Modularity cluster finding in financial time series ‎

In this paper, relying on the clustering of complex networks that can determine large scale features of ‎the network, we study 48 financial markets across the world. To this end, we develop a modularity ‎maximization method for directed and weighted networks. According to the linear correlation measure, ‎we construct the adjacency matrix, and by using the theory of random matrices, we divide the space of ‎eigenvalues of our matrix into two irrelevant and relevant fragments. By considering the temporal ‎window and its evolution over time series, our results demonstrate that in the vicinity of so-called ‎financial crisis clusters, which are often affected by geographical characteristics, are formed and from the ‎perspective of complex networks, they show more random behavior‎.‎‎