Two new papers published this month show the impressive ability of ChatGPT, an artificial intelligence chatbot, in market-relevant tasks. The technology was tested in determining whether headlines were good or bad for a stock and deciphering whether Federal Reserve statements were hawkish or dovish. In both tests, ChatGPT excelled and displayed its ability to parse nuance and context with exceptional accuracy.
According to Slavi Marinov, head of machine learning at Man AHL, ‘It’s one of the rare cases where the hype is real.’ The first paper tested ChatGPT’s ability to decipher Fedspeak, and two researchers from the Federal Reserve found that it was the closest to humans in identifying whether the central bank’s statements were dovish or hawkish. The second paper tested whether headlines were good or bad for a stock, and ChatGPT accurately explained its classifications.
The ability to turn reams of text from news articles to tweets and speeches into trading signals is not new on Wall Street. However, the results of these studies suggest that the technology developed by OpenAI has reached a new level in terms of parsing nuance and context. The advances demonstrated by ChatGPT look set to open up whole worlds of new information and make the tech more accessible to a broader community of finance professionals.
To Marinov, while there’s no surprise machines can now read almost as well as people, ChatGPT can potentially speed up the whole process. Previously, the quant hedge fund manually labeled each sentence as positive or negative for an asset to give the machines a blueprint for interpreting the language. With ChatGPT, that process can potentially be done faster and more efficiently.
The two new papers suggest ChatGPT can pull off similar tasks without even being specifically trained. The technology shows promise in making NLP more accessible and may open up new opportunities for finance professionals. Bloomberg LP also released a large language model for finance last month.
Post Your Comments