Correlation network analysis for multi-dimensional data in stocks market

• Published in: Physica A Vol. 429; pp. 62 - 75
• Main Authors: Kazemilari, Mansooreh, Djauhari, Maman Abdurachman
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01-07-2015
• Subjects: Escoffier’s operator; Multivariate association; Stocks network analysis; Vector correlation; Multivariate association; Escoffier’s operator; Stocks network analysis; Vector correlation
• ISSN: 0378-4371; 1873-2119
• DOI: 10.1016/j.physa.2015.02.052

This paper shows how the concept of vector correlation can appropriately measure the similarity among multivariate time series in stocks network. The motivation of this paper is (i) to apply the RV coefficient to define the network among stocks where each of them is represented by a multivariate time series; (ii) to analyze that network in terms of topological structure of the stocks of all minimum spanning trees, and (iii) to compare the network topology between univariate correlation based on r and multivariate correlation network based on RV coefficient.
•Stocks network based on opening, highest, lowest, and closing prices is introduced.•Escoufier’s RV coefficient is used to measure the similarity among stocks.•Information in multi-dimensional network is filtered using minimal spanning tree.•The topological properties of stocks are analyzed by using centrality measures.•30 Dow-Jones stocks are used to illustrate the advantage of RV coefficients network.