FURI | Spring 2026

AI-Driven Portfolio Optimization Using News Sentiment-Based Volatility Forecasting and Reinforcement Learning

This study evaluates whether financial news sentiment can improve portfolio optimization through volatility forecasting. Results show that integrating sentiment-based signals into a reinforcement learning framework improves risk-adjusted returns and reduces drawdowns compared to traditional models. By incorporating real-time information from news data, this approach enables more adaptive and risk-aware investment decisions. Future work should explore multimodal data sources and test the framework across different market conditions and asset classes.