SYMARMA: a new dynamic model for temporal data on conditional symmetric distribution (Q148334)

The authors propose a new class of models under the assumption that the random component has conditionally a continuous symmetric distribution, while the dynamic component has the form of ARMA structure. A conditional maximum likelihood estimator of the unknown parameter is derived and an iterative procedure for the estimation is presented. Some simulation studies are given to present the performances of the obtained estimator. The application on the time series of excess return in the daily closing prices in some indexes is discussed.