Warren Buffett once said that ‘diversification is protection against ignorance. It makes little sense if you know what you’re doing.’ I’m a Buffett fan, but this quote always bothered me because we know what we’re doing and we’re highly diversified.
The question of how many stocks you need to have a diversified portfolio seems less asked today than it was when we got started 20 years ago, perhaps because you can buy an exchange-traded fund (ETF) with thousands of stocks instantly, without a commission. In other words, why not have a diversified portfolio?
A simple way to think about how many stocks you ‘need’ is to look at the Dow Jones Industrial Average (DJIA), which only has 30 stocks. Since 1965, when the total return (price and dividend) data starts for the DJIA, the annual return through June was 10.49 percent.
To my surprise, the S&P 500 returned 10.50 percent over the same period. For this particular period, the more diversified portfolio did better, but just by 0.01 percent, which definitely falls into the ‘noise’ category.
I was even more surprised, though, to see that the volatility numbers were very similar too: the annualized standard deviation for the DJIA was 14.84 percent, and somewhat remarkably, 14.85 percent for the S&P 500.
In some ways, this quick and dirty analysis supports what some of the early academic papers say, which is that 30 stocks are probably the right number. Of course, the papers are a little different in the sense that they selected random groups of 30 stocks over and over, rather than have one portfolio determined by the committee at Dow Jones.
The original paper that suggested 30 stocks was simply looking at the reduction of total risk in the portfolio, as measured by standard deviation. That paper, by Fischer and Lorie, said that you could get 95 percent of the benefit of diversification with 30 stocks.
A more recent paper by Sur & Price looked at that paper and reaffirmed that about 30 stocks are correct if the only thing that you care about is the reduction in risk. Importantly, though, they looked at several other measures and concluded that a 30-stock portfolio only captures 86 percent of the diversification that’s possible.
And, if you’re comparing yourself to a benchmark, you should expect big differences between you and the benchmark each year. In three out of every 10 years, you should expect the difference between your portfolio and the benchmark to exceed six percent. In one out of 20 years, you expect the difference to be 12 percent.
If those differences are to the positive, then you probably won’t mind a whole lot, but if they are to the negative, then you won’t be happy and will have some catching up to do. And doubling the number of stocks to 60 doesn’t help much on this measure – you really need a few hundred to do the trick.
Although a series of 30-stock portfolios might do better on average, there were no doubt some that did vastly better, and some that did vastly worse. How can you be sure you’ll get one of the good ones?
A study by Crittenden and Wilcox looked at all of the stocks in the Russell 3000 (large, mid, and small-cap) and found that 39 percent of them lost money, and almost 19 percent of them lost more than 75 percent. A whopping 64 percent underperformed the index that they were part of and only 25 percent were responsible for the gains in the index.
Given the choice between the Dow & the S&P 500 since 1965 where the risk and return summary statistics are effectively the same, it seems obvious to me that you would choose the S&P 500 when it’s so easy to buy with one click. In 1965, that wasn’t possible, so owning 30 individual stocks made more sense.
I also don’t believe that you need fewer stocks to beat the market. Take the S&P 500 equal weight index – it has 500 stocks, but instead of owning them by their market cap like the regular index, it takes even positions in each stock.
An ETF following this strategy since 2003 beat the S&P 500 by 1.37 percent per year over that period. There have been periods of major underperformance, but there were enough periods with major outperformance to offset the bad ones.
From a portfolio perspective, we’ve answered the question of how many stocks to own by effectively taking ‘all of them.’ The three US benchmarks that we use for the US have 1,500 stocks, and we have exposure to all of those, plus a few more since some of the funds we use have the Russell 1000 and 2000 as their universe.
We take the view that it isn’t the number of stocks that matters, but how you weight them.