Forecasting the stock market using ARIMA and ARCH/GARCH approaches

Abstract

Forecasting stock returns forecasting is a crucially important topic in the study of finance, econometrics, and academic studies, and involves an in-depth study on time series. This thesis aims to examine the most representative companies on the São Paulo Stock Exchange, and based on that data, predict the behavior of future stock returns using several different forecasting methods. In time series analysis, ARIMA models are used in many situations and usually present good results; nevertheless, to determine which model best suits the data, others must be tested. When considering the high volatility of the data and factoring in the economic situation of the country that is being analyzed, other techniques must be considered, especially the ARCH family ones. Those techniques are primarily used to predict data involving Stock Markets worldwide. An accurate prediction can bring advantages for the companies who make those predictions and benefit the stakeholders directly since it provides enough information to make better decisions towards the future.