Czech Stock Market Analysis

Abstract

The analysis of relative returns of selected stocks at Prague Stock Exchange has been performed. As a rule, the kurtosis of the return distribution was greater than that of the standard normal distribution. Second, Box-Jenkins ARIMA models have been employed for return time series modelling. Instead of autoregressive terms of first order, higher terms are also necessary (particularly tenth order). Third, ARCH and GARCH models allowing for conditional heteroskedasticity provide closer approximation of return time series. In most cases, GARCH (1,1) is quite satisfactory.