Today’s WSJ has a good article on what is known as “financial engineering” at MIT. This program at MIT illustrates perfectly the fallacy of thinking higher math is a sure path to making money in the stock market — in this case by pricing financial options. A couple of years ago, Long Term Capital Management (LTCM) nearly brought the world financial system to its knees when its sophisticated trading program in derivatives went off the rails. This despite the fact that LTMC had on its team two Ignoble Prize winners in economics, plus some of the biggest names in mathematical economics.
Essentially what happened was the guys at LTCM factored out of their equations the possibility of very large but highly unlikely events. Unfortunately, highly unlikely events happen in real life. Thus, when something unlikely did happen — in this case, the East Asian financial panic — all their calculations went up in smoke. It’s like a program designed to beat the odds in Las Vegas: as long as you allow the person placing the bets to double his bet indefinitely, it is easy to demonstrate that he has a near perfect chance of winning against the house eventually, at whatever confidence interval you may care to choose. The flaw in the approach, of course, is that you are not allowed to double your bet indefinitely — on Wall Street as in Las Vegas.
So far I’ve been unable to append a copy of the WSJ article below, since it requires special access and I’ve lost my password. But let me note the main motive for these not-so-hot-shots in electrical engineering wanting to go to Wall Street in the first place: the firms that are foolish enough to hire them, are willing to pay them a higher salary than they can earn as real engineers! Clearly, someone has a vested interest in a certain point of view. (And just as clearly, if these young not-so-hot-shots really knew how to make money this way, they would be trading on their own account.)
Bottom line: human behavior is not mathematically predictable to the degree these people expect. It is a labile thing, which means that trying to use higher math to leverage something valuable out of it, is about as promising as trying to use a wet sponge as a lever.
Posted by lukelea at 07:06 AM