군집화(Clustering)를 이용한 멀티팩터모형의 스타일 운용에 대한 사례 연구

As the public pension market, particularly including the National Pension, and various other public and retirement pension markets continue to expand, the development of differentiated management strategies by institutions managing these funds (such as direct management institutions and Outsourced Chief Investment Officers (OCIO)) is emerging as a significant research topic. Recently, major pension fund management institutions like Japan's GPIF have been actively adopting multi-factor models and artificial intelligence (AI) systems, and private asset management companies are also making extensive efforts to apply these technologies to their management strategies. This study aims to respond to such market demands by classifying fund styles through clustering techniques and verifying whether it is possible to construct low-cost, high-efficiency diversified portfolios based on this classification. The main contents of the study are as follows: First, it analyzes the style management strategies of domestic equity funds using a multi-factor model. Second, it applies the K-means clustering technique to cluster fund styles and derive the strategic characteristics of each cluster. Third, it compares and analyzes how the clustered fund styles differentiate from the style classification criteria commonly used by existing management institutions (such as large-cap/small-cap, value/growth). Lastly, it verifies whether it is possible to achieve high-efficiency diversification effects using clustered funds and the applicability of active management strategies. This study is expected to contribute to asset management companies in establishing more sophisticated and differentiated management strategies by empirically demonstrating the potential of enhancing fund management efficiency through the use of data analysis and artificial intelligence technologies.