GFS vs ECMWF Demand-Hub Outlook: 2026-05-18T00:00Z

Executive Summary

The 15-day outlook shows a significant bullish divergence, with the GFS model projecting materially stronger weather-driven gas demand than the ECMWF. This difference is driven by the GFS forecasting both higher cooling demand (CDDs) in the South and more late-season heating demand (HDDs) in the East. Overall forecast confidence is low, reflected in a convergence score of 0.48, warranting caution.

Hub-Level Weather

The national demand-weighted picture shows the GFS forecasting 115.3 total CDDs and 39.8 HDDs, compared to the ECMWF's 89.9 CDDs and 34.5 HDDs. The primary drivers of this model disagreement are:

• South Central (Houston): GFS is substantially hotter, projecting 234.5 cumulative CDDs versus just 170.6 from the ECMWF. This is the largest single point of CDD divergence.
• East (New York City): GFS is notably colder, forecasting 54.1 cumulative HDDs compared to the ECMWF's 38.6 HDDs, indicating more significant late-season heating needs.
• Midwest (Chicago): The highest weighted hub (0.35) also shows a warmer GFS solution, with 80.1 CDDs versus the ECMWF's 64.7 CDDs.

Demand Implications (Natural Gas)

The GFS solution implies a stronger call on natural gas for power generation to meet cooling load, primarily in Texas and the Southeast. Simultaneously, its colder Eastern forecast would support higher residential/commercial demand than the ECMWF.

Compared to the prior model cycle, the GFS trended significantly colder, adding +8.4 weighted HDDs. This run-over-run shift adds to the bullish demand risk and uncertainty in the forecast. The ECMWF remained more stable, with only a minor change (+2.1 weighted HDDs).

Model Agreement & Caveats

Forecast confidence is low. The model convergence score of 0.48 indicates substantial disagreement on the atmospheric pattern, particularly in the 8-15 day period. The GFS's forecast for stronger southern heat and a cooler Eastern trough is a less consistent solution at this time.

Traders should treat the GFS's higher demand forecast with skepticism until subsequent model runs show better agreement with the ECMWF. The current low confidence suggests volatility is likely as models attempt to resolve this pattern.

Data is sourced from Open-Meteo for the specified forecast initialization time.