Sammendrag
This paper studies the Nordic electric-market relationship between one-week and one-month forward market prices versus realized average next week and next month spot prices. The forward products are traded at the Nordpool Financial Market and the spot-market prices are settled at Nordpool Spot. The microstructures of the forward and spot markets are used to evaluate a formal model based on actual reaction from new market information from the limited number of market participants. The objective is directional prediction ability applying an Artifical Neural Network (ANN) to predict next week and month spot market prices. From ANN directional accuracy short and long positions are traded at the Nordpool financial market for delivery in the spot market inducing risk. The forecasts are evaluated with both statistical and economic criteria. In terms of statistical criteria, the ANN model shows directional accuracy at 5% for most of the years and at the 1% level for the whole period. In terms of economic criteria, the paper shows a directional accuracy for short (majority) and long forward market position that is surprisingly high. For the whole period the weekly contracts show a success-ratio of 2.61 and for monthly contracts the success-ratio are as high as 7. The ANN prediction model therefore shows very promising results, inducing first that the spot energy market may inefficiently price electricity and second, application of the ANN-prediction model can be very profitable at the Nordpool Spot electricity market bearing some risk.
Vis fullstendig beskrivelse
