Previsão do preço de ações de empresas do setor de energia elétrica da [B]³ (Brasil, Bolsa, Balcão), por meio de modelos autoregressivos

Abstract

The electric power sector in Brazil stands as one of the fundamental cornerstones of the national economy and daily life. This study aims to analyze and make predictions using time series models on assets within the electric power sector of the Brasil, Bolsa, Balcão [B]³ for the first 10 weeks of 2023. The selection of the most appropriate model for the data will be determined through hypothesis testing and comparisons using model fitting criteria. The methodology for constructing time series models follows the principles of the Box e Jenkins theory. After definition the of ARIMA models (2,1,2) and ARIMA (3,1,4) models for the assets, opting for the logarithmic version of the original time series as the best approach. As a result, the empirical evidence found demonstrated that the best model adjustments to make efficient and reliable forecasts for the series were the ARIMA models.