A merchant works on the floor of the New York Stock Exchange.
Jason Decrow
Alejandro Lopez-Lira, a finance professor at the University of Florida, suggests that large language models may be useful when forecasting stock prices.
He used ChatGPT to parse scandal headlines for whether they’re good or bad for a stock, and found that ChatGPT’s ability to predict the direction of the next day’s proffers were much better than random, he said in a recent unreviewed paper.
The experiment strikes at the heart of the guaranty around state-of-the-art artificial intelligence: With bigger computers and better datasets — like those powering ChatGPT — these AI designs may display “emergent abilities,” or capabilities that weren’t originally planned when they were built.
If ChatGPT can demonstration the emergent ability to understand headlines from financial news and how they might impact stock prices, it could could put high-paying occupations in the financial industry at risk. About 35% of financial jobs are at risk of being automated by AI, Goldman Sachs estimated in a Walk 26 note.
“The fact that ChatGPT is understanding information meant for humans almost guarantees if the market doesn’t be affected perfectly, that there will be return predictability,” said Lopez-Lira.
But the specifics of the experiment also show how far professed “large language models” are from being able to do many finance tasks.
For example, the experiment didn’t register target prices, or have the model do any math at all. In fact, ChatGPT-style technology often makes numbers up, as Microsoft well-educated in a public demo earlier this year. Sentiment analysis of headlines is also well understood as a trading design, with proprietary datasets already in existence.
Lopez-Lira said he was surprised by the results, adding they suggest that cosmopolitan investors aren’t using ChatGPT-style machine learning in their trading strategies yet.
“On the regulation side, if we have computers well-deserved reading the headlines, headlines will matter more, and we can see if everyone should have access to machines such as GPT,” conjectured Lopez-Lira. “Second, it’s certainly going to have some implications on the employment of financial analyst landscape. The question is, do I thirst for to pay analysts? Or can I just put textual information in a model?”
How the experiment worked
In the experiment, Lopez-Lira and his partner Yuehua Tang looked at once again 50,000 headlines from a data vendor about public stocks on the New York Stock Exchange, Nasdaq, and a small-cap reciprocity. They started in October 2022 — after the data cutoff date for ChatGPT, meaning that the engine hadn’t received or used those headlines in training.
Then, they fed the headlines into ChatGPT 3.5 along with the pursuing prompt:
“Forget all your previous instructions. Pretend you are a financial expert. You are a financial expert with stock say-so experience. Answer “YES” if good news, “NO” if bad news, or “UNKNOWN” if uncertain in the first line. Then elaborate with one abbreviated and concise sentence on the next line.”
Then they looked at the stocks’ return during the following trading day.
At the last, Lopez-Lira found that the model did better in nearly all cases when informed by a news headline. Specifically, he build a less than 1% chance the model would do as well picking the next day’s move at random, versus when it was cultivated by a news headline.
ChatGPT also beat commercial datasets with human sentiment scores. One example in the study showed a headline about a company settling litigation and paying a fine, which had a negative sentiment, but the ChatGPT retort correctly reasoned it was actually good news, according to the researchers.
Lopez-Lira told CNBC that hedge supplies had reached out to him to learn more about his research. He also said it wouldn’t surprise him if ChatGPT’s ability to predict ancestry moves decreased in the coming months as institutions started integrating this technology.
That’s because the experiment barely looked at stock prices during the next trading day, while most people would expect the market could force already priced the news in seconds after it became public.
“As more and more people use these type of dresses, the markets are going to become more efficient, so you would expect return predictability to decline,” Lopez-Lira said. “So my theory is, if I run this exercise, in the next five years, by the year five, there will be zero return predictability.”