Abstract
The aim of this contribution is to analyse historical daily closing quotes of the key price index of the Prague Stock Exchange from an econometric point of view. In greater detail, a particular class of discrete-time state space models appropriate for this type of univariate financial time series is introduced.
It combines various modelling con- cepts altogether; it joins a local level model, a linear ARMA process, and conditionally heteroscedastic innovations. The suggested modelling class is examined in different settings of parameters.
The final model is selected with respect to information and prediction criteria. Moreover, it is statistically verified and further investigated, i.e. checked for outliers or structural breaks.
Additionally, it is compared with other frequently applied models.