Neural Networking

Publish date:
Updated on

We’ve got enough math nerd neurons firing in our heads to actually enjoy reading stories about new ways of doing financial modeling. What we don’t especially enjoy is articles about new models that rely on vague biological metaphors and don’t tell us anything about how the models actually work. In fact, we have a word for these articles—useless.
Today’s Reuters story on a “neural network” computer models being developed by Germany's Commerzbank AG is a good example of useless. It lets Commerzbank's head of alternative investment strategies, Mehraj Mattoo, rave about a computer model that “tries to take everything into account and learn as it goes along.” Wow. Sounds exciting. How’s it do that? Oh, what, never mind. Pay no attention to the little man behind the curtain.
It’s understandable that money managers wouldn’t want to disclose much about how their financial model works. But should journalists really just print the managers bragging about the models? Would they let a politician get away with saying he had a perfect plan to keep Social Security solvent but couldn’t disclose it? A football team manager announce a new strategy that he wouldn’t discuss but he was sure would win?
Eh. Maybe we're being too hard on the Reuters story. The neural network financial modeling thing does sound cool.
Neural networks invade hedge fund industry [Reuters]