MATSUMOTO Lab - Electricity spot price forecast (en)

Electricity spot price forecast

We update our daily forecasts of electricity prices (JEPX spot prices). Using six different methods (including machine learning algorithms) we generate forecasts and combine them through an advanced online learning ensemble approach.

Based on weekly weather forecasts from the Japan Meteorological Agency, we produce price predictions up to 7 days ahead and evaluate their accuracy.

For related charts and data, including electricity market trading prices, grid information, weather conditions, and imbalance prices, please refer to the relevant pages.

Data sources:

Japan Electric Power Exchange (JEPX): http://www.jepx.org
Japan Meteorological Agency: https://www.jma.go.jp
OCCTO Wide-area Reserve Margin Disclosure: https://web-kohyo.occto.or.jp/kks-web-public/

Forecast data download (JSON format)

Access to Forecast Data via API

The public API above currently provides day-ahead forecasts for four areas (Tokyo, Kansai, Hokkaido, and Kyushu) free of charge.

For companies, research institutions, public organizations, and other entities that wish to continuously access forecast data up to six days ahead, API keys are issued free of charge on an organization-by-organization basis. If you are interested, please contact us using the form below.

Registered users can also access selected results and validation tools for a new forecasting model covering all nine JEPX areas, in addition to the existing API data.

Real-time forecasts from the new model
For the four currently supported areas (Tokyo, Kansai, Hokkaido, and Kyushu), forecasts up to six days ahead can be viewed together with probabilistic forecasts at the q1, q5, q25, q50, q75, q95, and q99 quantiles.
Historical forecast accuracy evaluation
Forecast errors (MAE) from January 2025 onward can be examined for all nine areas over user-selected periods, including monthly and time-of-day analyses.
For the four areas covered by the existing model, forecast errors from the existing and new models can also be compared directly. In evaluations using the same periods and samples, the new model reduced MAE by approximately 17% on average across the four areas compared with the existing model.

Continuous API access to forecast data from the new nine-area model, including probabilistic forecasts, is planned to be provided to companies and other organizations through a Data Transfer Agreement (DTA) under the industry–academia collaboration framework of Institute of Science Tokyo. If you are interested in such access, please contact us using the form below.

API keys are, in principle, issued on an organization-by-organization basis. We may not be able to respond to applications submitted anonymously, under an individual name, or without a clearly identifiable organizational affiliation.

API Key Request Form: https://forms.gle/qxkCBWXHSVJvxNp9A

▼ About the Forecasting Model (Click to Expand)

■ Model Overview

We forecast electricity prices at 30-minute intervals using an ensemble approach that combines six different methods: a Generalized Additive Model (GAM), regularized regression (Ridge regression), and four machine learning algorithms: k-nearest neighbors (kNN), artificial neural networks (ANN), support vector regression (SVR), and random forest (RF).

The GAM captures long-term trends through smooth seasonal patterns such as annual cycles, while the other models (Ridge, kNN, ANN, SVR, and RF) are designed to capture short-term fluctuations.

Our ensemble forecasting approach is based on an online machine learning technique called Multivariate Probabilistic CRPS Learning (Berrisch and Ziel, 2023, 2024). Due to the online nature of this learning method, the weights assigned to individual forecasting agents (i.e., models) are updated daily, giving more importance to those that have shown higher accuracy in recent predictions.

For instance, the GAM receives higher weight when long-term trends dominate, while machine learning models take precedence during periods of strong short-term weather sensitivity. This dynamic weighting scheme allows for overall more accurate predictions compared to any single model. Model performance is visualized using cumulative Mean Absolute Error (MAE) comparisons.

For daily average price forecasting, we apply a quantile regression approach, specifically a quantile generalized additive model (qGAM), to generate probabilistic forecasts (i.e., percentile forecasts). Since predictions of higher and lower percentiles are particularly sensitive to the input data, we estimate separate models for short- and long-term variations and combine them to ensure robustness.

The relevant papers are listed below.