Every morning, a new 16-day temperature forecast re-prices US heating and cooling demand. Every Thursday, the EIA reports whether storage kept up. Every Friday, the COT shows who was positioned for it. Energy Market Intelligence runs that loop daily — with demand weighted by where people actually live, not where the weather stations are.
Explore the Demo → Current gas signal →Natural gas demand is not an estimate — it is a physical response. When the temperature drops below 65°F, furnaces switch on; above it, air conditioners do. Degree days translate that response into a number, and the number moves billions of dollars of storage and futures positioning every week. The entire market runs on a forecast that resets every morning — which is why our energy report is the only one we update daily.
But a degree day computed in the wrong place is worse than none at all. A frigid morning in a sparsely populated county moves no meaningful demand; two degrees of cold in the corridor between Boston and Washington moves a lot. That is why we weight every forecast grid point by population before it becomes a demand number — and score it against normals built the same way, so a deviation is always apples to apples. When the 16-day outlook shifts, how many of those degree days landed where the people are?
Storage is the market's memory of every forecast that already verified. Each Thursday's EIA print settles last week's argument and starts the next one: is the surplus to the 5-year average growing because supply is strong, or because the weather never showed up? Crude runs on a different engine — global balances, OPEC, refining — but the positioning discipline is the same, and the COT book tells you who is leaning which way in both.
Every SoftSignal market is covered the same way: who holds the positions, what the balance sheet says, what conditions are doing, and what the climate background implies. For energy, each layer looks like this.
This is the raw material of the gas market: heating demand peaks in deep winter, cooling demand in high summer, and the shoulder months in between are where storage gets rebuilt — or doesn't.
| Month | Heating degree days | Cooling degree days |
|---|---|---|
| January | 27.8 | 0.1 |
| February | 25.2 | 0.2 |
| March | 18.5 | 0.5 |
| April | 10.9 | 1.5 |
| May | 4.5 | 4.1 |
| June | 1.1 | 8.6 |
| July | 0.2 | 12.1 |
| August | 0.2 | 11.2 |
| September | 1.6 | 6.6 |
| October | 7.7 | 2.2 |
| November | 16.9 | 0.5 |
| December | 25.0 | 0.2 |
Unweighted average across all 91 grid points, shown here for seasonal shape. The report itself weights each point by population within its region — the methodology that separates demand that matters from weather that doesn't. How the weighting works →
The storage year: gas is injected April through October and withdrawn November through March. The shoulder pivots are where end-of-season storage math gets decided — and where forecast surprises carry the most leverage per degree.
The same report URL updates in place all week — the link from Monday's email shows Friday's data by Friday night. ENSO context refreshes monthly.
The core read, updated every morning: nine-region demand forecast with normals deviations, EIA storage context, Henry Hub fundamentals, and COT positioning for the gas and oil complex.
See it in the demo →WTI crude and RBOB positioning in cross-market context, alongside gold, silver, and copper — for reading energy flows against the broader macro book.
See it in the demo →The public one-page read: managed-money regime, week-over-week change, and demand-forecast stress — updated after every CFTC release, with an RSS feed.
Current gas signal →The same one-page treatment for the crude book — positioning percentile and weekly flow at a glance.
Current WTI signal →The full report — this morning's demand forecast, storage context, and the latest positioning — is in the open demo. No email, no card.