Benchmarking Pre-Trained Time Series Models for Electricity Price Forecasting

TL;DR

This paper benchmarks various pre-trained time series models for electricity price forecasting, comparing their performance against traditional statistical and machine learning methods across multiple European countries.

Contribution

It provides a comprehensive evaluation of state-of-the-art TSFMs in the context of electricity price forecasting, highlighting the effectiveness of traditional models like MSTL.

Findings

Chronos-Bolt and Time-MoE perform comparably to traditional models

The MSTL model consistently outperforms TSFMs across countries

No TSFM statistically outperforms the biseasonal MSTL model

Abstract

Accurate electricity price forecasting (EPF) is crucial for effective decision-making in power trading on the spot market. While recent advances in generative artificial intelligence (GenAI) and pre-trained large language models (LLMs) have inspired the development of numerous time series foundation models (TSFMs) for time series forecasting, their effectiveness in EPF remains uncertain. To address this gap, we benchmark several state-of-the-art pretrained models--Chronos-Bolt, Chronos-T5, TimesFM, Moirai, Time-MoE, and TimeGPT--against established statistical and machine learning (ML) methods for EPF. Using 2024 day-ahead auction (DAA) electricity prices from Germany, France, the Netherlands, Austria, and Belgium, we generate daily forecasts with a one-day horizon. Chronos-Bolt and Time-MoE emerge as the strongest among the TSFMs, performing on par with traditional models. However, the biseasonal MSTL model, which captures daily and weekly seasonality, stands out for its consistent performance across countries and evaluation metrics, with no TSFM statistically outperforming it.