The Art of Investment Portfolio Curation through Centrality Metrics (An Enchanting Network Analysis of Tehran Stock Exchange's Top 50 Companies)

The stock market, as a prominent financial domain, presents a formidable challenge in comprehending and evaluating an extensive array of stocks employing centrality measurements to portray the key variables within a network, companies' stocks can be visualized and comprehended. The utilization of stock network analysis facilitates a comprehensive understanding of the entire network through diverse visualization techniques. This study delves into the data of the top 50 companies listed on the Tehran Securities Exchange during the period from January 1, 2019, to July 6, 2021. Employing unsupervised machine learning tools such as Community Detection algorithms and network analysis methods like Louvain and Girvan-Newman, we construct a stock network. Subsequently, we compute five Centrality Metrics, including Degree Centrality, Closeness Centrality, Eigen Centrality, Betweenness Centrality, and PageRank, for these companies. By formulating a similarity matrix based on these criteria for the remaining stocks in the network, we determine a portfolio of 25 stocks suitable for investment, derived from the ranking of stocks according to the Centrality Metrics