Modeling KRX Bank's prices using time series analysis

Abstract

Using the AR, MA, ARMA, ARIMA, ARCH, and GARCH models, an appropriate model for given financial data comprised of the closing prices of the KRX Bank is found. An empirical approach that changes the values of p, d, and q to values between 0 and 3 is used. The basic assumptions are thoroughly checked using intuitive methods. In other words, the data are divided into five categories to show the stationarity in the variance. If the model is not stationary in terms of its variance, differencing or transformation of the given data is conducted. Subsequently, parameters are estimated using a maximum likelihood estimator and the most suitable model for the data is determined through several tests. Moreover, the relationships among the components in the KRX Bank data are considered using correlations and a principal component analysis to understand how they affect the data for the KRX Bank. In addition, another model for the data using regression with 10 items as multivariate variables is found.