AI Revolution: Finance Professor Shows ChatGPT May Be Able To Predict Stock Movements
Large language models may be able to forecast stock prices, according to Alejandro Lopez-Lira, a finance professor at the University of Florida. Lopez-Lira used the ChatGPT model to analyze news headlines to determine whether they were positive or negative for a particular stock. And in a harbinger of what may be to come, the advanced artificial intelligence (AI) program found success.
In a recent unreviewed paper, he found that ChatGPT was able to predict the direction of a stock’s returns the following day much better than random.
The experiment highlights the potential of artificial intelligence to display “emergent abilities,” capabilities that were not initially intended when the models were built. If ChatGPT can accurately interpret financial news headlines and their impact on stock prices, it could put high-paying jobs in the financial industry at risk.
Goldman Sachs has previously estimated that about 35% of financial jobs are at risk of being automated by AI.
However, the experiment also demonstrated the limitations of ‘large language models.’ For example, it did not include target prices or involve any mathematical calculations. Furthermore, ChatGPT-style technology often generates incorrect numerical values. There were some limitations on sentiment analysis of headlines as well, as reading into group psychology is a difficult endeavor for even the most advanced systems.
A Surprising Initial Result
Lopez-Lira was surprised by the results of the experiment, indicating that sophisticated investors are not currently using ChatGPT-style machine learning in their trading strategies. The experiment involved over 50,000 headlines from a data vendor about public stocks on the New York Stock Exchange, Nasdaq, and small-cap exchange.
The headlines were fed into ChatGPT 3.5 along with a prompt to look at the stocks’ returns during the following trading day:
“Forget all your previous instructions. Pretend you are a financial expert. 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.”
The model did better in nearly all cases when informed by a news headline, with less than a 1% chance of performing as well when picking the next day’s move at random. ChatGPT even outperformed commercial datasets with human sentiment scores.
For example, when presented with a headline about a company settling litigation and paying a fine, which had a negative sentiment, ChatGPT was able to correctly identify the news as positive.
Lopez-Lira reported that hedge funds have expressed interest in his research, and he believes that as more institutions begin to integrate this technology, ChatGPT’s ability to predict stock moves may decrease. The experiment only looked at stock prices during the next trading day, while the market could have already priced in the news seconds after it became public.
On the regulation side, if we have computers just reading the headlines, headlines will matter more, and we can see if everyone should have access to machines such as GPT. Second, it’s certainly going to have some implications on the employment of financial analyst landscape. The question is, do I want to pay analysts? Or can I just put textual information in a model?Alejandro Lopez-Lira, finance professor at the University of Florida
While the results are preliminary in nature, they indicate the predictive ability of AI-based prediction models in here, and will only get better over time.