When Does Centrality Matter in Portfolio Choice? – Roope Rihtamo (Department of Mathematics and Statistics, UTU)

This paper examines the inverse relationship between network centrality measures and optimal weights of a variance minimizing portfolio allocation. We revisit previous results and their underlying assumptions and consider alternative perspectives and refinements proposed in the literature in order to provide a deeper understanding of the suggested relationship. Building on this, we derive the necessary and sufficient conditions under which the inverse relation between centrality and optimal weights holds. Our analytical and computational results show that certain simple and natural covariance structures satisfy these conditions. Furthermore, we empirically validate our theoretical insights using return data from the constituents of the S&P500 index. Our findings offer investors guidance on when and how centrality-based measures can be used in portfolio selection. This work contributes to bridging complex network theory with classical portfolio optimization, offering practical implications applicable to various investment strategies.