Structured Event Representation and Stock Return Predictability
Analysis
This research paper explores the use of large language models (LLMs) to extract event features from news articles for predicting stock returns. The authors propose a novel deep learning model based on structured event representation (SER) and attention mechanisms. The key finding is that this SER-based model outperforms existing text-driven models in out-of-sample stock return forecasting. The model also offers interpretable feature structures, allowing for examination of the underlying mechanisms driving stock return predictability. This highlights the potential of LLMs and structured data in financial forecasting and provides a new approach to understanding market dynamics.
Key Takeaways
- •LLMs can be effectively used to extract event features from news articles.
- •A structured event representation (SER) model improves stock return prediction.
- •The SER model offers interpretable feature structures for understanding market dynamics.
“Our SER-based model provides superior performance compared with other existing text-driven models to forecast stock returns out of sample and offers highly interpretable feature structures to examine the mechanisms underlying the stock return predictability.”