GFS vs ECMWF Demand-Hub Outlook: 2026-11-05 12:00 UTC

Executive Summary

The 12Z model cycle presents a cooler divergence, with the ECMWF projecting a more aggressive early-season heating load than the GFS across the 15-day window. Nationally, the ECMWF delivers 211.04 weighted cumulative heating degree days (HDDs) compared to the GFS at 196.93 HDDs, reflecting a +14.11 HDD spread driven primarily by stronger cold-air retention across the Eastern and Southern hubs. Furthermore, the ECMWF trended notably colder run-over-run, adding +18.28 weighted cumulative HDDs relative to its 00Z cycle.

Hub-Level Weather

Demand is heavily anchored by the Midwest and Northeast/Mid-Atlantic population corridors, which account for 75% of the aggregate index:

• Midwest (Chicago, 35% weight): ECMWF totals 281.49 HDDs versus 270.52 HDDs on the GFS (mean daily absolute divergence of 3.66 HDDs). While both models project sharp cold arriving around November 13–15 (daily HDDs peaking near 26), they disagree significantly on timing and duration: ECMWF maintains colder conditions November 09–10 (17.45 and 12.13 HDDs vs. GFS 9.96 and 7.82 HDDs) and re-intensifies cold on November 18–19 (20.60 HDDs vs. GFS 12.40 HDDs), whereas the GFS features a mid-period cold spike on November 16–17.
• East (New York City, 30% weight): ECMWF exhibits a distinctly colder bias, forecasting 235.51 cumulative HDDs against the GFS at 212.40 HDDs (+23.11 HDD gap; 2.54 mean absolute daily difference). Key divergence windows occur on November 11 (ECMWF 13.26 HDDs vs. GFS 7.60 HDDs), November 14 (ECMWF 25.95 HDDs vs. GFS 20.71 HDDs), and November 19 (ECMWF 24.19 HDDs vs. GFS 15.30 HDDs).
• South (Atlanta, 10% weight): Atlanta reflects the widest localized dispersion, with ECMWF estimating 160.46 HDDs versus the GFS at 116.80 HDDs (mean daily absolute difference of 3.92 HDDs). ECMWF drives substantial cold air southward between November 08–16 and on November 19 (16.54 HDDs vs. GFS 7.94 HDDs).
• Northeast (Boston, 10% weight): Boston is the lone hub where GFS projects higher heating load (253.92 HDDs vs. ECMWF 247.56 HDDs), though both converge closely on peak cold during November 14–17.
• South Central (Houston, 15% weight): Space heating needs remain minimal (9.71 HDDs on GFS vs. 7.06 HDDs on ECMWF), with lingering mild temperatures generating marginal cooling load (45.04 CDDs on GFS vs. 40.94 CDDs on ECMWF).

Demand Implications (Natural Gas)

The colder 15-day evolution points to an acceleration of early-winter residential and commercial (res/comm) space-heating burn, supportive of front-month Natural Gas (NG) demand.

With both models showing cumulative weighted CDDs dropping into low single digits (GFS 7.76 CDDs vs. ECMWF 6.89 CDDs), power-sector cooling burn has functionally concluded for the season. Demand will be entirely dictated by heating load in the Midwest and East. If the colder ECMWF path verifies—reinforced by its +18.28 HDD upward shift—aggregate heating demand will push well above seasonal normals heading into mid-November, expediting initial net storage withdrawals. Conversely, the GFS solution provides intermittent warm breaks (notably November 09–11 and November 18–19) that would keep baseline heating demand more tempered.

Model Agreement & Caveats

• Convergence Score: The aggregate convergence score stands at 0.2693, accompanied by a weighted mean absolute daily HDD difference of 2.71 HDDs. This relatively low agreement score indicates substantial synoptic timing uncertainty beyond Day 6 (November 11 onward), particularly regarding the depth and longevity of Arctic air masses penetrating the Midwest and Southeast.
• Forecast Reliability: Given the low convergence score and high day-to-day variance in Day 8–15 projections, market participants should avoid fully leaning into the cold ECMWF extreme until consecutive cycles confirm spatial alignment with ensemble means.
• Data Caveats: Open-Meteo point extractions utilize gridded numerical weather prediction (NWP) outputs (NCEP GFS 0.13° and ECMWF IFS 0.25°). Point-location sampling does not capture localized microclimates or suburban heating demand variations within large metropolitan areas, nor does raw degree-day calculation account for wind chill and cloud cover impacts on actual pipeline pull.