Storage response to percentile Tier 1

API name: storage_percentile · Source: EIA

How full a natural gas storage region is relative to its own 5-year seasonal average, tested against forward Henry Hub returns. Doesn't clear the strictest scorecard tier alone (81% sign consistency vs. an 85% bar), but is the single strongest, most cross-family-robust predictor we've found - confirmed independently 4 separate times (redundancy testing, two combined-signal backtests, and by elimination in an unrelated cross-market investigation).

Economic effect

+0.47x realized volatility at 60 trading days (~1 quarter), for the Lower 48 national aggregate.

Servable regions

Region	Subject id	Tier
Lower 48 (national)	`NW2_EPG0_SWO_R48_BCF`	Tier 1
Salt	`NW2_EPG0_SSO_R33_BCF`	Tier 3
Midwest	`NW2_EPG0_SWO_R32_BCF`	Tier 3
South Central	`NW2_EPG0_SWO_R33_BCF`	Tier 3
Pacific	`NW2_EPG0_SWO_R35_BCF`	Tier 3
Mountain	`NW2_EPG0_SWO_R34_BCF`	Excluded
Nonsalt	`NW2_EPG0_SNO_R33_BCF`	Excluded
East	`NW2_EPG0_SWO_R31_BCF`	Excluded

Why Mountain is excluded despite the largest raw effect (+0.74x) of any region tested: the worst out-of-sample stability of any storage region (3%, vs. 51-100% everywhere else), and a redundancy test showing its apparent edge evaporates to a 50% walk-forward hit rate - a coin flip - once the Lower 48 aggregate is controlled for. Full story: Why we didn't ship our best backtest result. Nonsalt and East are excluded for the same reason (redundant with R48, not independent information).

Query it

GET /v1/factors?factor=storage_percentile&subject=NW2_EPG0_SWO_R48_BCF

Full validation methodology and every real number behind these verdicts: FACTOR_CATALOGUE.md.

Free early access to the API: Get early access →