Trading patterns of institutional investors: applications of machine learning

Abstract

ABSTRACT In financial literature, institutional investors hold profound influence on stock dynamics. In Taiwan’s stock market, institutional investors dominate largely due to the nation’s unique financial regulations mandating daily trading data disclosure. This high-frequency data, distinct to Taiwan, offers an unparalleled opportunity for in-depth market analysis. Despite their importance, scant research employs machine learning to predict these investors’ dynamic movements. Our study fills this gap, leveraging Taiwan’s unique 2019–2022 transactional data with machine learning techniques. Impressively, insights derived yielded an 8.8% average cumulative return in just 20 days, highlighting the potential of understanding and leveraging these dynamics.