재무연구 제 권 호 (2017년 5월)
Asian Review of Financial Research, Vol., No..
pp.1554~1576
pp.1554~1576
A network approach to optimal portfolio : Application to the Korean stock market
Seok-won Ahn Division of Business Administration, Chosun University, Korea
Yong-Choel Kim Lubar School of Business, University of Wisconsin-Milwaukee, Po Box 742, Milwaukee, Wisconsin 53201-0742, USA.
Gabjin Oh Division of Business Administration, Chosun University, Korea
Portfolios constructed from a sparse correlation-based network of stocks have rich and practical implications for stock index formulation and portfolio management. A hub portfolio consisting of approximately 20 stocks with a high degree of connections with remaining stocks shows a significantly positive correlation with the KOSPI index, and represents the total stock market with fewer stocks. An outlier portfolio containing stocks with small number of connections with other stocks shows minimum portfolio risk with about one-fifth of total number of stocks and higher risk-adjusted performance than Markowitz’s random portfolio. A minimum variance portfolio (MVP) constructed with 50 outlier stocks in a sparse network of low interconnections suggests the existence of an optimal reference portfolio and the sensitivity of individual stocks to the MVP as an alternative risk measure similar to the market β risk.
Seok-won Ahn
Yong-Choel Kim
Gabjin Oh
Portfolios constructed from a sparse correlation-based network of stocks have rich and practical implications for stock index formulation and portfolio management. A hub portfolio consisting of approximately 20 stocks with a high degree of connections with remaining stocks shows a significantly positive correlation with the KOSPI index, and represents the total stock market with fewer stocks. An outlier portfolio containing stocks with small number of connections with other stocks shows minimum portfolio risk with about one-fifth of total number of stocks and higher risk-adjusted performance than Markowitz’s random portfolio. A minimum variance portfolio (MVP) constructed with 50 outlier stocks in a sparse network of low interconnections suggests the existence of an optimal reference portfolio and the sensitivity of individual stocks to the MVP as an alternative risk measure similar to the market β risk.
correlation-based network,outlier and hub portfolio,diversification benefit,convexity of diversification.
