Reverse engineering trading positions of individual power plants from public data

We present a method for reverse engineering the unknown individual power plant positions from the known values of aggregated generation, power plant nameplate capacities and renewable forecasts by solving a quadratic integer program. We demonstrate the validity of our method for the British power market where sufficient data exists for validation. There, the transmission system operator publishes the values of indicated aggregated generation every half an hour, for delivery up to a few days in the future. Additionally, the operator publishes power plant physical parameters, and eventually power plant positions just before the delivery. Our method enables us to reconstruct individual power plant positions resulting from the day-ahead auction and to capture changes in positions during the intraday market. Applications of this method enable tracking of unusual activity on the market, the anticipation of transmission systemrelated actions, as well as estimating power plant running costs more accurately.