Visualization of a Stock Market Correlation Matrix
This paper presents a novel application of Neighbor-Net, a clustering algorithm developed for constructing a phylogenetic network in the field of evolutionary biology, to visualizing a correlation matrix. We apply Neighbor-Net as implemented in the SplitsTree software package to 48 stocks listed on the New Zealand Stock Exchange. We show that by visualizing the correlation matrix using a Neighbor-Net splits graph and its associated circular ordering of the stocks that some of the problems associated with understanding the large number of correlations between the individual stocks can be overcome. We compare the visualization of Neighbor-Net with that provided by hierarchical clustering trees and minimum spanning trees. The use of Neighbor-Net networks, or splits graphs, yields greater insight into how closely individual stocks are related to each other in terms of their correlations and suggests new avenues of research into how to construct small diversified stock portfolios.