Estratégia automatizada de decisão multicritério no mercado financeiro

Abstract

Investment automation has been a challenge since the beginning of stock markets. With the evolution of computational power, this dream is getting closer to reality. In this context, this paper investigates, develops and applies some computational techniques to automate investments. It was proposed a backtesting tool, with strategy based on multiple technical indicators. The goal with the indicators was to create a committee, responsible for the decision-making using signals as output, such as buy, hold or sell. This approach is made in a way that removes some psychological aspects of human traders that have significant impact on the decision making process under uncertainty conditions. Similar to other approaches it uses technical indicators, however it is different from usual approaches, this method focuses on usage for day trade operation of mini future contracts of index Bovespa, with candles of 5 minutes frequency. Considering the data used, from 2015 to 2021, the strategy used was not able to overcome the reference model used, the classic buy & hold. Even though, the proposed tool seems to be relevant (i) to improve the data acquisition process, that can be challenging depending of the equity and frequency; (ii) for data analysis, with customs metrics and visualizations; (iii) and for optimization and validation of custom and complex strategies, using python, that could be more challenging to implement using MQL5 on Metatrader.