A1719
Title: Multivariate probabilistic forecasting of electricity prices with trading applications
Authors: Alla Petukhina - HTW Berlin (Germany) [presenting]
Ilyas Agakishev - Humboldt University of Berlin (Germany)
Karel Kozmik - Charles University (Czech Republic)
Wolfgang Karl Haerdle - Humboldt University at Berlin (Germany)
Milos Kopa - Charles University (Czech Republic)
Abstract: A recently introduced approach is extended to probabilistic electricity price forecasting (EPF) utilizing distributional artificial neural networks, based on a regularized distributional multilayer perceptron (DMLP). This technique is developed for a multivariate case EPF with incorporated dependence. The performance of a fully connected architecture and an LSTM architecture of neural networks are tested. The empirical data application analyzes two day-ahead electricity auctions for the United Kingdom market. This creates the opportunity to buy in the first auction for a lower price and sell in the second for a higher price (or vice versa). Utilizing forecasting results, trading strategies are developed with various investors' objectives. It is found that, while DMLP shows similar performance compared to the benchmarks, the algorithm is considerably less computationally costly.