Construct network portfolio in stock market
In stock market, stocks are vertices and edges are connectedness between them. Thus, we can apply network theory to stock market and construct stock network. A stock connected to many other stocks is in a central position and can be measured by centrality. We manage to demonstrate the importance of network risk by portfolio risk decomposition. By centrality constraint, we can construct portfolio by constraining portfolio centrality to control additional network risk to classical portfolio theory, which we call network portfolio. Empirical results show that network portfolio outperform global minimum variance portfolio and equally weighted portfolio by both return and risk.
Applied Mathematic PhD student in Southwestern University of Finance and Economics, and visiting student in IRTG group, Humboldt University of Berlin