Day-ahead Electricity Price Forecasting by Iterative Model Selections

This paper provides iterative model selection for forecasting day{ahead hourly electricity prices, while accounting for fundamental drivers. The iterative procedure is based on the automatisation of the forecasting process, by allowing for switching across several model specifications. Forecasts of demand, infeed from renewable energy sources, traditional fossil fuel prices, and physical flows are all included in linear and nonlinear specifications, ranging in the class of ARFIMA-GARCH models. Results support the adopting a flexible structure that is able to adapt to market conditions. Predictions, made for the northern Italian hourly electricity prices and compared by using the Diebold-Mariano test and the Model Condence Set, indicate a strong predictive power from forecast demand at any hour and from RES mainly at peak hours, as well as a non-diminishing role of natural gas and CO2 prices, and a high level of signicance of electricity weighted inflows, especially during the morning hours.