Prediction of abnormal returns on bidding firms in mergers and acquisitions

This study investigates the predictability of abnormal returns for bidding firms following M%A announcements in China, focusing on deal characteristics and firm size. Using 1,253 transactions from 2010-2018, we apply a K-nearest neighbours (KNN) model to classify whether post-announcement abnormal returns are positive or negative across Day +1, Day +2, and CAR (1, 2) windows. Predictions draw on four variables: form of acquisition, payment method, industry-relatedness, and firm size. Results show strong predictability for small acquirers - those in the bottom 15%-30% by assets - achieving over 70% accuracy across windows. Predictive power declines with larger firms and approaches randomness for the full sample. The findings indicate that abnormal returns are more systematically predictable when involving smaller bidders and clear structural deal features, offering insights for investors and managerial decision-making in short-term market reactions.