Redes Complexas aplicadas à otimização de portfólio de ações
Carregando...
Arquivos
Citações na Scopus
Tipo de produção
Dissertação
Data
2023
Autores
Varga, Felipe Souza
Orientador
Rodrigues, Paulo Sérgio Silva
Periódico
Título da Revista
ISSN da Revista
Título de Volume
Citação
VARGA, Felipe Souza. Redes Complexas aplicadas à otimização de portfólio de ações. 2023. 91 p. Dissertação (Mestrado em Engenharia Elétrica) - Centro Universitário FEI, São Bernardo do Campo, 2023. Disponível em: https://doi.org/10.31414/EE.2023.D.131620.
Texto completo (DOI)
Palavras-chave
Redes Complexas,Mercado financeiro,Investimentos
Resumo
A compreensão dos mercados financeiros é algo que desperta grande interesse nas pessoas.
Eles foram estudados por diferentes áreas do conhecimento, tais como: economia, matemática
e física. Apesar deste interesse, apenas recentemente que estes mercados vêm sendo estudados
pela perspectiva de Redes Complexas. Assim, este trabalho modela e estuda o mercado
financeiro brasileiro pela perspectiva dessas redes. O mercado brasileiro foi escolhido como
objeto de estudo, pois o foco da literatura identificada é direcionado aos mercados financeiros
de países desenvolvidos ou aqueles cujas economias são muito fortes no contexto internacional.
As redes são construídas com base no coeficiente de correlação calculado para todos
pares de Ações do mercado. Estas correlações são obtidas a partir dos retornos logarítmicos
diários das Ações divididos em janelas de tempo. As redes construídas formam redes completas,
entretanto, para análise de suas estruturas, é necessário filtrar as informações da rede. Com
esse fim, são geradas Minimum Spanning Tree (MST) das redes. Então, é realizada a extração
e o estudo das propriedades físicas destas árvores. Desta forma, será possível estudar as dinâmicas das redes de mercado financeiro ao longo do tempo. Parte desse estudo será focado em
compreender a reação destas propriedades a crises financeiras. Uma vez que estas propriedades
foram compreendidas, elas são utilizadas para realização da seleção de portfólios de Ações. Para
isso, são selecionadas Ações com menor centralidade da rede, uma vez que são tais Ações que
costumam gerar maior diversificação para o portfólio. Essa ideia corrobora com a teoria moderna
de portfólio de Markowitz, na qual é descrito que a otimização de portfólios é dependente
do grau de diversificação de seus ativos. Os portfólios gerados foram avaliados e comparados a
portfólios de referência, tais como: Portfólio de Markowitz, Portfólio Igualmente Ponderado e
Índice Ibovespa
Understanding financial markets is something that arouses great interest in people. They were studied by different areas of knowledge, such as: economics, mathematics and physics. Despite this interest, these markets have only recently been studied from the perspective of Complex Networks. Thus, this work models and studies the Brazilian financial market from the perspective of these networks. The Brazilian market was chosen as the object of study, as the focus of the identified literature is directed to the financial markets of developed countries or those whose economies are very strong in the international context. The networks are built based on the calculated correlation coefficient for all pairs of stocks in the market. These correlations are obtained from the daily logarithmic returns of the Shares divided into time windows. The constructed networks formcomplete networks, however, to analyze their structures, it is necessary to filter the network information. For this purpose, MST of networks are generated. Then, the extraction and study of the physical properties of these trees is carried out. In this way, it will be possible to study the dynamics of financial market networks over time. Part of this study will focus on understanding the reaction of these properties to financial crises. Once these properties have been understood, they are used to carry out the selection of stock portfolios. For this, Stocks with lower network centrality are selected, since these are the Stocks that usually generate greater diversification for the portfolio. This idea corroborates Markowitz’s modern portfolio theory, in which it is described that portfolio optimization is dependent on the degree of diversification of its assets. The generated portfolios were evaluated and compared to reference portfolios, such as: Markowitz Portfolio, Equally Weighted Portfolio and Ibovespa Index
Understanding financial markets is something that arouses great interest in people. They were studied by different areas of knowledge, such as: economics, mathematics and physics. Despite this interest, these markets have only recently been studied from the perspective of Complex Networks. Thus, this work models and studies the Brazilian financial market from the perspective of these networks. The Brazilian market was chosen as the object of study, as the focus of the identified literature is directed to the financial markets of developed countries or those whose economies are very strong in the international context. The networks are built based on the calculated correlation coefficient for all pairs of stocks in the market. These correlations are obtained from the daily logarithmic returns of the Shares divided into time windows. The constructed networks formcomplete networks, however, to analyze their structures, it is necessary to filter the network information. For this purpose, MST of networks are generated. Then, the extraction and study of the physical properties of these trees is carried out. In this way, it will be possible to study the dynamics of financial market networks over time. Part of this study will focus on understanding the reaction of these properties to financial crises. Once these properties have been understood, they are used to carry out the selection of stock portfolios. For this, Stocks with lower network centrality are selected, since these are the Stocks that usually generate greater diversification for the portfolio. This idea corroborates Markowitz’s modern portfolio theory, in which it is described that portfolio optimization is dependent on the degree of diversification of its assets. The generated portfolios were evaluated and compared to reference portfolios, such as: Markowitz Portfolio, Equally Weighted Portfolio and Ibovespa Index