Forecasting of Frequency Containment Reserve Prices Using Econometric and Artificial Intelligence Approaches

The forecasting of control reserve prices is essential in order to participate reasonably in the auctions. Having identified a lack of related literature, we therefore deploy approaches based on auto-regressive and exogenous factors coming from econometrics and artificial intelligence and set up a forecasting framework. We use SARIMA and SARIMAX models as well as neural networks and forecast based on a rolling one-step forecast with re-estimation of the models. It turns out, that the combination of auto-regressive and exogenous factors yields the best results compared to approaches solely considering auto-regressive or exogenous factors. Further, the artificial intelligence approach outperforms the econometric approach in terms of forecast quality, whereas for the further use of the generated models, the econometric approach has advantages in terms of interpretability.