Arnold Kling  

The Quants

Adverse Selection by Law... The Henderson Club...

The Quants, by Scott Patterson, is the latest indulgence in my taste for financial crisis porn. It reminds me a lot of Roger Lowenstein's When Genius Failed. Lowenstein's book, which narrates the rise and fall of the infamous hedge fund Long Term Capital Management, is highly readable, but ultimately it fails to explain why the firm failed. Did it hold unsound positions, or did it suffer from an excess of temporary losses on what were basically good bets (sometimes this is known as the gambler's ruin problem)?

Patterson seems tempted to lump all of the quants together and to blame the gambler's ruin problem. Early in the book, he offers an anecdote in which Ed Thorpe, an early quant, sets out to use card counting to win at blackjack. A key is to avoid the gambler's ruin problem by reducing the size of your bets as you run low on capital.

Still, the gambler's ruin story has some problems. First, not all quant firms followed the same strategy. Some firms' strategies actually succeeded to during the market meltdown. Second, if it was a gambler's ruin problem, then the losses in the quant positions should have turned around by now. It does not appear that they have.

For me, the most interesting take-away from the book was just how fragile the financial system has become. For example, Patterson says that the stock market crash of October 1987 very nearly caused the Chicago Mercantile Exchange to go under. I wonder how many of the people who think that putting derivatives trading on an exchange is a cure-all are aware of that episode.

My favorite chapter in the book is called "The August factor." It describes a near-implosion in quant firms that took place in August of 2007. On August 10, Patterson suggests, the financial system might have collapsed, except for two decisions. First, Cliff Asness of the hedge fund AQR decided to fight the trends in the market by taking on more risk. Second, Goldman Sachs added liquidity to its stressed quant funds. Patterson writes,

If the selling had continued--which likely would have happened if Goldman Sachs hadn't bailed out its GEO fund--the results could have been catastophic, not only for the quants but for everyday investors...Just as the implosion of the mortgage market triggered a cascading meltdown of quant funds, the losses by imploded quant funds could have bled into other asset classes, a crazed rush to zero that could put the entire financial system in peril.

The picture I get is of a financial trading system that has grown far too large and far too complex for anyone to understand. I came away from the book with even less confidence in the ability of regulators to ensure its safety. By the same token, it reinforces my view that we need to make the system easy to fix, because we will never be able to make it impossible to break.

COMMENTS (4 to date)
bjk writes:

"the losses by imploded quant funds could have bled into other asset classes, a crazed rush to zero that could put the entire financial system in peril."

In 2007 the S&P was over 1400 most of the year . . . "A crazed rush to zero" was not exactly in the cards at the time. Maybe a rush to 1200, but that looks pretty good from here.

Walt French writes:
The picture I get is of a financial trading system that has grown far too large and far too complex for anyone to understand.
Bachelier's Theory of Speculation, treating random walks to value options, was published in 1900; Einstein re-discovered the math of it a few years later; Ito's stochastic calculus was recognized in 1945; Black & Scholes used Ito's stochastic calculus in their 1970's option pricing model. Many, many more have toiled in these fields, some garnering Nobel recognition.

If it's true (and I'd guess it bit of a stretch) that one individual getting out of the right side of the bed one morning determined whether there was a meltdown or not, it's not clear what a model could possibly tell us. This math is complex, but we have a long history of using stochastic math in finance; even more unpredictably inter-connected systems such as the internet are studied, designed and managed with this near-chaos in mind. Perhaps others will update this post to show how very smart, focused, but mortal economists and systems types have made more recent advances than my little history above, which stops almost 40 years ago.

I would say that the statement, “no, we do understand this much better than it seems,” is consistent with this blog's apparent purpose in celebrating and applying good thinking to important issues of national policy. Unfortunately, policy is often a shouting match between players whose primary focus is the zero-sum payoffs where they have strong interests, and who allow others, many of which have very small soapboxes on which to stand, to promote the positive-sum aspects.

PS for @bjk-- I'm sure you're aware that during the August “Quant Meltdown,” relevant market indexes barely budged, a point that Lo has made effectively. What I still don't know -- I haven't read the book, and doubt that it could actually describe what might have happened -- is whether massive losses at involved asset managers such as BGI could have led to refusal to trade with them; that was essentially what torpedoed Lehman a few months later. And we all know how well that worked.

EclectEcon writes:

I wonder whether the meltdown that was averted in August, 2007, would have been any worse than the one experienced in autumn, 2008. It might actually have been better for the economy to have had the shake-out occur over a year earlier.

I'm still puzzled why more people didn't bet the shake-out would occur.

TKM writes:

If it were possible to mathematically factor in greed then we would have our answers.

Comments for this entry have been closed
Return to top