Comparative study of ARIMA methods for forecasting time series of the mexican stock exchange

Abstract

Predicting volatility in stock market price indices is a major economic problem. The idea of forecasting time series is that the patterns associated with past values in a data series can be used to project future values. The study of volatility can be applied to solving these economic problems, because volatility allows measuring the risk of asset portfolios, since it shows the behavior of the variation of asset prices. In order to be able to predict effectively the future behavior of a time series, it is necessary to know the attributes of the series with the correct prediction method and thus to be able to define training patterns. The accurate selection of the attributes evaluated in a time series defines the impact on prediction accuracy. In this work the study of kurtosis and the comparison between different ARIMA methods for the solution of time series of the Mexican Stock Exchange and the Makridakis contests are shown.