ChatGPT may be able to predict stock movements – Here’s how

As artificial intelligence grows increasingly smarter with the use of bigger computers and better datasets, they may display “emergent abilities” – skills that weren’t originally planned when they were built.

“Our analysis reveals that ChatGPT sentiment scores exhibit a statistically significant predictive power on daily stock market returns. By utilizing news headline data and the generated sentiment scores, we find a strong correlation between the ChatGPT evaluation and the subsequent daily returns of the stocks in our sample,” an yet-to-be-peer-reviewed research paper explains. 

Professor Alejandro Lopez-Lira and his fellow researcher Yuehua Tang used over 50,000 news headlines about public stocks on the New York Stock Exchange, Nasdaq, and a small-cap exchange. The starting date was October 2022 – to ensure that these headlines had surfaced after ChatGPT’s current learning cutoff date. 

ALSO READ: ChatGPT faces demands from Italian data protection agency before resuming services

The headlines were then fed to GPT 3.5 with instructions that the AI platform pretend to be a ‘financial expert’ and react.

“You are a financial expert with stock recommendation experience. Answer “YES” if good news, “NO” if bad news, or “UNKNOWN” if uncertain in the first line. Then elaborate with one short and concise sentence on the next line,” ChatGPT was told. 

When compared against the stocks’ returns the next day, researchers found that ChatGPT had performed well in nearly all cases after being informed by a news headline. He found a less than 1% chance the model would do just as well if picking the next day’s move at random. ChatGPT also beat commercial datasets with human sentiment scores. 

“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.

ALSO READ: How ChaosGPT plans to destroy humanity

While the researchers concluded that basic models such as GPT-1, GPT-2, and BERT could not accurately forecast returns, advanced language models would have greater success. 

“Our results suggest that incorporating advanced language models into the investment decision-making process can yield more accurate predictions and enhance the performance of quantitative trading strategies,” the paper’s abstract explains.