index means accepting whatever concentration that market has.

The dimensions that matter

Diversification across companies reduces the effect of any single business failing. Across sectors, it reduces the effect of one industry facing a shock; across countries, of one economy or government doing so. Across asset types — equities, bonds, cash, property — it addresses the risk that one entire class performs badly for years.

Across currencies, it changes what happens to your purchasing power when exchange rates move, which is a separate question again.

Correlations are not constant

Assets that normally behave differently have repeatedly moved together during severe market falls, as investors sell whatever they can. This means diversification often provides less protection precisely when it would be most useful, which is a well-documented and awkward fact. It does not make diversification pointless; it means the expected benefit should be estimated modestly rather than assumed.

Holding some genuinely uncorrelated cash is the version of this that has consistently worked, at the cost of a lower expected return.

The limit of the benefit

The reduction in fluctuation from adding holdings diminishes quickly, and most of the available benefit within a market arrives well before you own hundreds of companies. Beyond that point, additional diversification within the same market changes very little. What remains is the risk affecting the whole market, which cannot be diversified away by owning more of that market.

Reducing that residual requires a different asset class rather than a longer list.

Diversification is not free

It guarantees you will always own something performing badly, which is uncomfortable and is the point. It also caps the upside, since you will never be concentrated in whatever performed best.

Practically, trading breadth for the chance of a spectacular outcome is a choice, and it should be a deliberate one rather than an accident of holding what was familiar. What mix is appropriate for any individual depends on circumstances and horizon, and is a matter for regulated advice rather than a general rule.

The takeaway

Check what your funds actually hold. Four funds owning the same twenty companies is one bet with four fee lines.

Write the number down before you decide. It usually decides for you.