Journal of Asset Management and Financing (Mar 2021)
Establishment of a Non-Linear Financial Network Based on its Typological Characteristics Based on Graph Theory (A Study in Tehran Stock Exchange)
Abstract
AbstractThe purpose of this study is to introduce a financial network based on non-linear relationships between stocks to optimize the portfolio of investors,identify the leaders of the Iranian stock market using centrality criteria and finally clustering non-linear financial network.In this study,the top 100 companies listed on the stock exchange with the highest capital registered in the 11-year period (December 2009 to January 2020) were selected.The results show that according to the degree centrality, the stocks of Sepahan Cement,Omid Capital Financing,and Omid Investment,according to the criterion of closeness centrality,Ghadir investment stocks, investment of National Development and Khuzestan Steel, According to the closeness centrality,Ghadir investment stocks,National Development and Khuzestan Steel Investment Group, according to the betweenness centrality,Ghadir Investment stocks, Sepahan Cement and National Development Investment and according to the bottleneck centrality, the stocks of Khuzestan Steel,Sepahan Cement and International Building Development have the most impact on the stock market and were identified as market leaders. To categorize the top stocks, the fast greedy algorithm was used, in which the network was divided into 11 clusters, and each of these clusters represents the largest relationship between the shares of companies in the financial network. Introduction:A stock portfolio is a collection of the best stocks in which each stock has a certain return and risk. What is very important in forming a portfolio with the least amount of risk is to find stocks that have the least amount of relationship with each other. In order to examine the relationship between the stocks of different companies and consequently the selection of the optimal stock portfolio, there are different methods and techniques that can be used. One of the best techniques for identifying and selecting the optimal portfolio of diversified stocks is to identify the relationship and correlation and then clustering between different stocks and grouping them based on the important factors that investors consider for investing. Using this technique, stock selection and the formation of an optimal portfolio of different groups is done, which in addition to being able to solve the problem of expected returns of investors, also the problems caused by the investment risk in the stock market can be solved. One of the most important problems in modern financial discussions is finding efficient methods for presenting and summarizing data produced by the stock exchange, and this information is displayed in thousands of forms, each of which separately represents the price movement of each stock. As the number of stocks increases, the analysis of these forms will become more complex According to recent research, the complex network method is highly recommended for visualizing and summarizing stock data and examining the relationship between stock prices. Using complex network analysis, a clear picture of the internal structure of the stock exchange can be provided Analyzing stock market statements, examining how they evolve over time, and describing patterns within the stock market are important and useful for developing and designing investment strategies. Therefore, the purpose of this study is to create and introduce a financial network based on stock relationships in companies listed on the Tehran Stock Exchange, which will be provided by a minimum spanning tree. This network will be examined by the centrality measures and among the stocks of companies, top stocks and stock market leaders will be examined according to different measures, and finally the top stocks clustered will help to investors in order to optimize the portfolio and maximize Investment profit. Method and Data:The present study is applicable in terms of purpose, quantitative in terms of implementation process, retrospective and post-event in terms of time. R software is used to analyze data. The daily data of 100 companies that had the most market capital in Tehran Stock Exchange were received in 243 working days from "Tehran Stock Exchange site" from 2009 to 2019. This data corresponds to 11 solar years that have been selected as a sample to make a spanning tree and compare companies based on them. The financial network was converted to logarithmic returns using adjusted closing price. The concepts of graph theory and prim algorithm were used to explore the relationships and distances between stocks to construct a minimum spanning tree. Findings:The Findings show that according to the degree centrality, the stocks of Sepahan Cement Companies, Omid Capital Financing, and Omid Investment Management, according to the criterion of closeness centrality , Ghadir investment stocks, investment of National Development Group and Khuzestan Steel, According to the closeness centrality, Ghadir investment stocks, National Development Group and Khuzestan Steel Investment Group, according to the betweenness centrality, Ghadir Investment Company stocks, Sepahan Cement and National Development Investment and according to the bottleneck centrality, the stocks of Khuzestan Steel Company, Sepahan Cement and International Building Development have the most impact on the stock market and were identified as market leaders. To categorize the top stocks, the fast-greedy algorithm was used, in which the network was divided into 11 clusters, and each of these clusters represents the largest relationship between the shares of companies in the financial network. Conclusion and discussion: This study sought to investigate the nonlinear relationship between the most valuable stocks in the stock market. In addition to creating a network to identify relationships between stocks, market leaders were also identified who can influence the network based on various measures. Finally, to optimize the stock portfolio, all stocks in the network were clustered to reduce portfolio risk.
Keywords
