AI claims to find hidden stock market signals, but only in a lab

Researchers built an AI system that claims to find hidden patterns in stock market data. This system uses large language models to generate trading signals, but only within a controlled testing environment.

Quantitative traders constantly search for reliable signals in noisy financial markets. This paper shows a new way to use AI to search for these signals. It tries to avoid the common problem of AI creating complex formulas that look good on paper but fail in the real world.