Electricity price forecasting across Norway's five bidding zones in the post-crisis era

2026-04-29 · Source: Takara TLDR - Daily AI Papers · Field: Energy & Utilities — Energy Markets & Policy, Renewable Energy Systems · Depth: Advanced, medium

Summary

A comprehensive evaluation of electricity price forecasting across Norway's five Nord Pool bidding zones reveals that the 2021-2022 energy crisis and increased integration with Continental Europe have significantly altered price formation, rendering older forecasting models unreliable. Researchers constructed a multimodal hourly dataset from 2019-2025 and tested eight forecasting model families, including LightGBM, ARX, and deep learning architectures, using rolling-origin backtesting and feature ablation. LightGBM consistently achieved the best performance across all zones, with Mean Absolute Error (MAE) ranging from 1.64 to 5.74 EUR/MWh. The ridge ARX model proved competitive in northern zones. While lagged prices and calendar variables alone often yield high accuracy, external features like reservoir levels and gas prices are crucial for stratifying forecast errors, especially under stressed market conditions.

Key takeaway

For AI Engineers developing electricity price forecasting models in volatile markets, you should prioritize LightGBM for its superior performance across diverse zones. While simple features like lagged prices are effective, integrate external data such as reservoir levels and gas prices to improve error stratification and model robustness, particularly when anticipating or operating under stressed market regimes. This approach enhances forecast reliability and supports better decision-making.

Key insights

LightGBM excels in Norwegian electricity price forecasting, but external factors are vital for accuracy under market stress.

Principles

• Hydropower dominance impacts electricity price formation.
• Market integration alters historical price patterns.
• Model interpretability aids decision-making.

Method

A multimodal hourly dataset (2019-2025) was used with rolling-origin backtesting, leave-one-group-out feature ablation, and conditional regime analysis to evaluate eight model families.

In practice

• LightGBM offers superior electricity price forecasting.
• Ridge ARX is a strong linear benchmark in northern zones.
• Lagged prices and calendar variables are highly effective features.

Topics

• Electricity Price Forecasting
• Norwegian Electricity Market
• Nord Pool Bidding Zones
• LightGBM
• ARX Models

Best for: AI Engineer, AI Scientist, Machine Learning Engineer, Research Scientist

Related on AIssential

• Electricity price forecasting across Norway's five bidding zones in the post-crisis era — LightGBM excels in Norwegian electricity price forecasting, but external features are crucial for understanding errors in stressed markets.
• AI-Powered Forecasting: How Prediction Markets Like Polymarket Are Enhancing Tech Trend Predictions — AI integration with prediction markets like Polymarket significantly enhances tech trend forecasting accuracy and speed.
• Investigating Calibration Challenges in Probabilistic Electricity Price Forecasting — Current electricity price forecasting prioritizes sharpness over calibration, yielding unreliable uncertainty estimates.
• Beyond Polarity: Multi-Dimensional LLM Sentiment Signals for WTI Crude Oil Futures Return Prediction — Multi-dimensional LLM sentiment, beyond polarity, improves WTI crude oil futures return prediction.
• The Hard Truth About Machine Learning for Amazon FBA Sellers — Amazon FBA demand forecasting with ML is challenging due to data hostility, requiring specialized techniques beyond standard approaches.
• Ensemble High-Resolution Weather Forecasting on AMD Instinct GPU Accelerators — Diffusion models enable robust ensemble weather forecasting by generating diverse atmospheric states to quantify uncertainty.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Takara TLDR - Daily AI Papers.