Extracting the Structure of Press Releases for Predicting Earnings Announcement Returns
This work addresses the problem of real-time stock return prediction for investors and analysts, offering an incremental improvement by integrating advanced NLP methods with traditional financial data.
The study tackled predicting stock returns on earnings announcement days using textual features from press releases, finding that content is as informative as earnings surprises and that FinBERT provides the highest predictive power, with models combining to enhance explanatory strength.
We examine how textual features in earnings press releases predict stock returns on earnings announcement days. Using over 138,000 press releases from 2005 to 2023, we compare traditional bag-of-words and BERT-based embeddings. We find that press release content (soft information) is as informative as earnings surprise (hard information), with FinBERT yielding the highest predictive power. Combining models enhances explanatory strength and interpretability of the content of press releases. Stock prices fully reflect the content of press releases at market open. If press releases are leaked, it offers predictive advantage. Topic analysis reveals self-serving bias in managerial narratives. Our framework supports real-time return prediction through the integration of online learning, provides interpretability and reveals the nuanced role of language in price formation.