How many stocks does a portfolio need to mimic the market reliably? It’s become fashionable to say not too many: maybe as few as 30. New research claims that’s delusional.
It’s work by Roni Israelov, a former principal at AQR Capital Management and now chief investment officer for NDVR. After examining risk and return in model portfolios of various sizes during the 25 years through 2019, Israelov and his colleague Yin Chen found that a fully diversified portfolio needs way more stocks than people think to avoid “unlucky outcomes.”
They were testing an increasingly popular notion in academic circles that says stuffing just a few dozen stocks into a fund is enough to ensure it behaves like the broader market. While that holds true a lot of the time, Israelov found, it falls apart at the margin -- too many bad outcomes befall investors with too few stocks. Depending on what’s in the basket, a less-blessed investor can suffer way worse returns or higher volatility.
“The traditional studies essentially just assume everything has the same expected return and the same volatility,” Israelov said in an interview. “That’s a large part of where the problem lies -- you know, stocks don’t have the same average returns.”
The way he sees it, the purpose of a diversified portfolio is to reduce the role of luck in performance. By his estimate, achieving that requires at least 200 stocks.
The reasons people are wrongly persuaded by the concentration argument are complicated but might be summarized as follows. Looked at from the top down, portfolios made up of 20 or 30 or 100 or 200 stocks all tend to bounce around in a similar way. Their “average volatility” is basically the same, leading analysts to conclude that the experience of owning them must be the same, too.
Israelov decided to test the idea in a cumbersome but particularly real-world fashion, by letting a computer pick out and build thousands of random stock portfolios over two decades and see what they actually do. The idea was to check if the top-down approach was masking wrinkles in the returns ordinary investors actually pocketed from portfolios of different sizes.
He found that it was. While average volatilies among different-sized combinations were similar, smaller portfolios in Israelov’s computer-generated research kept spitting out a weirdly high number of bad outcomes when compared with the market as a whole. The more stocks you added, the less often they did this.
The finding is a broadside against a popular view that argues that burdens including the cost of keeping track of a large number of holdings means managers should put a lid on how many stocks they own.
While diversification isn’t the goal of every investor, a look at exchange-traded funds shows they generally keep a fairly low level of holdings. According to data compiled by Bloomberg, the median size of actively managed ETFs is 35 stocks. For all ETFs it’s 104.