While artificial intelligence has many uses within financial services, it’s not well-suited to predicting the stock market, because the data is too dynamic, unstable, and broad.  

Artificial intelligence is amazing, but don’t expect it to predict the stock market for you, according to Marty Chavez, Sixth Street vice chairman and former chief information officer at Goldman Sachs.  

At its core, AI is simply “statistical pattern matching,” Chavez said during Bloomberg’s recent Invest conference.  

“It’s extremely powerful and it’s interesting, but I do not see AI achieving what some would call the ‘Holy Grail,’” he said. “Everybody wants to know, ‘What’s the S&P going to be in six months?’ I can’t tell you, and neither can AI.” 

The strengths of AI don’t translate to such a broad and unstable use case, he said.  

“The successes that we’ve seen in AI — and they are amazing, they are brilliant — all come from one profound realization,” he said, which is that AI is great at dichotomizing data. It can identify cats based on gobs of labeled training data, because the concept of a cat is stable. The stock market, by contrast, doesn’t have the same consistency, so that kind of pattern matching doesn’t translate well to Wall Street, he said: “The stock market is notoriously not a stable distribution.” 

While broad, generalized stock market predictions may never come to pass, artificial intelligence can provide tradable signals about very specific outcomes based on specific signals – for example, analyzing the tenor of Fed statements to predict market moves.