Finance Professor Claims ChatGPT Can Predict Stock Performance

We have some questions.

AI Stock

ChatGPT is a real jack of all trades. The AI chatbot can draft cover letters, tinker with code, and even breathe life into a Furby.

And now, maybe OpenAI’s blockbuster tool could even make you rich — emphasis on the “maybe.” According to University of Florida finance professor Alejandro Lopez-Lira, large language models (LLMs) like it could be used to forecast stock prices. As CNBC reports, it’s an intriguing use of the tech that could potentially have some pretty substantial economic implications.

Just last month, Goldman Sachs estimated that around 35 percent of financial jobs are already on the chopping block due to AI.

But we’re still a far cry from a point where ChatGPT can turn you into a millionaire overnight.

Predictable Returns

In a yet-to-be-peer-reviewed paper, Lopez-Lira found that ChatGPT was able to predict the direction of stock prices well enough to make a difference in returns just by parsing financial headlines.

“The fact that ChatGPT is understanding information meant for humans almost guarantees if the market doesn’t respond perfectly, that there will be return predictability,” Lopez-Lira told CNBC.

But there are plenty of limitations to the professor’s experiment. For one, ChatGPT never actually analyzed target prices or did any calculations.

Worse yet, LLMs have a strong tendency to botch arithmetic.

And furthermore, Lopez-Lira predicted that if everybody in the finance industry would start analyzing headlines using LLMs like ChatGPT, their advantage could diminish completely given enough time, wiping out any potential gains.

“As more and more people use these type of tools, the markets are going to become more efficient, so you would expect return predictability to decline,” he told CNBC. “So my guess is, if I run this exercise, in the next five years, by the year five, there will be zero return predictability.”

More on ChatGPT: ChatGPT Is Consuming a Staggering Amount of Water

Share This Article

Go to Source