Clusterization of Indices and Assets in the Stock Market

K-means clustering algorithm has been used to classify major world stock market indices as well as most important assets in the Warsaw stock exchange (GPW). In addition, to obtain information about mutual connections between indices and stocks, the Granger-causality test has been applied and the Pearson R correlation coefficients have been calculated. It has been found that the three procedures applied provide qualitatively different kind of information about the groups of financial data. Not surprisingly, the major world stock market indices appear to be very strictly interconnects from the point of view of both Granger-causality and correlation. Such connections are less transparent in the case of individual stocks. However, the “cluster leaders” can be identified which leads to the possibility of more efficient trading.