Diversification is often treated as an unalloyed good. It's not. The more I learn, the more I appreciate that Warren Buffett said something as perfect as can be about the subject: "Diversification is protection against ignorance."
Most investors should own diversified, low-cost funds
Diversification is a volatility-control strategy that requires little knowledge on your part. As long as 1) two assets aren't perfectly correlated and 2) the expected return on one of the assets isn't too low, it follows as a matter of math that owning a combination of the two can be expected – not guaranteed – to provide a better volatility-adjusted pay-off than owning only one.
If you know little, diversification is a no-brainer. In fact, you want to diversify as much as possible. Moreover, you want to do it as cheaply as possible, as the benefits of diversification don't require expertise. There is a trade-off. If you know something – say, you can actually identify undervalued stocks – then at some point diversification hurts you by diluting your edge. An extreme example would be someone privy to news that's certain to send a stock's price rocketing. It would be crazy for him not to put a huge chunk of his wealth into the stock, assuming he's not breaking the law. The more you know, the more diversification hurts you.
Most investors understand that they should diversify a lot. However, some hurt themselves by behaving inconsistently: They diversify a lot while implicitly behaving as if they know a lot. A big subset of this group is investors who own lots of different expensive funds. Owning one expensive fund is a high-confidence bet on the manager. Well-done studies estimate that the percentage of truly skilled mutual fund managers is in the low single digits.
It would be strange if your process for assessing managers turns up lots and lots of skilled ones, because there aren't many in the first place. If you see skilled managers everywhere, chances are your process is broken or not discriminating enough. It would be even stranger if you bet on many of them. Doing so dooms you to getting index-fund-like results while paying hefty fees. It makes little sense to pay 1% or more of assets on an aggregate portfolio with hundreds of positions and market-like behaviour.
An exception is if you assemble a portfolio of extremely concentrated fund managers. Owning 10 funds with 10 stocks each put together will look like a moderately concentrated fund manager. This is a model some successful endowments, hedge funds, and mutual funds use.
Most investors should own diversified, low-cost funds. Those who believe they know something should concentrate to the extent that they're confident in their own abilities. A big danger is that humans are overconfident; many will concentrate when they should be diversified.
A young investor with lots of room to make mistakes and a passion for investing should consider forming a portfolio of "play money" with a handful of his best ideas. Over time, he can learn whether he knows what he's doing and either size up or down his bets. An advantage of a concentrated "skill" portfolio is it becomes quickly apparent if an investor knows what he's doing. This can prevent a lot of heartache down the road. An older investor near or in retirement just beginning to learn about investing cannot take the risk of self-exploration. He should stick to low-cost, highly diversified funds.
