Forecasting energy markets using support vector machines

- We use a state-of-the art classification methodology.
- We test our model at the peak load period.
- We used an extensive set of explanatory variables according to literature.
- The generalization ability of the model was evaluated for a 200 day period.

In this paper we investigate the efficiency of a support vector machine (SVM)-based forecasting model for the next-day directional change of electricity prices. We first adjust the best autoregressive SVM model and then we enhance it with various related variables. The system is tested on the daily Phelix index of the German and Austrian control area of the European Energy Exchange (ΕΕΧ) wholesale electricity market. The forecast accuracy we achieved is 76.12% over a 200 day period.