Distributional neural networks for electricity price forecasting

We present a novel approach to probabilistic electricity price forecasting which utilizes distributional neural networks. The model structure is based on deep network containing so-called probability layer, i.e., the outputs of are parameters normal or Johnson’s SU distribution. To validate our approach, we conduct comprehensive study complemented by realistic trading simulation with day-ahead prices in German market. proposed outperforms state-of-the-art benchmarks over 7% terms continuous ranked score and 8% per-transaction profits. obtained results not only emphasize importance higher moments when modeling volatile prices, but also – given that essence risk management provide important implications for managing portfolios power sector.