Briefs

January 28, 2002. U.S. News & World Report writes about a VirtualHamilton neural network, which beat the market by nearly 3 percentage points a year between 1930 and 1997. The program, named after William Hamilton, former editor of the Wall Street Journal in 1900s, was created by a team from New York University and Yale, which analyzed his writings and used it to create the program.

The article discusses several other AI-based investment programs, and the possibility of reverse-engineering the expertise of a mutual fund manager by tracking his trades and feeding them into a neural network.

What will happen if computers eventually get so good at the prediction game that all investors are made of silicon rather than carbon? Then the computers, says Ray Kurzweil, "will be trying to outpredict each other."

See the article at

http://www.usnews.com/usnews/issue/020128/biztech/28ai.htm