Modelling of Control Reserve Prices with Econometrics and Artificial Intelligence

In Germany, Transmission System Operators (TSOs) procure different reserve power products meeting different quality requirements through tenders. This leads to market segments for primary, secondary (automatic Frequency Restoration Reserve, aFRR) and tertiary control reserve power within the market for ancillary services. Besides reserve power markets, potential suppliers face several opportunities to market their flexible generation capacity. Therefore, forecasting of prices in several markets is crucial for market participants, for both taking the trading decision and placing optimal bids.
This study focuses on the German aFRR market, which is the largest European one and consists of a two-part pay-as-bid auction for each 4-h-slice of the day. The modeling and forecasting of prices of ancillary services markets has been comparatively rare in the literature. We therefore explore methods to forecast aFRR prices and to identify different bidding strategies.
Most of price forecasting models originate from either approaches based on time series analysis or from artificial intelligence or machine learning approaches. We apply SARIMA as well as regression models to aFRR price forecasting. ... mehrThe presence of intermittent spikes and nonlinearities makes the prediction very challenging.