CEPEX: Enhancing Corporate Event Prediction with Stock Price Dynamics and GRU-Attention Networks

Accurate corporate event prediction (CEP) presents a significant challenge for decision-making in financial markets. In this work, we reframe the CEP task as a problem of forecasting the optimal next step in a corporate life-cycle sequence, and introduce CEPEX, a novel predictive framework that enhances a Gated Recurrent Unit (GRU) attention-based model by uniquely incorporating quantified stock price dynamics alongside historical event sequences. Evaluated on the comprehensive Chinese Corporate Event Prediction (CCEP) dataset, CEPEX achieves a state-of-the-art accuracy of 84.28%. A key contribution of our work is the demonstration that stock market data provide notably stronger predictive power for future corporate actions than financial news. This framework improves decision-support systems in financial informatics and offers a new approach for predictive control in dynamic market environments.