Distributional neural networks for electricity price forecasting
Abstract
We present a novel approach to probabilistic electricity price forecasting which utilizes distributional neural networks. The model structure is based on a deep neural network containing a so-called probability layer, i.e., the outputs of the network are parameters of the normal or Johnson's SU distribution. To validate our approach, we conduct a comprehensive forecasting study complemented by a realistic trading simulation with day-ahead electricity prices in the German market. The proposed distributional deep neural network outperforms state-of-the-art benchmarks by over 7% in terms of the continuous ranked probability score and by 8% in terms of the per-transaction profits. The obtained results not only emphasize the importance of higher moments when modeling volatile electricity prices, but also - given that probabilistic forecasting is the essence of risk management - provide important implications for managing portfolios in the power sector.
- Publication:
-
Energy Economics
- Pub Date:
- September 2023
- DOI:
- 10.1016/j.eneco.2023.106843
- arXiv:
- arXiv:2207.02832
- Bibcode:
- 2023EneEc.12506843M
- Keywords:
-
- C44;
- C45;
- C46;
- C22;
- C53;
- Q47;
- Quantitative Finance - Statistical Finance;
- Statistics - Applications;
- Statistics - Machine Learning
- E-Print:
- Enrgy Economics, 125 (2023) 106843