Improving cross-temporal forecasts reconciliation accuracy and utility in energy market

Abstract

Wind power forecasting is essential for managing daily operations at wind farms and enabling market operators to manage power uncertainty effectively in demand planning. Traditional reconciliation methods rely on in-sample errors for forecast reconciliation, which may not generalize well to future performance. Additionally, conventional aggregation structures do not always align with the decision-making requirements in practice, and evaluation metrics often neglect the economic impact of forecast errors. To address these challenges, this paper explores advanced cross-temporal forecasting models and their potential to enhance forecasting accuracy and decisions. First, we propose a novel approach that leverages validation errors, rather than traditional in-sample errors, for covariance matrix estimation and forecast reconciliation. Second, we introduce decision-based aggregation levels for forecasting and reconciliation, where certain horizons are tailored to the specific decisions required in operational settings. Third, we assess model performance not only by traditional accuracy metrics but also by their ability to reduce decision costs, such as penalties in ancillary services. Our results show that using validation errors improves the accuracy by more than 7% across different temporal levels. We also demonstrate that statistical-based hierarchies tend to adopt less conservative forecasts and reduce revenue losses. On the other hand, decision-based reconciliation offers a more balanced compromise between accuracy and decision cost, while saving computational time by 2%–3% for simpler models and up to 93% for more advanced models, making them attractive for practical use.