Cliff Asness's Staff Built Him A Robot Warren Buffett

Publish date:

Do you want to invest like Warren Buffett? Sure you do. You know who will tell you how? Strangely, some guys at AQR:*

[W]e create a portfolio that tracks Buffett’s market exposure and active stock-selection themes, leveraged to the same active risk as Berkshire. We find that this systematic Buffett-style portfolio performs comparably to Berkshire Hathaway.

They acknowledge that Robo-Buffett doesn't incur transaction costs that flesh-Buffett does (because R.-B. is as of yet just a simulation) but, that aside, "comparably" is an understatement:

Whee! Go Robo-Buffett! Who, intriguingly, looks a lot like ... AQR:

Collectively, these four standard [Fama-French, momentum etc. - ed.] factors do not explain much of Buffett’s alpha .... Since Buffett’s alpha cannot be explained by standard factors studied by academics, his success has to date been considered a sign of unique skill or as “alpha”.

Our innovation is to also control for the Betting Against Beta (BAB) factor of Frazzini and Pedersen (2010) as well as the quality factor (QMJ, “quality minus junk”) of Asness, Frazzini, and Pedersen (2012b). A loading on the BAB factor reflects a tendency to buy safe (i.e., low-beta) stocks while shying away from risky (i.e., high-beta) stocks. Similarly, a loading on the quality QMJ factor reflects a tendency to buy high-quality companies, that is, companies that are profitable, growing, and paying out dividends (see Asness, Frazzini, and Pedersen (2012b) for details).

We see that Berkshire loads significantly on the BAB and QMJ factors, reflecting that Buffett likes to buy safe, high-quality stocks. Controlling for these factors drives the alpha of Berkshire’s public stock portfolio down to a statistically insignificant annualized 0.1%, meaning that these factors almost completely explain the performance of Buffett’s public portfolio. Thus, a significant part of the secret behind Buffett’s success is the fact that he buys safe, high-quality, value stocks.

From this they build a "Buffett-style"** strategy consisting of basically picking stocks whose previous performance corresponds most closely to those loadings, rebalancing every month. And it does well - both in the sense of "outperforms the market" and in the sense of "correlates reasonably well with Buffett's actual performance."***

I am usually a bit unmoved by this sort of exercise - you can fit anyone's record to some set of factors chosen after the fact; saying "Warren Buffett's investing looks like a robot choosing safe, high-quality, value stocks" is true because you built the robot to look like Buffett.**** But I enjoyed this one. It's nice to see that this particular robot works - both because, y'know, yay robots, and because it's some evidence that Buffett is doing what he says he's doing. Your image of Warren Buffett could be either "kindly grandpa clutching his dog-eared copy of Graham & Dodd" or "savvy media manipulator whose folksy value investor front masks aggressive derivatives dealings and whose performance is mostly driven by halo effects and the ability to drive a hard bargain with embattled issuers who need to rent that halo." We've occasionally indulged in the latter view here.

But I guess you should feel good about how closely Warren tracks his robot cousin's performance? The robot, after all, is a bloodless (simulated) value investor, whose (pretend) performance is never boosted by favorably priced private investments in bank preferred, and who can't drive up a company's stock just by (simulating) investing in it. All that he has, he gained from (fake) stock-picking.

And, amusingly, he's outperforming Buffett since the mid-'90s. The authors don't really get into why - given transaction costs, etc., you can't take this outperformance too seriously - but they hint at an answer when they say "Berkshire initially focused on small firms ... and only became biased towards large stocks in the later time period. Hence, Berkshire’s diminishing returns could also be related to capacity constraints." It almost looks like the advantages of Berkshire's size and fame - the ability to get great terms out of companies and have stocks go up after you buy - are dominated by the disadvantages - the difficulty of putting money to work without moving the market. Maybe you really would be better off with the robot.

Frazzini, Kabiller, and Pedersen - Buffett's Alpha [Yale, pdf]

* Via this article in Pensions & Investing which unfortunately I cannot entirely endorse; it is all "oooooooooooh huge shocker Warren Buffett uses leverage OMGOMG!" and, I mean, are you freaked out that an insurance company has liabilities? If so, I submit to you that you should not be running a pension and/or investment. The AQR paper is totally sensible on this and actually gets into Berkshire's sources of leverage (principally AAA-ish debt and insurance float - cheap!) and determines that Buffett's access to abnormally cheap leverage is not the main factor in his outperformance.


*** Actually it has a 75% correlation to Berkshire's public stock portfolio, but only 47% to Berkshire's stock price. Meh.

**** I like how they congratulate him for anticipating their paper based on him:

In summary, if one had applied leverage to a portfolio of safe, high-quality, value stocks consistently over this time period, then one would have achieved a remarkable return, as did Buffett. Of course, he started doing it half a century before we wrote this paper!

Like - that's not a coincidence. If he'd applied leverage to a portfolio of risky, low-quality growth stocks fifty years ago, you wouldn't have written the paper.