Hyperion - Client

U.S. Natural Gas Weekly Overview — October 23, 2026

Written by Robert Vaughn | Oct 23, 2026, 6:45:01 PM

Prepared from SynMax storage-call tracking, Hyperion S/D and storage balances, EIA weekly storage, EIA-930 generation, and GFS weather. Figures refresh each Friday.

Last Week Recap

The EIA reported a +77 Bcf injection for the week ending October 16, 2026. SynMax had modeled +88 Bcf, an over-call of 11 Bcf. That is the season's largest miss and only the third call since July to fall outside the ±5 Bcf band. Due to the larger than normal miss, we are currently looking into how our power burn models utilize the EIA 930 dataset. Early tests of the modeling changes point to roughly 0.5 Bcf/d increase in our power burn numbers for the week. This would have put our injection call close to 84-85 Bcf; still a large miss but closer to the market's range of expectations. Furthermore, it’s worth noting that the changes we’re testing would slightly improve our weekly S/D balances relative to the EIA’s posted storage numbers in our historical dataset.

Seasonal scorecard: Since July 3, the model's mean absolute error is 3.3 Bcf with a bias of +0.8 Bcf (13 of 16 calls within ±5 Bcf). The last two prints widened what had been an essentially unbiased record. We are watching whether the over-call persists as heating demand builds.

Working gas inventories now stand at 3,644 Bcf as of October 16. That is 42 Bcf (1.2%) above the five-year average but 154 Bcf (4.1%) below last year's level. Looking ahead, the next storage weeks are modeled at +71 Bcf (week ending Oct 23) and +75 Bcf (Oct 30), which would lift inventories to roughly 3,790 Bcf by October 30. A further +53 Bcf is modeled for the week ending Nov 6 (~3,843 Bcf).

Week ending October 23

Call: +71 Bcf, slightly below the five-year average build for the week (+73 Bcf) and last year's +74 Bcf.

  • Weather: Population-weighted total degree days for the week rise to 67.0 (from 56.9), essentially in line with last year (68.1). Heating load is starting to arrive: residential and commercial demand rose a combined 4.6 Bcf/d week on week.
  • Power burn: Electric-power gas demand fell 3.5 Bcf/d to 30.9 Bcf/d. That was the largest single loosening item and offset most of the heating gain.
  • Supply & exports: Dry-gas production rose 1.5 Bcf/d to 111.2 Bcf/d. LNG feedgas rose 0.7 Bcf/d to 20.5 Bcf/d.
  • Net: The daily balance tightened by 2.4 Bcf/d, from 12.6 to 10.2 Bcf/d of implied injection.

Next week (ending Oct 30): Degree days ease to 62.4, well below last year's 90.4, when an early cold snap cut the build to +33 Bcf. That sets up a +75 Bcf injection against a five-year average of ~+70. The year-on-year deficit should narrow sharply, by about 40 Bcf, in that report.

Renewables and Gas Burn

  • Holding demand fixed, each GW of wind and solar has replaced ~0.68 GW of gas-fired generation. That comes from a daily US48 regression since mid-September (gas generation vs total load and wind+solar output; R² = 0.99). At typical heat rates, a sustained 10 GW wind surge removes roughly 1.2 Bcf/d of gas burn.
  • Week ending Oct 16: wind did the work. Wind output jumped 26 GW to a 62 GW weekly average, its highest since mid-September. Gas generation fell 19 GW even though total load was essentially flat (−2 GW). Gas's share of generation dropped to 40% from 44%, and wind/solar/hydro rose to ~25%.
  • Implication: With cooling load gone, wind variability is now the main swing factor for power burn through November. A windy week can take 1–2 Bcf/d out of power demand, about the same size as a modest temperature miss. That makes wind forecasts a key input to the near-term storage calls.

LNG

Total U.S. LNG feedgas averaged 20.5 Bcf/d for the week (up from 19.6 Bcf/d) and peaked near 21.1 Bcf/d on October 18, a record-high range. Cove Point is fully back after its Sep 20 – Oct 11 turnaround. Golden Pass continues to commission at roughly 0.8 Bcf/d and remains the key watch item for the winter LNG call.

Regional Storage Divergence

The national surplus continues to mask a sharp regional split. The Lower 48 sits 42 Bcf above the five-year average. The South Central salt tier, however, is 43 Bcf (14.7%) below its five-year average and 22% below last year heading into winter. Salt is the highest-deliverability storage and does the heavy lifting in cold snaps, so the deficit matters more for winter reliability and Gulf basis risk than the headline surplus suggests.

Region Stocks (Bcf) vs 5-yr avg vs last year
East 900 +26 (+3.0%) +1 (+0.1%)
Midwest 1,065 +23 (+2.2%) +5 (+0.5%)
Mountain 257 +14 (+5.6%) −15 (−5.5%)
Pacific 297 +18 (+6.5%) −14 (−4.5%)
South Central — nonsalt 875 +5 (+0.6%) −60 (−6.4%)
South Central — salt 251 −43 (−14.7%) −69 (−21.6%)
Total Lower 48 3,644 +42 (+1.2%) −154 (−4.1%)

Long-Term Storage Outlook

End-of-season baseline levels and modeled peaks/troughs:

Season endpoint Baseline level Modeled peak / trough
End-injection Oct 31, 2026 3,802 Bcf peak 3,888 Bcf @ Nov 17, 2026
End-withdrawal Mar 31, 2027 2,124 Bcf trough 2,077 Bcf @ Mar 19, 2027
End-injection Oct 31, 2027 3,845 Bcf —

SynMax's Baseline storage scenario uses the short-term GEFS forecast, followed by the average demand across the ten most recent weather years. Averaging demand outcomes, rather than temperatures, preserves the weight of cold snaps and heat waves.

Year-over-year S/D for the withdrawal season (Nov–Mar 2026/27 vs 2025/26): Supply growth continues to outpace demand. Dry-gas production is +4.9 Bcf/d year over year, only partly offset by higher LNG feedgas (+2.1 Bcf/d; Plaquemines, Corpus Christi Stage 3, Golden Pass). Heating demand is little changed. On net, the implied daily balance loosens by about 2.4 Bcf/d versus last winter (−11.1 Bcf/d average net draw vs −13.5 Bcf/d).

Source Dashboards