Construção de modelo preditivo para investimento em bolsa de valores considerando três importantes ativos da Bovespa

Abstract

Forecasting financial time series is challenging and complex, being the target of several studies and applications of different techniques. An important element to be considered is the volatility that has as one of its main functions the measurement of the risk of financial instruments, being the GARCH model one of the most important tools for this form. The financial market is exposed to several factors that cause asset price fluctuation. As logistic regression is a technique used to describe the behavior between a binary dependent variable and independent variables, the modeling of Logistic Regression itself, together with the GARCH (Generalized Autoregressive Conditional Heteroskedasticity) model, will be the focus the study of this work, because the main challenge will be to predict the movement of the following day: whether positive or negative, of three shares of B3, Itaú, Petrobrás or Taesa. The results showed that Logistic Regression, selecting good predictor variables, proved to be an adequate model capable of overcoming the natural movement of the market, that is, it obtained a higher profitability than the "buy and hold" in the analyzed period.