The idea that ChatGPT could be able to predict stock movements is becoming more and more prevalent in FinTech. Predicting stock prices and stock movements are, of course, two different things. While it is nearly impossible to exactly predict stock prices, it is definitely possible to predict stock movements since stock movements are based on patterns. The ability to predict stock movement was first developed in the 1960s, by Edward O. Thorp, an American mathematician and hedge fund manager who applied quantitative techniques to the stock market. It was, however, Jim Simons, billionaire hedge fund manager of Renaissance Technologies, and mathematician, who expanded quantitative analysis into Wall Street. Both mathematicians and hedge fund managers have been extremely successful and made quantitative analysis a chief investment strategy among all the strategies used by money managers and professional speculators to predict the market. Therefore, the idea of AI predicting stock movements is unsurprising.
Alejandro Lopez-Lira, a finance professor at the University of Florida, said that large language models may be useful when forecasting stock prices. The finance professor claimed that he used ChatGPT to parse news headlines for whether they are good or bad for a stock, and found that ChatGPT’s ability to predict the direction of the next day’s returns was very much better than random. AI may be able to predict stock movements based on the algorithmic structure that enables it to discern patterns in the market, but it will not be able to correctly forecast stock prices.
The reason behind that is that stock prices are simply based on supply and demand and are therefore not correlated to a company’s fundamentals; at least in the short-term. It would be impossible for ChatGPT, for example, to determine that the price of Microsoft would go from $287 a share when the market opens to $295 a share when the market closes. During trading hours, anything can happen. Stock prices are determined by market forces (supply and demand) and market forces are based on human psychology. Human psychology is something that machines cannot entirely fathom no matter how well-programmed they are to determine the mechanics of human behavior. All it takes is one piece of information to make the market move from one direction to another. Suppose ChatGPT predicts that the price of Berkshire Hathaway (BRK) Class A stock will go from $400,000 to $405,000 today, but Warren Buffett suddenly passes away in the middle of that same day. This prediction will be completely inaccurate because the news of Buffett’s passing will plunge BRK stock price. Investors will react to the news and sell their BRK shares, which will make the price of the stock fall while ChatGPT did not factor Buffett’s death in its prediction, and therefore failed to anticipate how the market would react to such news.
ChatGPT can surely predict stock movements because stock movements are based on companies’ fundamentals. In the long-term, a company's stock price is supposed to increase or decrease based on the company’s performance. Great performance reflected by increasing revenues will logically increase the stock price of a company. Poor performance reflected by decreasing revenues and poor management in the company’s leadership will obviously decrease its stock price. Stock movements are based on data. Data help us make informed decisions for the future based on past performances because data is trends and trends are based on patterns. But past performances do not guarantee future results. Stock movements are signals about market conditions.
ChatGPT is surely useful for us to make informed decisions about the way we seek to invest because it gives us an overview of market conditions. But we shouldn’t make investment decisions based on price predictions because even if ChatGPT is able to predict stock prices, there is a strong chance that those predictions will be inaccurate and make investors lose money.