Using Machine Learning to Prevent Losses in the Brazilian Stock Market During the Covid-19 Pandemic

The Covid-19 Pandemic caused unprecedented changes in our society, going from human behavior modification areas to stock market crashes around the world. In fact, while the virus spread among several countries, investors watched great stock market losses during the period, especially in Brazil, in which the Ibovespa index presented a fall of almost 50% in its value. Thus, there is a need to investigate strategies to, at least, mitigate the losses during a stock market crisis, and, recently, Machine Learning techniques have played a key role in this process. Therefore, this work aims to propose an investigation of Machine Learning techniques in order to prevent stock market losses in the Brazilian stock market during the Covid-19 pandemic. We study commonly used algorithms in the financial area: Linear Regression, Support Vector Machines, Random Forest, XGBoost, Multilayer Perceptron, and an Ensemble composed by the combination of all of the mentioned models, all of them fed with historical stock prices and technical indicators. Our results show that, when properly tuned, some of the Machine Learning models could even bring a little profit during the Covid-19 pandemic, and, finally, we also present some guidelines for investors choice when considering investing in a future market crisis.