Using influential nodes of stock network to measure Tehran Stock Exchange index

Abstract

In the capital market, the stocks which have more impact on the other stocks, are called “influential stocks.” If we can find the influential stocks of market, we can analyze present and future status of the market only by investigating them. This paper proposes a method for determining the influential stocks and estimating the market index using them. The influential stocks are recognized based on the market correlation-based network. This network is a graph, which the nodes are the existing stocks in the market and the edges are the similarity of the price fluctuations of corresponding stocks. Since the network follows power-law distribution, there are only few nodes with extreme high degree, i.e., influential stocks. The results of the experiments show that the market index calculated using the influential stocks fluctuates very similar to the main index, which includes all the stocks. In addition, the proposed method reduces the sudden and minor fluctuations of the main index.