A Novel Criterion of Electricity Price Forecast for Demand-side Responses Participating in the Electricity Market

This paper proposes a novel criterion for evaluating electricity price forecasting results for demand-side responses (DRs) who participate in the electricity market. Generally, the DR needs to predict the market price and arrange its bidding and operation schedule according to the forecast result. The mean-square-error (MSE) or the R-squared coefficient is used for evaluating the forecast result conventionally. However, it is shown in this paper that a forecast result with a good MSE or R-squared value is not necessarily more beneficial for the DRs. Instead, the proposed novel criterion can reflect the influence of different forecast results on the DR's market revenue and help the DR identify which forecast result is better regarding economic benefits. The proposed criterion emphasizes the accuracy of predicting the timing of the price peaks and dips, which is more crucial information to the DRs than the total numerical forecast precision over time. Results comparing different forecast methods on the clearing price of the day-ahead energy wholesale market in Japan, the JEPX spot market, are reported. The better forecast method for the DR is identified by the proposed criterion successfully.