Buffett's Hedge-Fund Bet Was a Virtual Sure Thing
Ten years ago, Warren Buffett bet $500,000 that an S&P 500 index fund would beat the average of five fund-of-funds selected by the investment manager Ted Seides over the subsequent 10 years. The bet ends in December, but Seides has already conceded defeat.
People have been arguing for decades over whether active management produces superior post-fee returns. The answer depends on which active funds, over which time periods, using which evaluation criteria. Each side can cite high-quality studies and neither changes position. In bets, the parameters are set in advance, so fair-minded people accept the result.
I am disheartened that people are ignoring the outcome to claim support for their pre-exiting beliefs. The most common claim, originating from Buffett himself and echoed by Siedes and others, is that fund-of-funds lost due to high fees; actually, the funds lost before fees. 1 The problem wasn’t that high fees erased their performance advantage; the problem was that they underperformed to begin with. 2
Blaming fees is not only false but disingenuous. It takes large amounts of data over many decades to address whether active managers underperform after fees. Even if they underperform on average, there would only be about a 51 percent chance that a specific fund would underperform in a specific decade. Given Buffett’s goal to educate the public about investing, it would have been irresponsible to rely on a bet that was almost as likely to do harm as good. 3 And I’m confident that if the bet had gone the other way, neither Buffett nor the other commentators would have changed their minds about the value of active management, making it less than honest to use the win to insist other people change.
What does the bet actually prove? Siedes chose funds with a 0.97 correlation to the S&P 500. 4 You can’t get this by accident or by choosing investments related to S&P 500 stocks; you need to buy S&P 500 stocks. 5 You can tinker with the weights, or add a side-bet, but basically you’re running an S&P 500 index fund. 6 And even if tweaks return 10 percent per year, since they represent only 5 percent of risk, you beat the S&P 500 by only 0.5 percent. Seides paid a suicidal 3.25 percent for side-bets. Nevertheless, that was not the reason he lost.
Seides selected funds that went up and down only 56 percent as much as the S&P 500. 7 The only way for Seides to win the bet is if the S&P 500 went down so much over the 10 years -- more than 7.5 percent per year -- that avoiding 44 percent of the losses overcame the 3.25 percent fee.
The S&P 500 had negative 10-year returns only twice 8 since 1871 and never lost 4 percent per year, much less 7.5 percent. It’s strange that Seides conceded the bet when his funds were down 34 percent to the S&P 500. The S&P 500 has lost more than 34 percent four times, so he had more chance of winning at that point than he did when the bet commenced. 9
Therefore, the lesson of the bet is like the curious incident of the dog in the nighttime: 10 not who won the bet, but why no one offered a portfolio with a fighting chance of winning. 11 There are reasons a manager might decline. The bet is not a good test of hedge-fund value, and a manager might prefer to avoid misleading the public. Many managers shun publicity and few want investors who misunderstand their fund’s value proposition -- which has nothing to do with beating the S&P 500 over 10 years. There are compliance and contractual issues in publicizing performance. Still, you might expect one manager defend the honor of the industry.
Perhaps the key is that in 2007 there was excessive enthusiasm for hedge funds, and managers’ main problem was discouraging unrealistic expectations. Losing the bet would be embarrassing, but winning it would be worse, further misleading the public and attracting investors who could never be satisfied. 12
Buffett did win the bet. Not by finding a patsy, and not because the laws of probability made his bet a virtual sure thing. He stood up courageously when opinions of hedge funds were foolishly high and his willingness to back index funds in dramatic public fashion -- and the unwillingness of managers to accept his challenge -- probably saved many investors from costly investment errors. Unfortunately, he and his fellow travelers are undoing the good today by misrepresenting the lesson of the bet.
Professionals offering actionable insights on markets, the economy and monetary policy. Contributors may have a stake in the areas they write about.
Unfortunately, the fund-of-funds Seides selected are secret, so the only data we have comes from Buffett and ends in December 2016. At that point, the S&P 500 was up 85 percent, versus 22 percent for the funds. Buffett has access to all the data, and estimated fees accounted for 60 percent of the pre-fee return, suggesting that the funds returned about 55 percent before fees.
The next most common interpretation I read is the results prove Buffett’s investment talent. The problem with that is that his investment results were not relevant to the bet. Buffett didn’t bet that he could beat the fund-of-funds, he bet that the S&P 500 index would.
Then there are writers who claim the bet proves that active managers as a group achieve the index return before fees, and therefore underperform after fees. Here the problem is that the active managers significantly underperformed the index before fees. If it’s possible for a large group of managers to underperform, then it has to be possible for a large group to outperform.
And that even if it won, if it did, it would create the false impression that active managers almost always underperform after fees. That’s dangerous, because it’s easily refuted by observation, leading investors to distrust the sensible advice to stick to low-cost, tax-efficient index funds absent a strong reason to consider alternatives.
This is extraordinarily high. The correlation of the S&P 500 to other U.S. equity indexes is around 0.90, with international equity indices about 0.60. To be clear, it’s the total portfolio of all funds that had the 0.97 correlation, individual funds varied in their correlations.
Now this is not a bad strategy if you want to maximize your chance of beating the S&P 500. You invest in the index, and make some small tweaks that you hope will result in a small edge.
Technically, you need to take 95 percent of your variance (the correlation coefficient squared) from S&P 500 stocks.
In fact, S&P 500 index funds do not have a correlation of 1.00 with the index; 0.99 or 0.98 are common, and many index funds in indexes with less liquid components have correlations below 0.97.
The 0.97 correlation tells us that the 56 percent was not an accident. He must have planned for something close to this.
If you bought right before the Great Depression around 1928 or late enough in the internet bubble to catch the market crashes in 2000 and 2008. This statement uses proxies for the S&P 500 before 1957.
The S&P 500 has more sharp crashes than it has negative return 10-year periods. So being only 34 percent down with a year to go was a good result for Seides; instead of needing 10 years of minus 7.5 percent returns, he only needed 2017 to have a crash like 2008.
From the Sherlock Holmes story "The Silver Blaze":
Gregory (Scotland Yard detective): "Is there any other point to which you would wish to draw my attention?"
Holmes: "To the curious incident of the dog in the night-time."
Gregory: "The dog did nothing in the night-time."
Holmes: "That was the curious incident."
In fact, it’s trivially easy to win this bet: Just put your money in an S&P 500 index fund, plus make one bet with a very high probability of winning, such as writing one deep out-of-the money put. Or, there are plenty of hedge funds with reliable after-fee alpha; just pair one of these with an S&P 500 index fund to bring the beta up to one, and you have maybe an 80 percent chance of winning.
By 2017, the industry had been battered by all sorts of criticism, about half of it justified, so perhaps there would be a taker.
To contact the editor responsible for this story:
Max Berley at email@example.com