Paul Wilmott says Wall Street Relies on Too much Math.

[Subramaniam]

An article that blamess mathematician's lack of questioning the assumptions behind many of the models.. Entire article below.
http://www.bloomberg.com/apps/news?pid=20601087&sid=aD4y2TBuy4gQ

Wall Street Relies on ‘Too Much Mathematics,’ Paul Wilmott Says

By Thomas R. Keene and Shannon D. Harrington
Sept. 21 (Bloomberg) -- Financial markets have grown too dependent on mathematicians who use models to anticipate price moves and need to start injecting common sense into the equation, said Paul Wilmott, a London-based author and quantitative finance instructor.
“There is too much mathematics in this business,” Wilmott, author of “Paul Wilmott on Quantitative Finance,” said in a Bloomberg Radio interview. “I just want people to stop and think for once.”
Wilmott, based in London, has warned that so-called quants who use mathematics to forecast how markets will behave can overlook errors in the models, leading to flawed predictions.
“People don’t really question those assumptions enough,” said Wilmott, who founded the Diploma in Mathematical Finance at Oxford University, according to his Web site. “If the assumptions are wrong, then obviously the models and what follows can be wrong as well.”
Reliance on such models has led to a shift in the types of people hired by Wall Street, he said. “You go back 20 years, and people running finance, they were maybe history graduates,” Wilmott said. Now, much of the industry is run by mathematicians, he said. “A lot of mathematicians do not have that common sense that the old guard had,” he said.
To contact the reporter on this story: Thomas R. Keene in New York tkeene@bloomberg.net; Shannon D. Harrington in New York at sharrington6@bloomberg.net.
Last Updated: September 21, 2009 12:20 EDT

 

[joel_b]

I think 'common sense' can be programmed to a certain extent. It's just the evolution of software design. In my own experience you start with a 'perfect' trading model, and then as it evolves, the logic becomes fuzzier over time, and thus more intuitive. My trading models are not quant-level, but a human could look at the trade resuts and say 'hey that makes sense', and it no longer appears that a computer did it, which would have been the case in the 'perfect' model.

I think trading models should be aiming at producing profit, but the maximization of profit needs to have more intuitive constraints. Perhaps some of the information you may need has not yet been commoditized into a data feed, though. Single events need to be executed with vision of overall events in the marketplace, and outside of the marketplace. I'm sure this is already being done in many respects, and perhaps Wilmott is focusing more on the firms who are trying to implement more rudimentary (relatively) models, without enough real-world logic.

 

[Andy Nguyen]

Reliance on such models has led to a shift in the types of people hired by Wall Street, he said. “You go back 20 years, and people running finance, they were maybe history graduates,” Wilmott said. Now, much of the industry is run by mathematicians, he said. “A lot of mathematicians do not have that common sense that the old guard had,” he said.

Nice. I'm curious if the CQF program that Paul runs is producing the old guards or the new ones.

 

[Peter]

Nice. I'm curious if the CQF program that Paul runs is producing the old guards or the new ones.

That's the kind of pithy reply I wish I had written.

Well said, Andy.