![[2602.01157] Deep Time-Series Models Meet Volatility: Multi-Horizon Electricity Price Forecasting in the Australian National Electricity Market](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.01157] Deep Time-Series Models Meet Volatility: Multi-Horizon Electricity Price Forecasting in the Australian National Electricity Market
Summary
This paper explores the effectiveness of deep time-series models for forecasting electricity prices in the volatile Australian National Electricity Market, revealing gaps in current methodologies.
Why It Matters
Accurate electricity price forecasting is crucial for market stability and investment decisions. This study highlights the limitations of existing deep learning models in highly volatile conditions, emphasizing the need for improved forecasting strategies that account for market dynamics.
Key Takeaways
- Deep learning models often underperform in volatile electricity markets compared to standard baselines.
- Forecasting accuracy varies significantly across different intraday periods, particularly during price spikes.
- Existing models struggle with extreme price conditions, indicating a need for volatility-aware strategies.
Computer Science > Machine Learning arXiv:2602.01157 (cs) [Submitted on 1 Feb 2026 (v1), last revised 13 Feb 2026 (this version, v2)] Title:Deep Time-Series Models Meet Volatility: Multi-Horizon Electricity Price Forecasting in the Australian National Electricity Market Authors:Mohammed Osman Gani, Zhipeng He, Chun Ouyang, Sara Khalifa View a PDF of the paper titled Deep Time-Series Models Meet Volatility: Multi-Horizon Electricity Price Forecasting in the Australian National Electricity Market, by Mohammed Osman Gani and 3 other authors View PDF HTML (experimental) Abstract:Accurate electricity price forecasting (EPF) is increasingly difficult in markets characterised by extreme volatility, frequent price spikes, and rapid structural shifts. Deep learning (DL) has been increasingly adopted in EPF due to its ability to achieve high forecasting accuracy. Recently, state-of-the-art (SOTA) deep time-series models have demonstrated promising performance across general forecasting tasks. Yet, their effectiveness in highly volatile electricity markets remains underexplored. Moreover, existing EPF studies rarely assess how model accuracy varies across intraday periods, leaving model sensitivity to market conditions unexplored. To address these gaps, this paper proposes an EPF framework that systematically evaluates SOTA deep time-series models using a direct multi-horizon forecasting approach across day-ahead and two-day-ahead settings. We conduct a comprehensive empirical study ...