Uma análise por redes complexas dos ativos financeiros da bolsa de valores do Brasil: caracterização, predição e simulação de investimentos

Abstract

Complex Networks are promising tools for understanding Complex Systems. It is a field of interdisciplinary studies, effective in modeling several phenomena of strategic and/or market interest. Its applications are present in several areas of knowledge such as Biology, Education, Sociology, Health and Economics, as in the Stock Market. In this sense, this work has analyzed the Brazilian Stock Market, as well as the external variables related to it in the context of complex networks. To this end, based on the concept of moving networks, 473 unweighted and undirected complex networks of B3 financial assets were developed with relationships based on the Pearson’s correlations between the logarithms of asset returns. The financial data evaluated comprises the period of 43 months, between January 2, 2017 and July 31, 2020. The results suggest that the variables Ibovespa index, foreign capital flow in B3 and Selic rate, basic interest rate, influence the topological structure of the asset correlation networks. The sectors participating in B3 showed, in general, statistically significant negative correlations between the sectorial returns and the global metrics coming from the respective sectorial networks. In addition, the most profitable assets, as well as those with lower risk evaluation presented, on average, lower degree centrality, PageRank, eigenvector centrality, among others, when compared to the local metrics of the other assets observed. The investment simulations based on local asset metrics showed portfolio options with a higher annual rate of return than the financial investments in the Ibovespa and Selic reference indices.