Balancing forecast errors in continuous-trade intraday markets

Abstract

Forecasting the output of photovoltaic (PV) and wind power systems inevitably implies inaccuracies, requiring balancing efforts prior to delivery. This paper takes the perspective of an operator who aims at compensating PV or wind power forecast errors in the continuous-trade intraday market. We combine a trade value concept with options valuation and dynamic programming to optimize volume and timing decisions of an individual power plant operator without market power. The model employs a multi-dimensional binomial lattice, with trade value maximized at every node to help formulating bids in view of correlated, uncertain production forecast and price patterns. Inspired by the German electricity market’s characteristics, we test the sensitivity of the model’s trade activities to changing parameters in 50 different scenarios. It shows that the model effectively outbalances price against volumetric risks. Trades are executed early and with large batch sizes in the case of price volatility. In contrast, increasing forecast error uncertainty leads to trade delays. High transaction costs trigger batch size reductions and ultimately further trade delays. Running 10,000 simulations across ten scenarios, we find that the model translates its flexible trade execution into a competitive advantage vis-a-vis static bidding strategy alternatives.