Getting Buffett's number

October 17, 2017 8:56 PM

Warren Buffett, Oracle or Orangutan? A statistical analysis of the Sage of Omaha's investing aptitude.
posted by storybored (18 comments total)

Not that the underlying idea is bad but

The orang-utan will not be flipping by himself, just as Berkshire Hathaway does not compete in a vacuum. So, our trials will include not one orang-utan, but a whole rainforest full of them. Five hundred to be exact, one for each company in the Standard & Poor’s 500 Index (S&P 500) – an industry standard for “the market” (x = 500).

And the parts of the text under that make it sound like the author thinks the S&P 500 is a set of investors and not a broad index of equity prices.
posted by dismas at 9:10 PM on October 17, 2017 [6 favorites]

How does a paper that doesn't know what an average is win a statistical writing award? Per the Zacks stock screener, 233 of the S&P 500 members have beaten the index since the start of the year, while 288 have outperformed over the past 5 days.

PS: There are only 499 companies on the S&P 500 that perform independently of Berkshire Hathaway, as 10 seconds of googling will verify.
posted by Homeboy Trouble at 9:18 PM on October 17, 2017 [6 favorites]

The Drunkard's Walk, 2008, Leonard Mlodinow, [GoodReads]
posted by lazycomputerkids at 9:19 PM on October 17, 2017

It's an introduction to a statistical concept, maybe not a great finance piece though. Berkshire has a different risk profile from the S&P 500, extremely different tax treatment, and of course it's going to perform differently.
posted by miyabo at 9:37 PM on October 17, 2017 [3 favorites]

Orangutans are great apes, not monkeys.
posted by lazycomputerkids at 9:48 PM on October 17, 2017 [7 favorites]

Sometimes an article deserves the WELL ACTUALLY treatment.
posted by clawsoon at 10:33 PM on October 17, 2017 [1 favorite]

I would highly recommend The Myth of the Rational Market, which is a sort of intellectual history about ideas around the stock market.

I think believing in the drunkard's walk appeals to a lot of lefty types* because you get to say these rich people aren't that smart**. The problem is believing that also requires believing markets are 100% efficient, at least in the narrow sense of "as good as humanly possible." Fox (author of the above book) comes down fairly on the side of the efficient market not existing, and claims that the academic consensus is that it indeed does not.

Briefly illustrated: There was a bubble in the '90s and it was obvious there was a bubble in the '90s. (i.e., Palm, a subsidiary of Cisco, should not have been valued as more than the entire company.) It was hard to make money knowing there was a bubble partly because you needed to be patient. Furthermore, money managers have a systematic incentive not to be contrarian: You are less likely to be fired for follow-the-herd behavior that doesn't work out than for an equivalently risky contrarian gamble that didn't work out. This biases the whole system.

Hard to make more, but not impossible. Buffett for example had a lot of patience and an independent stream of cash to invest. He actually got out of the market in '66^† because he couldn't find investments that he thought made sense. How many investors have the nerve, insight and capability to do that? And among the consistent winners the people who can follow this sort of strategy are over-represented. It's not surprised to find ten orangutans who are doing really well, but when they're all at the same zoo?

If you don't trust me that the whole book is worthwhile here's a Cosma Shalizi review that focuses on different aspects than I did but gives you a flavor. And also has the added benefit of being a chance to read something by Cosma Shalizi.



* This could be properly understood to mean "This is how I feel personally, so I feel I'm 100% justified in projecting it on other people."
** Spoiler alert: They aren't that smart regardless.
† I'm not an investor type and am typing from memory so if I have the details of this anecdote not quite right I don't care.
posted by mark k at 11:17 PM on October 17, 2017 [6 favorites]

There are only 499 companies on the S&P 500 that perform independently of Berkshire Hathaway, as 10 seconds of googling will verify.

The word "independently" needs to be used very carefully outside the context of a controlled experiment. Given that all of the companies in the S&P 500 belong to the same economic system, and that major shocks tend to affect all members, their individual time series are correlated, and as such they are not statistically independent. As a random system, even though there are 500 elements in the S&P 500, there are not 500 effectively independent elements, or "degrees of freedom." The actual number is likely much smaller.

When comparing hedge funds and investment banks, the correlation of activity is even more pronounced. Investors are (a) all working from the same data, and (b) are all observing one another's behavior. Attempts to quantify the number of effective degrees of freedom among investors have suggested something alarming: Even when firms are "independent" in the legal sense, they are highly interdependent in the statistical sense. In effect, the entire investment sphere behaves like a random system with at most two or three independent contributors. This increases the odds of booms and busts considerably, as extreme events in either direction are much more likely (put another way, the tails of the distribution are "heavy").

We would probably be better off if the world of finance behaved a bit more like orangutans flipping coins. Since orangutans don't conspire and don't display herd behavior, we'd see a lot less boom-bust.
posted by belarius at 7:11 AM on October 18, 2017 [2 favorites]

Coming to this a day late, but I think the linked piece does a good job setting out its thesis and then following through. The question isn't whether Buffett has been more successful than the (rest of the) S&P 500, it's whether he's been more successful than chance. The conclusion, to spoil the piece, is: yes. By chance alone, .00961 companies out of 500 would be expected to do as well as Buffett has done.

That might be considered a stupid question to even ask, but there are many, many, many people who truly seem to believe that all stock-related activity is pure chance, nothing more. This piece counters that pretty effectively, I think.

I don't think the writer is attempting to say as much about Buffett as they are about the "random walk" hypothesis.
posted by pwinn at 7:43 AM on October 18, 2017 [1 favorite]

It was hard to make money knowing there was a bubble partly because you needed to be patient.

That's not the big problem so much as that actively betting against a stock, etc. isn't free. E.g., if you wanted to "bet against" Palm back in the day, there were various ways to do it, but they all carried costs. And if, for instance, your option expired out of the money, because the bubble had not yet popped, the premium was just gone. (An option with a longer period costs more.) If, worse, you had actually shorted the stock, your losses would be as great as your miscalculation on stock price. So, for instance, Ackman has taken at least paper and probably real losses on his shorts against Herbalife. "The market can stay irrational much longer than you can stay solvent," is what is often said in these circumstances.

I don't really think that "random walk" and "efficient market" theories are ineluctably linked. You can believe that the market even only somewhat approaches weak efficiency and yet also believe that humans are generally not good at figuring out how to exploit those inefficiencies, except to the extent that they or confederates have created them. This is consistent with the fact that there are approaching zero professional asset managers that consistently outperform an appropriate index, or even the S&P 500, net of fees. As far as I know, the Medallion Fund still mints money--they're the one hedge fund I'd give money to, if they weren't closed to outside investors (and if I were a qualified investor, which I am not)--but the same people have had bad stumbles with their follow-on funds.
posted by praemunire at 8:37 AM on October 18, 2017 [1 favorite]

This article confirmed two facts that weren't in dispute:

Warren Buffett has done better than anyone else on the market.
Warren Buffett has done better than orangutans flipping coins would be expected to be.

This article did not address the actual question at hand:

Given X equally skilled investors, one of them will rise to the top as the "best" due to random factors. Are Warren Buffett's results statistically better than this expected "best" such that we can confidently state that he is of above-average skill?
posted by explosion at 9:17 AM on October 18, 2017

the Medallion Fund still mints money--they're the one hedge fund I'd give money to

A popular theory on the Internet, which I can't vouch for, says that Medallion is using complex deals to essentially steal profits from Renaissance's other funds. Medallion is only open to Renaissance's own employees who are all under amazingly restrictive NDAs and are all filthy rich at this point so I don't find it hard to believe.
posted by miyabo at 10:20 AM on October 18, 2017

(i.e., Palm, a subsidiary of Cisco, should not have been valued as more than the entire company.)

The Palm Company made handheld computer devices, the likes of which have transformed TV, telephone, and everything else in the past 15 years. That mobile internet technology and compact computer design wasn't up to par with modern times or that Palm wasn't as clever as Samsung and Apple isn't a sign of a bubble.
posted by The_Vegetables at 1:09 PM on October 18, 2017

Also I'm not sure I agree that "A Random Walk" is really all that applicable to Warren Buffett, as compared to your average stock broker. Your average stock broker doesn't get to talk to the President of the US, lobby tax law, sue competitors, or structure the companies they are invested in. Nor does he or she get to hire independent auditors to review the books of potential corporate purchases like a conglomerate like Berkshire Hathaway does.
posted by The_Vegetables at 1:20 PM on October 18, 2017 [4 favorites]

A popular theory on the Internet, which I can't vouch for, says that Medallion is using complex deals to essentially steal profits from Renaissance's other funds.

This simply can't account for Medallion's performance during the majority of its existence, inasmuch as it was the only fund of what was then called Renaissance Technologies up until ~2007.

Certain of Renaissance's tax strategies which might be called, ah, aggressive, came to light a couple of years ago, but they were matters of leverage and taxation, not underlying alpha.

I would think the particular current set of Renaissance employees would have more to lose from ripping off the large institutional clients in RIFF and RIEF than to gain at this particular point in time, but I guess nothing's impossible.
posted by praemunire at 7:15 PM on October 18, 2017

The Palm Company made handheld computer devices, the likes of which have transformed TV, telephone, and everything else in the past 15 years. That mobile internet technology and compact computer design wasn't up to par with modern times or that Palm wasn't as clever as Samsung and Apple isn't a sign of a bubble.

The sign of the bubble has nothing to do with those things and everything to do with the valuation and stock price. And unlike most ridiculous valuations during the bubble this was objectively, undeniably obvious at the time because if its weird nature as an almost-wholly owned subsidiary of 3Com (not Cisco, sorry.)

Here's the passage (from Myth of the Rational Market:)

Palm was a wholly owned subsidiary [of 3Com . . . ] In March 2000, 3Com took Palm public, selling 5 percent of its stake onto the market and announcing plans to spin off the rest of Palm to 3Com, at a ratio of just over 1.5 Palm shares for every 3Com share, before the end of the year. The 3Com shares thus represented 95 percent ownership of Palm plus full owernship of 3Com's other [profitable] business [ . . . ] They should have traded for well more than 1.5 times the price of Palm stock. Instead they sold for less than Palm stock did.

"It took five months to correct," recalled 3Com CEO Eric Benhamou.

Put another way, you could buy one share of Palm stock directly on the market, or spend less and get more (1.43 shares, specifically) indirectly by owning 3Com and yet people were still buying Palm directly instead. This makes no sense and there's no way to make it make sense.

In case you're wondering Palm would indeed drop 90% over the next year.
posted by mark k at 8:02 PM on October 18, 2017 [1 favorite]

An undefined future spinoff date represents a financial risk. This paper is a very wonky, but in financial terms it posits a good reason (based on available data and derived(synthetic) data) why the difference in price existed, based on pricing the uncertainty of the split date and the limitations of trading on Palm stock.

A solution to the Palm—3Com spin-off puzzles”

FTA:
In the after-hours of May 8, 3Com announced that it would spin-off its shares of Palm on July 27, 2000, well
before its original estimated spinoff date of December 2000.14 The following day, 3Com’s share price rose
more than 10%, from $43.69 to $48.25. Palm’s share price fell almost 10%, from $32.25 on May 8 to $29.13
at the close on May 9. How can we explain the opposite movement of Palm and 3Com prices when 3Com
owns 95% of Palm? The May 8 announcement sharply changed the distribution of the spin-off time, and the
uncertainty disappeared.

If you want to call that a bubble, that's fine but it's really a bubble that 3Com themselves was able to create - and very possibly profit from if they were slick. To prevent this type of uncertainty and price disparity, companies should be required to set a fixed distribution date. Lending and trading fees explains the rest of the price differential, but I'm not well versed enough to say anything about that.
posted by The_Vegetables at 8:03 AM on October 19, 2017

Regardless of how one has reached a certain number of years of overperformance, once this is achieved, the investment choices that someone like Buffett makes, influence the actual performance of the companies he choses or ommits. And as The_Vegetables points out, with immense wealth comes both access to increased knowledge of the market and the opportunity to influence key parameters in the market.

And mark_k, the efficient market hypothesis, which would indeed imply that Buffet could be no better than an Urangotang, is anathema to lefty types (but by no means only lefty types) who, consider it yet another dogma of neoliberal market fundamentalism, with no empirical evidence to back it up. Lefties would probably argue about the mechanism of Buffett's success, but not its randomness. As you point out the paper is in essence an implicit argument against EMH.
posted by talos at 5:23 AM on October 20, 2017