Researchers have developed an online algorithm for multivariate distributional regression to forecast electricity prices, addressing the underexplored multivariate nature of day-ahead prices. This method efficiently models conditional means, variances, and dependence structures, utilizing online coordinate descent and LASSO-type regularization for scalability in high-dimensional covariate spaces. A case study on German day-ahead market data demonstrated the approach's ability to yield interpretable and well-calibrated joint prediction intervals, with an open-source Python implementation available in the ondil package. AI
Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →
IMPACT Provides a novel statistical method for high-dimensional time-series forecasting, potentially improving market predictions.
RANK_REASON Academic paper on a statistical forecasting method with an open-source implementation.