Variable renewable energy has a growing impact on electricity markets and power systems in many regions of the world. In this context, the 17th International Conference on the European Energy Market EEM20 set up a competition to develop probabilistic forecasting tools of wind production at a regional level. This paper proposes an adaptive approach for regional wind power forecasting. A physics-oriented preprocessing of the data delivers analog weather patterns and windpower-related variables, then a k-means clustering of wind farms further reduces the dimension of the problem. The generated representative features feed a Quantile Regression Forests model that produces sharp and reliable predictions. As a result, our model won the competition with a relative improvement of the average pinball loss of 6.7% and 14.7%, compared to the teams ranked second and third respectively.

In many countries, electricity prices on day-ahead auction markets result from a market clearing designed to maximize social welfare. For each hour of the day, the market price can be represented as the intersection of a supply and demand curve. Structural market models reflect this price formation mechanism and are widely used in prospective studies guiding long-term decisions (e.g. investments and market design). However, simulating the supply curve in these models proves challenging since estimating the sell orders it comprises (i.e. offer prices and corresponding quantities) typically requires formulating numerous techno-economic hypotheses about power system assets and the behaviors of market participants. In this paper we propose a method for the parameterization of sell orders associated with production units. The estimation algorithm for this parametrization makes it possible to mitigate the requirement for analytic formulation of all of the above-mentioned aspects and to take advantage of the ever-increasing volume of available data on power systems (e.g. technical and market data). Parametrized orders also offer the possibility to account for various factors in a modular fashion, such as the strategic behavior of market participants. The proposed approach is validated using data related to the French day-ahead market and power system, for the period from 2015 to 2018.

Anticipating electricity prices on the day-ahead market has become a key issue for both risk assessment and revenue optimization. In this paper, we propose to generate time series of prices with an hourly resolution using a structural model that simulates a simplified market clearing process. The aggregated supply curves in this model are composed of orders based on the available capacity of generation units. The ask prices are parametrized, and the calibration is performed by applying statistical learning to historical market and power system data. To reflect the strategic behavior of market participants, these prices depend on the scarcity of power at the national level. The model's performance is assessed based on the case of France with a one-year horizon and data from 2013-2015. This approach illustrates how open data on the electric power system enable links to be drawn between technical constraints and price formation. Index Terms-day-ahead markets, electricity prices, statistical learning, structural model The authors wish to thank the French Environment and Energy Management Agency (ADEME), the Association pour la Recherche et le Développement des Méthodes et Processus Industriels (ARMINES) and Coruscant SA for financially supporting this research.