Every investor learns the same rule early: don't put all your eggs in one basket. It's good advice, and the maths backs it up. Spreading capital across enough uncorrelated positions genuinely lowers a portfolio's risk without giving up expected return. The mistake is assuming that if some diversification is good, more must always be better. It isn't. Past a certain point, each additional holding does little to reduce risk and dilutes the thinking that went into the portfolio in the first place.

The first serious attempt to measure this dates back to 1968, when Evans and Archer published their landmark study in the Journal of Finance. Using US equity data, they found that a portfolio of ten to fifteen randomly selected stocks eliminated roughly 80% of company-specific risk. Each additional stock contributed steadily less. Later research by Reilly and Brown put the useful range slightly higher, finding that twelve to eighteen holdings captured around 90% of the maximum diversification benefit available. Even Benjamin Graham, writing decades earlier, settled on a range of ten to thirty.

The exact number moves with market conditions. Statman's 1987 work argued that rising correlations between stocks meant investors needed closer to thirty holdings to replicate the broader market's risk profile, and today's more interconnected global markets likely push that number higher still. But the curve's shape hasn't changed. Diversification benefit rises quickly at first, then flattens. Somewhere past that flattening point, a portfolio stops getting meaningfully safer and starts getting harder to manage well.

The same dynamic shows up one level up, in how many managers or funds an investor holds rather than how many stocks. Add enough managers to a portfolio and their individual bets start to cancel each other out, whether or not any single manager intended it. The technical term for the extreme version of this is closet indexing: a fund that markets itself as active but holds a portfolio so similar to its benchmark that it can't meaningfully outperform it. Researchers Cremers and Petajisto developed a measure called Active Share to capture this. Non-index funds with an Active Share below 60% are generally classified as closet indexers.

An investor doesn't need to hold a closet index fund to suffer the same fate. Hire enough managers with genuinely different processes, and their aggregate holdings can still average out to something close to the market. The fees, however, don't average down to an index fund's. This is diversification's quiet cost: not higher risk, but a portfolio that pays for conviction it no longer actually holds.

This is where the current shift gets harder to execute well. Hamilton Lane's 2026 Global Private Wealth Survey found diversification, alongside performance, among the top reasons private wealth clients invest in private markets. Used as a reason to build broader private markets exposure, that's sound reasoning. But when diversification becomes a reason to spread capital across an ever-larger number of managers and vehicles, the trade-offs look different from those in public markets, and can be less forgiving.

A public portfolio with too many stocks becomes an expensive proxy for an index, which is an annoyance you can unwind by selling. A private portfolio with too many managers becomes something harder to fix: capital locked up across a wide spread of vehicles, each individually reasonable, that collectively add operational complexity without adding much genuine risk reduction, because many of the underlying deals were competing for the same auctions in the first place. A family office correcting for public-market concentration by adding private managers faster than it can properly diligence them risks solving one concentration problem by creating a liquidity and oversight problem in its place.

The pressure to add another manager is not easing. Churchill Asset Management's research on the US middle market, published this September, found that funds raised by emerging managers, including first-time and newly spun-out general partners, accounted for more than 70% of new private equity fund launches in 2025, and that 38% of limited partners expect new manager formation to keep outpacing consolidation over the next three to five years. As year-end portfolio reviews approach, the pressure to add managers can intensify. That matters particularly in a market where a growing number of new and relatively unproven managers are competing for allocations.

Meanwhile, capital raised by first-time funds in North America fell from roughly $54bn across North America in 2022 to about $7bn across just 48 funds in 2025. That fund count was the lowest since 2010, while the average interval between first and final close stretched to sixteen months. Fewer institutions are willing to underwrite an unproven manager, but the number of new funds in the market keeps climbing regardless. That combination of more choices, longer diligence and less institutional appetite to do the legwork creates exactly the environment in which a family office can end up adding its next manager because the search process ran out of time before year-end, rather than because the manager earned the allocation.

None of this argues for concentration for its own sake. A ten-stock portfolio built on ten independent, well-researched convictions is a different thing entirely from a forty-manager portfolio assembled to look thorough on a pitch deck. The distinction that matters is whether each new position is actually doing work, reducing a specific risk the portfolio doesn't already have covered, or whether it's there because more positions felt safer than fewer.

The practical test is simple to apply and uncomfortable to answer honestly: for every holding in a portfolio, can an investor say what it's there to do that nothing else already does? If the answer is no for a meaningful share of the portfolio, the issue usually isn't too little diversification. It's too little conviction, dressed up as caution.

The rush away from concentration risk is justified, and the data on US equity weighting supports it. But diversification was never the end goal; it was always a means of protecting conviction, not replacing it. The same test that applies to a ten-stock portfolio applies to a family office adding its fortieth manager this year: is this position doing work, or is it there because more felt safer than fewer?