Regularized quantile regression averaging for probabilistic electricity price forecasting

Abstract Quantile Regression Averaging (QRA) has sparked interest in the electricity price forecasting community after its unprecedented success Global Energy Forecasting Competition 2014, where top two winning teams track used variants of QRA. However, recent studies have reported method's vulnerability to low quality predictors when set regressors is larger than just a few. To address this issue, we consider regularized variant QRA, which utilizes Least Absolute Shrinkage and Selection Operator (LASSO) automatically select relevant regressors. We evaluate introduced technique – dubbed LASSO QRA or LQRA for short using datasets from Polish Nordic power markets. By comparing against number benchmarks, provide evidence superior predictive performance terms Kupiec test, pinball score test conditional accuracy, as well financial profits range trading strategies, especially regularization parameter selected ex-ante Bayesian Information Criterion (BIC). As such, offer an efficient tool that can be boost profitability energy activities, help with bidding day-ahead markets improve risk management practices sector.