Comparações entre modelos vetoriais autorregressivos e redes neurais artificiais no mercado de ações brasileiro

Abstract

This study aimed to compare the predictive performance of Artificial Neural Network and Autoregressive Vectors methods applied to the price returns of three companies in the banking sector. The results showed that both models, seeking to minimize the error, produced forecasts close to zero. Thus, both models had similar predictions, with the Artificial Neural Network generally having a smaller error. However, both models showed limitations in generating accurate forecasts.