A probabilistic forecast methodology for volatile electricity prices in the Australian National Electricity Market

TL;DR

This paper presents a probabilistic forecasting approach for highly volatile electricity prices in South Australia's NEM, using quantile regression and ensemble methods to improve accuracy over existing point forecasts.

Contribution

It introduces a novel ensemble forecasting framework with spike filtration and adaptive averaging, enhancing prediction accuracy in extreme volatility conditions.

Findings

Proposed model outperforms existing NEM point forecasts.

Ensemble averaging with varying training periods improves adaptability.

Quantile regression effectively captures price uncertainty.

Abstract

The South Australia region of the Australian National Electricity Market (NEM) displays some of the highest levels of price volatility observed in modern electricity markets. This paper outlines an approach to probabilistic forecasting under these extreme conditions, including spike filtration and several post-processing steps. We propose using quantile regression as an ensemble tool for probabilistic forecasting, with our combined forecasts achieving superior results compared to all constituent models. Within our ensemble framework, we demonstrate that averaging models with varying training length periods leads to a more adaptive model and increased prediction accuracy. The applicability of the final model is evaluated by comparing our median forecasts with the point forecasts available from the Australian NEM operator, with our model outperforming these NEM forecasts by a significant margin.