The world of finance has been rocked by the advent of ChatGPT – a cutting-edge artificial intelligence chatbot that could potentially revolutionize the way investments are analyzed. Two recent papers have used the technology to take on market-relevant tasks such as deciphering the meaning of hawkish and dovish statements by the Federal Reserve, as well as decoding whether stock-related headlines were positive or negative. The results showed that ChatGPT was able to accurately predict the market direction based on text, making it a groundbreaking advancement in the use of natural language processing.
Research conducted by Anne Lundgaard Hansen and Sophia Kazinnik of the Federal Reserve proved that the chatbot was more successful than existing models in discerning the intent of the Fed’s statements, even beating out ‘Bryson’ – a 24-year-old analyst chosen to be the human benchmark for the study. In a paper titled “Can ChatGPT Decipher Fedspeak?”, Hansen and Kazinnik demonstrated ChatGPT’s capability to explain its classifications of Fed policy statements in a manner similar to an analyst, which could be very advantageous due to the time saved.
“It’s one of the rare cases where the hype is real,” said Slavi Marinov, head of machine learning at Man AHL, who’ve long used language models to sift through text data from common sources such as Twitter and Reddit.
The second paper, “Can ChatGPT Forecast Stock Price Movements? Return Predictability and Large Language Models”, authored by Alejandro Lopez-Lira and Yuehua Tang of the University of Florida, suggests that the AI was able to use corporate news headlines to accurately project stock outcomes.
The research has been a reassuring sign for investors, as ChatGPT’s natural language processing look set to expand the horizons of artificial intelligence in finance beyond the work done by quant hedge funds. According to Marinov, while machines reading texts with the same degree of accuracy as humans was expected, the chatbot further reduces the time needed as it can discern implication from news headlines without a human-curated dataset.
Bolstering the field of natural language processing is Bloomberg LP’s large language model for finance, released just last month. With the application of AI to text data, analysts can now more easily tap into the realm of unstructured data and gain valuable insight from it.
