Wall Street: Man versus Automaton


A small biotech company thinks it can make trading rooms better by eliminating people. In the midst of the biggest financial crisis of this generation, a tiny biotech company called Gene Network Sciences based in Cambridge, Massachusetts, thinks it can make Wall Street smarter. How? By getting rid of the man.

Their idea: Take the supercomputers Gene Network Sciences already uses to help Pfizer and Biogen Idec invent drugs, and use them to help hedge funds trade in stocks, bonds, and other assets. “Computers and data are smarter than people. “, explains Colin Hill, the physicist who founded GNS and who will be the president of its new subsidiary Fina Technologies.

“We believe that the economy and the financial system are governed by complex networks, just as genes control cells and neurons control the brain. says Hill. “And we believe that by using artificial intelligence, we can start to determine this circuit from the raw data. »

Computers have already replaced a good number of people working on trading desks. Now Fina wants to replace the mathematical programming prodigies with more computers. Perhaps a tenth of fund managers are “quants” – short for quantitative trader – who run computer algorithms to buy and sell stocks with such frequency that these transactions often represent a third or even a half of the trading volume recorded on the New York Stock Exchange.

When it works, you can win fortunes. James Simon founded the quantitative company Renaissance Technologies in 1982, which is currently worth an estimated $7.4 billion. David E. Shaw, a bioinformatician founded the quantitative firm DE Shaw in 1988, currently worth some $2.7 billion. But sometimes a lot of these funds lose money at the same time – they all use the same math, and therefore make the same bad bets. Fina thinks she can avoid the problem by decommissioning the armies of mathematicians and their market forecasting equations to make way for a computer-powered spectacle.

Instead of trying to figure out how to predict the market using a human brain, Fina’s system would be able to copy all available data into a computer system that would instantly store the billions of possible prediction patterns. Imagine being able to put together the puzzle of a photo in all the many different ways and at the same time decline all the probabilities of assembly.
The idea comes from systems, from the network, or from biology. Genes and proteins interconnect in a complex network. By materializing these connections, companies hope to invent better drugs. Merck, in particular, uses, at the heart of its approach, a technology similar to that used by GNS.

This computational approach to biology attracted investors who were, in some cases, quants. Two years ago, Hill was having a drink in an upscale Manhattan bar with an investor, a GNS board member named Thomas Paul, who later became chief investment officer at Fortress Investment Group. For years they joked about using GNS technology for stock trading. That night, for some reason, the idea finally took shape.

Paul graduated from MIT (Massachusetts Institute of Technology), class of 1993, with a master’s and master’s degree in engineering and computer science, and, like many of his peers, he went to Wall Street, working first at Goldman Sachs then at Deutsche Bank before managing an 800 million dollar fund at Fortress.

He was prepared for the bizarre world of quants, he said, alluding to the MIT blackjack team – a different version of the team depicted in the movie 21, in which a group of college students realized that with the backing of investors, they could consistently break Las Vegas records in card counting.

“Blackjack was good practice. “says Paul. In card counting, as with most quantitative funds, the idea is to use a mathematical system to do just a little better than luck. With his blackjack team, they won it hands down about 57% of the time. For a fairly large chunk of quantitative funds, 51% would be fine.

But finance is “quite different,” he says, because blackjack has immutable rules and measurable odds.

“If you create a [trading] model that assumes the world is as it was from 2001 to 2007, that model would likely fail in 2009,” Paul explains. “People are going to get smarter realizing that quantitative solution modeling needs scrutiny and skepticism. I’m sure this past year has destroyed a lot of pattern based trading strategies. Those who have done it well are those who can adapt to a rapidly changing world. »

Control? Skepticism? How to acquire them by putting people out of the process and letting computers do the forecasting? Because GNS software can also make other predictions that should be more certain than that stock prices will rise or fall. In drug development, this could yield blood tests that can predict early on whether an experimental medicine will fail. A hedge fund could experiment with predicting volume or price ratios, and thus abandon a faulty algorithm before losing tons of money.

To find out if the biotech software could really work on Wall Street, Paul tried it out on a historical gas price model. An ordinary computational model was able to predict whether gas prices will rise or fall about 40% of the time. The GNS model was correct 79% of the time.

He and Hill have recruited another quant, a former classmate of Paul’s at MIT named Josh Holden, who will serve as Fina’s manager. They signed a contract with Reed Elsevier to invest in the company. Kevin Brown, a manager at Reed Elsevier’s investment fund, says he’s delighted, particularly because the technology can also be applied to online advertising targeting particular groups of customers and to detect internet fraud.

But just because they were able to predict gas prices with startling accuracy doesn’t mean this biotech company’s Wall Street adventure is going to turn out well. “It’s something I’ve heard from hundreds of quants over the years. says Andrew Lo, director of MIT’s Financial Engineering Laboratory. “We are going to see the successes, not the failures. “

Even the most legitimate and serious quants can be mistaken in thinking they have found the fountain of youth. Lo said. “This is a problem that is endemic to all quantitative analysts. One counterbalance to Lo’s healthy skepticism, he says, is the involvement of Paul Thomas. “He’s a very seasoned veteran. He is also a very good trader. »

Still, some wonder if having the automatons that this entails is really a good thing. “The market has become a casino. wrote Sydney M. Williams, one of the partners of Monness, Crespi, Hardt & Co., a research and trading firm in New York, in a note that made the rounds on Wall Street. It is “a place where a growing number of actors ignore the fundamental principles of each society. »

Leave a Reply

Your email address will not be published. Required fields are marked *