Forecasting Power Prices Using a Hybrid Fundamental-Econometric Model

Abstract

We investigate the performance of two hybrid forecasting models to predict the next-day base load electricity prices on the APX power exchange for Great Britain. Hybrid models have often been advocated as a synthetic method for forecasting, but it is an open question how well they perform at the day-ahead stage. A conventional hybrid approach combines a fundamental model, formulated with supply stack modeling, with an econometric model using data on price drivers. We extend this model to one based upon logistic smooth transition regression, which represents a regime-switching for periods of structural change. Empirical results for out-of-sample forecasting using both hybrid approaches are discussed and compared to non-hybrid time series forecasting methods.