Let me just start out by saying that, in my mind, “The Quants: How a New Breed of Math Whizzes Conquered Wall Street and Nearly Destroyed It,” by Scott Patterson (New York: Crown Business, 2010) is a terrific book. It has people and places, it has history, it has theory and it is very readable. I could not stop reading it once I got started. I highly recommend it to anyone.
Most recent books on the financial crisis focus on the events of September 2008, Fannie Mae (FNM), Freddie Mac (FRE), Lehman Brothers and AIG (AIG). This book focuses on August 2007 and the meltdown of hedge funds run by Quants. The terrain is not as familiar as that pertaining to the former date.
I will come back to this part of the book later. First I want to highlight some of the points made in the book that can get lost in the excitement of the story Patterson tells.
The initial point I would like to make is this: finance is all about information, and Information is just 0s and 1s. Continuing, you can say that 0s and 1s are just about computers. This is the essence of the story that resides in this book.
It is interesting to me that the beginning of the story Patterson tells is how math/physics whiz Ed Thorp, the Godfather of the Quants, started out on the path to “Quant-dom.” Thorp, as a new member of the MIT staff, took some of his early work on how to predict outcomes of roulette wheels to a well-known member of the MIT faculty named Claude Shannon.
Shannon is known as one of the founding fathers of Information Theory, a theory that has to do with the transmittal of information and the ability to receive and discern the message conveyed in the information transmitted. Information, however, is a technical matter, completely devoid of meaning and content: it is purely statistical and encodable. Information can be put into 0s and 1s.
This becomes the essence of Quant behavior: the collection of data connected with the effort to discern what message or messages are contained within the data. Thorp’s initial financial interest was in the pricing of warrants, specifically those connected with convertible bonds. His attempts to price warrants centered on the random movement of the prices of the bonds which led him into something called Brownian motion. Thorp then discovered the work of Louis Bachelier and his idea that prices followed a random walk. This idea was then connected with the “law of large numbers” which allowed Thorp to translate the data on pricing behavior into a bell curve which could produce another parameter called the standard deviation: the volatility of the prices.
The discovery of volatility was the key for Thorp, who then developed a system for pricing warrants that was based on the relative variability of different issues of convertible securities. He, thus, developed a quantitative model based on the two summary messages that were produced by a large sample of data on the prices of convertible bonds. Note that the quality of the company or the management is irrelevant -- the information needed is just 0s and 1s.
Thorp applied his model to the “real world” and found that it was very successful. He was one of the early pioneers of the hedge fund and he was enormously successful. He got out of the business before it crashed in the 2000s.
This set things going. There were massive amounts of data on stock market prices, bond prices, interest rates and other pieces of financial information going back for years. This was ripe material for Ph. D. candidates in the 1960s and beyond! But the basic approach still held: what message can be culled from all the data available and how could the information contained in this message be used to make money? Lots and lots of money.
It all had to do with information, nothing else; no valuation, no analysis of management, nothing else. Cash flows could be cut up this way and that because they were nothing more than information, just 0s and 1s. What is your problem? I can design cash flows that can be used to solve your problem: Interest Rate Swaps, Collateralized Debt Obligations, IOs and Pos and Credit Default Swaps. You name it.
Now let me fast forward to the quant fund group known as Renaissance Technologies and its star fund Medallion. This whole group was created by Jim Simons and it is “the most successful hedge fund in history.” It made it through the 2007-2008 collapse relatively unscathed.
What kind of team did Simons pull together to staff his funds? Cryptographers and people trained in speech recognition; in essence, people trained in Information Theory. They were trained to detect hidden messages in seemingly random strings of code. Sound familiar?
My next point is that the models used by the Quants are just that, models. They are not the real world! The Dean of the Efficient Markets Hypothesis (EMH), Gene Fama, is quoted in the book as saying that the EMH is just a hypothesis and thus it is incomplete and fallible. And this is the crucial thing about the models that the Quants use: they are all incomplete and fallible!
Why is this true? Because we, as human beings, never have complete information when building a model; we always have to work with incomplete information. This is what creates uncertainty -- we don’t know what the outcomes of our decisions will be. Our models are imperfect.
However, everyone uses models to make predictions. The use of models is not just a characteristic of being a Quant. Humans use all sorts models. “Leave the building if it is on fire,” and, “slow and steady win the race,” are just two non-quantitative models. Quants use formal models, mathematical models. Non-quants use less formal models, but they are sill models. So, using models is not a way to distinguish a Quant from a non-quant. But, more formal models may help us to make better predictions.
Uncertainty is not a given characteristic or property of a situation. Uncertainty arises because of incomplete information. Some situations give rise to probabilities that are relatively stable, but these are situations more consistent with the physical sciences and card games. Some uncertain situations, however, are more uncertain than others, especially those dealing with human behavior. And that is what this book is about.
From the start, quantitative finance was basically about the two pieces of information that could be used to describe the behavior of prices: the mean and the standard deviation. The latter was derived from the law of large numbers, which was connected with the implicit assumption that the distribution of security prices approximated a bell curve.
The problem is that these two measures do not necessarily describe how security prices will behave. They come from a model that is incomplete and fallible. Given that these models are incomplete and fallible they should not be considered as true or as “The Truth.” The models must be continually tested, modified and changed. This is apparently what the Medallion fund does, and this is what some of the more skeptical Quant leaders like Paul Wilmott and Nassim Nicholas Taleb do.
And this is the reason why the Quants won’t go away. The human quest is to continually build models that help people make better decisions or solve more difficult problems. It has been shown over the past forty years that Quant models can help generate millions, or billions of dollars in earnings. The models were not perfect. Patterson ends up his book describing how Quants, as we speak, are working to improve their models. The process will go on.
On the down side, Patterson has described how the Quants have played more and more of a role in the recent experiences of financial disorder. The severity of the October 1987 stock market crash has been connected with Portfolio Insurance, a Quant innovation. The 1998 meltdown of Long Term Capital Management -- another Quant short-circuit. The August 2007 financial crash is the most recent Quant-led crisis.
If the Quants are not going to go away, can they at least be regulated? They have not been regulated up to this point. The August 2007 experience is notable because of the absence of the Federal Reserve System. This meltdown came in what is known as the “shadow banking system” and not the true banking system. The Fed really didn’t seem to know what was going on.
The first catastrophe came when the Bear Stearns hedge funds were instructed to file for bankruptcy on July 30, 2007. The melt-down started in earnest on Monday August 6. On Tuesday, the Fed said it had decided to leave short-term rates alone. It was only on the morning of Friday, August 17, that the Fed lowered interest rates as the fear over “toxic assets” spread.
My guess is that regulation is always going to run behind the Quants. Right now, it is my belief that the Big Banks -- JPMorgan (JPM), Goldman Sachs (GS) and others -- are far ahead of the regulators, and this just a short time after the bailouts of 2008. Because of the nature of information, this will continue in the future with little or no letup. (See my posts on financial regulation in the Information Age here and here.)
Patterson closes his book with “Here come the Quants.”
But, the Quants are going to have to learn something themselves. Their efforts are not made in a black box that is isolated from the rest of the world. They are dealing in arbitrage, and in the world of arbitrage there is something called “the Law of One Price.” Success brings emulation. Emulation means that there will be more and more funds chasing the same arbitrage situations. That means arbitrage situations will go away or become less and less profitable over time.
In order to retain money or attract new money in such a situation, funds take on more and more leverage or more and more risk in an attempt to sustain performance. When or where the system will crack is unknown. That is what incomplete information and uncertainty are all about. In August 2007 the crack came in the area of subprime mortgages. Next time it will be somewhere else. But as we saw in 2007, the de-leveraging is catastrophic. It is not bell curve.
And this is what Patterson’s “The Quants” is all about!