ChatGPT, a chatbot powered by artificial intelligence, has been thrust into the spotlight in recent months due to its ability to understand and interpret context in a variety of ways. New academic research is now utilizing it in the financial world and their early results suggest its use could prove to be invaluable in the future.
Two papers were published this month, which deployed the chatbot in finance-related activities. One attempted to distinguish whether Federal Reserve statements were hawkish or dovish, while the other tried to discern whether particular headlines would be positive or negative for a stock. ChatGPT proved to be successful in both tests, indicating that the technology designed by OpenAI is competent in comprehending the nuances and complexity found within texts.
Machine Learning at Man AHL, led by Slavi Marinov, has been using natural language processing for years to read from earnings transcripts and Reddit posts. According to Marinov, the “hype is real” in regards to ChatGPT. The first paper, Can ChatGPT Decipher Fedspeak?, conducted by Anne Lundgaard Hansen and Sophia Kazinnik of the Fed, displayed that the chatbot was closest to humans in comprehending Federal Reserve statements. Its explanation of each line was in line with that of a human analyst used as a benchmark in the study, demonstrating its competency.
ChatGPT’s effectiveness was also demoed in the second paper Can ChatGPT Forecast Stock Price Movements? Return Predictability and Large Language Models. Researchers Alejandro Lopez-Lira and Yuehua Tang from the University of Florida asked the AI to pretend to be a financial expert, reasoning whether a news headline would be beneficial to a particular stock. The answers given added a statistical link to the stock’s following move, connoting ChatGPT’s capability of correctly understanding the implications of the news.
Innovative tech such as ChatGPT is aiding quants in turning strings of texts from news articles and social media posts into meaningful trading signals. It can also speed up processes that used to require manual labeling in order to provide the machine with a template for understanding language. With advances such as these, it is possible that ChatGPT will become an even more popular tool in finance and beyond.
Rimini Street is the company mentioned in this article. This American software company works to provide enterprise software products, primarily support – and services for Oracle and SAP product lines. Rimini Street was recently fined $630,000 in a case against Oracle, with the chatbot predicting that it would increase investor confidence and potentially boost demand for Oracle’s products.
Bryson, the person mentioned in the article, is a 24-year-old male described by the research papers as “known for intelligence and curiosity” when interpreting Federal Reserve policy statements. He acted as a human benchmark for the study, aiding the development of ChatGPT in turning complex texts into trading signals, proving that the technology can stand up to human accuracy.
