Testing Homogeneity Hypothesis of Network Stock Market Models

The present research presents the statistical investigation of a stock market. The basic assumption of the study is the stationarity of stock market relations. The assumption is formulated in terms of mathematical statistics as the equality hypothesis of multivariate normal correlation matrices. A mathematical problem is considered from the perspective of multiple hypothesis testing. The work touches the main methods and principles of multiple hypothesis testing: Roy’s union-intersection principle and Bonferroni correction. Then to test individual hypotheses the well-known methods for testing of covariance matrices equality are investigated. Box’s M-test was chosen under investigation, and further its properties were studied. In the final part of the research there are results of experiments conducted on real data and using the Box’s M-test and Bonferroni procedure.