Man Group Tests ChatGPT for Financial Sentiment Analysis
Investment firm Man Group PLC has explored ChatGPT's capabilities in financial sentiment analysis. Summer interns compared the AI's performance against traditional word-counting methods on financial texts.
Man Group PLC, an investment management firm, has investigated the effectiveness of ChatGPT in performing financial sentiment analysis. The study, conducted by summer research interns, aimed to determine if the artificial intelligence tool can match or surpass existing baseline methods.
The research focused on targeted sentiment analysis, a method that assesses the sentiment of a statement regarding a specific aspect. For example, in a sentence like 'assets under management increased, but our profits decreased,' targeted analysis can identify the positive sentiment towards assets despite a negative profit outlook.
Traditionally, financial sentiment analysis has relied on dictionary-based approaches, such as the Loughran McDonald (LM) lexicon, which categorizes financial terms. However, the advent of large language models (LLMs) like ChatGPT has prompted re-evaluation of these tools for complex text analysis.
The interns evaluated ChatGPT's ability to discern sentiment in financial texts using various prompting techniques. Their findings contribute to understanding how advanced AI models can be applied within the financial services sector for data interpretation.
While the firm has not released detailed results, the project highlights the ongoing exploration of AI's role in enhancing financial analysis processes.