Natural Gas Storage
As of the week ending July 17, 2026, the U.S. held 3,056 Bcf of working gas in underground storage — a weekly injection of 32 Bcf. That is 6.4% above the 5-year average for this week (2,871.2 Bcf) — close to normal. Versus the same week last year, storage is 19 Bcf lower.
Storage vs the 5-year average
The seasonal saw-tooth is the product: injections all summer, draws all winter. The question is never the level — it's the gap between the black line and the green one. Big deficits into November are what winter price spikes are made of.
Storage vs the gas price
The storage surplus/deficit (green, % vs the 5-year average) against natural gas — shown as UNG, the investable gas ETF, whose price includes the futures roll costs a real long position pays. The great winter squeezes (2021–22) launched from deficits; the 2015 and 2019–20 gluts ground the price down.
Does the storage report predict gas prices? We tested it
We joined all 603 weekly reports since January 2, 2015 to forward natural-gas returns (UNG, measured from the first close after each report's Thursday publication — no look-ahead). Two honest findings, one of them a null.
The weekly surprise has no directional edge. Reports that came in tighter than the seasonal norm did not precede above-average gas returns — in-line weeks actually did best, and both tails landed below them. If a "bullish storage surprise" tempts you to chase gas the next morning, the base rates say the move is already in the price by the close.
Deficits don't predict rallies — they enable squeezes. The only bucket that beat baseline was the deep-deficit quartile (storage more than 5.5% below its 5-year average): +1.14% over the next month vs -1.12% baseline. But the median deep-deficit outcome over the next quarter was -1.6% — the average is carried entirely by the right tail (best quarter +90.6%, worst -52.1%). A deficit is the precondition for a winter squeeze, not a forecast of one.
The baseline itself is the third lesson. Across all weeks, simply holding the gas ETF lost 3.52% per quarter on average — the structural futures-roll bleed. Every bucket should be read against that drag, and it is why "buy gas and wait" has been a losing default regardless of storage.
| Storage level vs 5-yr avg | Weeks | Next month | Win rate | Next quarter |
|---|---|---|---|---|
| Deep deficit (< -5.55%) | 150 | +1.14% | 51% | +1.03% |
| Mild deficit (-5.55% … 2.94%) | 151 | -2.54% | 41% | -5.8% |
| Mild surplus (2.94% … 9.22%)← now | 146 | -2.9% | 38% | -7.52% |
| Big surplus (≥ 9.22%) | 151 | -0.23% | 43% | -2.1% |
| All weeks (baseline) | 598 | -1.12% | 43% | -3.52% |
| Weekly report vs seasonal norm | Weeks | Next month | Win rate | Next quarter |
|---|---|---|---|---|
| Tighter than normal (< -12.8 Bcf) | 200 | -1.37% | 44% | -1.82% |
| In line (-12.8 … 14.6 Bcf)← now | 198 | +1.51% | 49% | -1.61% |
| Looser than normal (≥ 14.6 Bcf) | 200 | -3.48% | 38% | -7.06% |
Method notes: returns are UNG (investable; includes roll costs), entered at the first close on/after each report's publication Thursday; front-month NG futures dollar-moves agree on direction in 96% of weeks, so the findings are not an ETF artifact. "Seasonal norm" is the 5-year-average change for the same calendar week — we have no analyst-consensus history, so this measures surprise vs seasonality, not vs the survey. Bucket edges are full-sample terciles/quartiles, disclosed above; forward windows overlap week to week, which overstates the independence of the samples. Base rates, not advice.
Why traders and investors watch this
This is the most directly tradable series in the Energy section: natural gas futures reprice on the Thursday 10:30am ET print within seconds, and the injection or draw versus expectations is the weekly surprise the whole gas complex trades around. The surplus or deficit against the 5-year average is the market's anchor for winter risk — deficits heading into November are the historical setup behind gas price spikes, which flow through to gas-weighted producers, utility fuel costs, and CPI energy with a lag. Note the direction of causation for equities: the same reading cuts opposite ways for producers (deficit = pricing power) and gas-burning utilities (deficit = cost pressure). And because "storage moves gas" is the sector's most repeated claim, we tested it — the 600-week study below shows what the report actually preceded, including the null.
How this is computed — and what it can't tell you
EIA surveys underground storage operators weekly (Form EIA-912) and publishes every Thursday at 10:30am ET — the one scheduled report in energy that natural gas futures visibly trade on. We take the lower-48 total and compute, for each week, the average, minimum and maximum of the same calendar week across the prior five years. The chart starts in 2015, the first year with a full 5-year window behind it; underlying weekly data runs from 2010 (pre-Nov-2015 figures are EIA's own derived estimates from the older regional structure).
Limits worth knowing: storage is one supply buffer, not the whole balance — LNG exports, production, and weather all move prices independently. And the "5-year average" itself drifts: demand has grown, so a percentage vs the average is a cleaner read than the raw Bcf gap.
Sources, methodology & freshnessLast updated 2026-07-17 · Open ↓Close ↑
thetrading.tools PRO
Talk to this data with your AI
PRO is an MCP server that plugs Claude, ChatGPT, Cursor or any AI agent straight into everything on this site — the live market scores, ~90 economic indicators with full histories, and every signal study. Ask a market question; your AI answers from our data, dated and sourced.
Cancel anytime
How Natural Gas Storage Works
- 1The one scheduled event in energyEvery Thursday at 10:30am ET, the EIA publishes how much working gas sits in underground storage across the lower 48 (Form EIA-912). It is the closest thing natural gas has to a CPI print — futures reprice on the number within seconds, and the weekly injection/draw vs expectations is the tradable surprise.
- 2Level vs the 5-year seasonal normThe raw level means nothing without the season — storage fills all summer and drains all winter by design. So for each week we compute the average, minimum and maximum of the same calendar week across the prior five years. The surplus or deficit vs that average is the number the market actually anchors on.
- 3Percentage beats BcfDemand has grown structurally (LNG exports, power burn), so a fixed Bcf gap shrinks in importance over time. We express the surplus/deficit as a percentage of the 5-year average — the cleaner cross-year comparison.
- 4Full history, honest windowsWeekly data runs from 2010; the chart starts in 2015, the first year with a complete 5-year window behind it. Pre-Nov-2015 figures are EIA's own derived estimates from the older regional structure, stated rather than hidden.