Key facts
- Matthew Dixon, a quant and academic, tested the reliability of large language models.
- Dixon used OpenAI's GPT-4.1 Mini to predict US stock market sentiment over five days.
Matthew Dixon, a quant and academic focused on the trustworthiness of large language models (LLMs), conducted an experiment to assess their reliability. He utilized OpenAI's GPT-4.1 Mini, asking it to predict the US stock market's mood over a five-day period. The study aimed to uncover potential 'lexical risk,' a concern for quantitative analysts as LLMs become more integrated into financial analysis.