Back to Research
Sentiment Analysis
Sentiment Trading with Large Language Models
Finance Research Letters · 2024
What this means for traders
LLM-based sentiment reading produces Sharpe ratios more than twice as high as traditional dictionary methods.
Abstract
This study compares dictionary-based methods with modern LLMs for sentiment analysis. Strategies based on OPT-66B generated a Sharpe Ratio of 3.05, significantly outperforming traditional methods (Sharpe 1.23).
Sentiment AnalysisLLMsTrading StrategiesAlpha Generation
Read the full paperResearch papers are presented for informational purposes only. Past performance reported in academic studies does not guarantee future results. These papers are not personalised investment advice.