Don't bank on AI predicting the stock market anytime soon

There has been a lot of pontificating about the potential that generative AI has. From the industries and jobs it will upend to the efficiencies it will create, the power of AI can seem almost endless.

But for those on Wall Street, one question of AI reigns supreme.

Can it predict the markets?

The short answer: no.

At least, that’s according to Wall Street vet Marty Chavez.

But Chavez, who’s the vice chairman of Sixth Street and was a longtime partner at Goldman Sachs, didn’t leave it at that. He elaborated on why the financial markets don’t make for easy problems to solve for AIs.

“There’s a really profound technical reason why the AIs, I think, are not going to predict the stock market,” said Chavez while speaking at the Bloomberg Invest conference in New York. “The successes that we’ve seen in AI — and they are amazing; they are brilliant — all come from one profound realization, which is if you can take a large number of samples and you can dichotomize it.”

Chavez went on to explain how something like cats are much easier for AI to tackle.

“People love putting pictures of their cats on the internet and they say, ‘This is my cat.’ Now you’ve got this labeled data set. Well, we’ve gone to town on that. You can train up an AI. You give it a new image it hasn’t seen before, and it’ll tell you whether there’s a cat or not. And then it’ll tell you all the variations. It will tell you, ‘This cat is 10 years old, and it’s this breed, and it might be sick with this kind of disease.’ It’s just gotten better and better at that,” Chavez said.

But cats are not the same as the stock market (in case you were unaware). And while Siamese cats are going to remain Siamese cats for the foreseeable future, the same cannot be said for stocks.

“The concept of a cat is stable,” Chavez said.  “The stock market is notoriously not a stable distribution.”