Predicting Stock Price Direction on Earnings Announcement Days using Multi-modal Deep Learning
2026-05-25 • Machine Learning
Machine Learning
AI summaryⓘ
The authors studied how well different types of information—like company financial details, recent stock price trends, and news sentiment before earnings announcements—can predict if a stock's price will go up or down on the day of those announcements. They used advanced machine learning models, including LSTM and Transformer networks, comparing them to a simpler logistic regression method. Their findings show that the LSTM model is better at making cautious, precise predictions, while the Transformer model is better at catching sudden big changes. Including news sentiment consistently helped improve predictions across all models.
Earnings AnnouncementsStock Price PredictionNews Sentiment AnalysisLong Short-Term Memory (LSTM)Transformer ModelLogistic RegressionFundamental MetricsTechnical IndicatorsFinBERTMacro F1-Score
Authors
Manuel Noseda, Nathan Soldati, Marco Paina
Abstract
Predicting stock price movements during Earnings Announcements (EAs) is a significant challenge due to market noise and high-impact price discontinuities. In this study, we evaluate whether pre-announcement news sentiment, firm fundamentals, and recent market dynamics jointly predict the directional price movement of equities on EA days. We construct a multi-modal feature space combining 15 fundamental metrics, 3 price-based technical indicators and sentiment scores derived from financial news articles processed using FinBERT. We compare a Long Short-Term Memory (LSTM) network and a Transformer-based architecture against a logistic regression baseline, and further assess all models with and without sentiment features to quantify their incremental value. Our results indicate that while the LSTM demonstrates higher precision through a conservative safe-bet strategy, the Transformer model exhibits superior sensitivity in identifying volatile movements, achieving a higher macro F1-score, with ablation experiments showing a consistent benefit from incorporating news sentiment.