Extended ARMA models for estimating price developments on day-ahead electricity markets

In this paper extended models for estimating price developments on electricity markets are presented. The models consider deviations from the normality hypothesis of the prices. Based on an ARMA model combination with GARCH, Gaussian-mixture and switching-regime approaches are comparatively discussed. The comparison is based on historic electricity prices of the spot and two reserve markets in Germany. It is shown that the proposed extended models lead to significantly improved representations of the considered stochastic price processes. It is inferred that these models may be preferred for estimating price developments on electricity markets.