Leveraging NLP-Driven Sentiment Analysis for Enhancing Decision-Making in Supply Chain Management

The integration of Natural Language Processing (NLP) into asset pricing through the analysis of sentiment in financial news articles. We presents a comprehensive procedure to gather news articles, analyze sentiment using advanced NLP techniques, convert sentiment into a continuous time-series signal, and develop models for excess returns based on lagged sentiment. Both statistical and machine learning methods are employed to model excess returns but encounter convergence issues over extended periods in the sentiment-return relationship. While a sentiment-based trading strategy is formulated, it yields negative Sharpe ratios for three out of four companies. Nonetheless, it outperforms a baseline strategy that does not incorporate sentiment features, underscoring the informative nature of sentiment in financial decision-making. Further research is warranted to establish a consistent and dependable trading strategy based on sentiment in financial news. Therefore, this paper underscores the potential of NLP in asset pricing and highlights avenues for future exploration.