Estratégias de investimento baseadas em microestrutura de mercado

• Ano de defesa: 2019
• Autor(a) principal: Alef Willis Magno Miranda
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; Brasil; ICX - DEPARTAMENTO DE CIÊNCIA DA COMPUTAÇÃO; Programa de Pós-Graduação em Ciência da Computação; UFMG
• Palavras-chave em Português: Inteligência artificial; Mercado financeiro; Bolsa de valores; Modelagem de dados; Computação – Teses; Inteligência Artificial – Teses; Mercado Financeiro – Teses; Bolsa de valores – Teses; Modelagem de Dados – Teses
• Link de acesso: http://hdl.handle.net/1843/32279

The use of autonomous trading agents in stock markets is becoming more common over time, although it is yet little explored in Brazil. With the large amount of available financial data, the construction of new trading models is feasible. The main objective of this work is to propose new financial indicators based on market micro-structure in order to create new automated investment strategies. To achieve this goal, a stock exchange simulator is built, a set of financial indicators based on market micro-structure and the principles for labelling prices series are defined and, finally, an autonomous trading agent is constructed. Firstly, the simulator, capable of reproduction of the orders and trades sent to the stock exchange of past trading days, is presented. Secondly, the set of financial indicators based on market micro-structure aspects and the necessary principles to label price series are defined. Lastly, the autonomous agent based on such indicators is built and, then, experimental validation is performed analysing financial metrics of the agent. In order to perform the experimental validation, data from future contracts of dollar of 2018 from B3 are chosen. The financial results obtained from the agent are evaluated in multiple scenarios varying parameters of network latency and operational costs. Such financial results show the potential of using micro-structure market data to construct automated investment strategies.