Surge

Surge provides open probabilistic day-ahead load forecasts for the US power grid, covering all 53 balancing authorities that publish demand data to the EIA-930. The platform is designed for research and reference use, offering insights into how much electricity is expected to be consumed in each region over the next 24 hours. These forecasts are generated using the Chronos-2 model, which has been fine-tuned on seven years of public data to achieve accuracy comparable to utility-internal systems.

The tool presents its forecasts through an interactive map, allowing users to select a region visually or from a dropdown menu. For each balancing authority, Surge displays a probability range rather than a single forecasted value. This includes the median forecast and a shaded band representing the p10–p90 range, which covers the middle 80% of likely outcomes. The platform also contextualizes forecasted demand by showing how close it is to the authority's all-time peak, helping users gauge potential grid stress.

Surge includes additional metrics such as the Model Accuracy Score (MASE), which compares the model's performance to a naive baseline forecast. The tool is free to use and is described as open-source, with references to its code repositories provided. It is explicitly not intended for trading, regulated bidding, or decisions requiring bankability-grade data, positioning it as a resource for analysis and exploration rather than operational use.

By offering open access to high-quality load forecasts that match the accuracy of expensive proprietary models, Surge serves energy analysts, researchers, and others interested in the dynamics of the US electricity grid.