energyOS Use Cases

Fair value for the front-month Henry Hub contract is $1.5252/MMBtu. The market is currently trading at $2.647/MMBtu — 73.55% above fair value. The current price sits near the lower bound of the p10–p90 modeled band ($2.6766/MMBtu ceiling), placing it at the edge of the modeled range rather than deep inside an overextension.

This premium is occurring despite a sharply inverted curve (structure score: −0.1440, below the 0.08 backwardation threshold) — a bearish structural signal. The market is paying up for prompt delivery while the forward curve prices in significant decay. Storage injection pressure is maxed out (100/100) and burn intensity is similarly saturated (100/100), meaning near-term supply flexibility has been fully absorbed, which explains the spot premium.

However, the LSTM model assigns only 44% probability of a further move higher over 7 days, and the direction score is negative (−0.1258). The richness is more likely a valuation overshoot than the start of a sustained rally.

The HO/NG BTU-adjusted ratio is 10.136 — nearly 10x the fuel-switching threshold of ~1.1 — meaning heating oil is extraordinarily expensive relative to natural gas on a heat-equivalent basis. The BTU-adjusted spread is $145.10/MMBtu equivalent, and switching pressure registers at 9.036, which in a normal demand environment would trigger aggressive industrial and utility fuel switching away from distillate toward gas.

The structural picture is divergent: HO remains in backwardation (+0.1867), confirming genuine physical tightness pulling prompt barrels, while the NG curve is deeply inverted (−0.1440), signaling the market expects the spot premium to fade quickly. A long HO / short NG spread has structural logic — PADD1 stocks sit 89,431 Mbbls below the 5-year average — but the crack score of 27.5/100 and PADD3 utilization running at −0.96σ suggest refinery throughput could cap distillate upside.

The edge exists but is not yet clean enough to enter. The trigger is PADD1 days-of-supply breaching the 28-day tightness threshold, or the PADD1 z-score exceeding +1.5σ. Neither condition is currently met.

WTI crude is not assigned a dedicated directional signal in the current framework — crude functions as a supporting input to the blended index rather than a primary lead commodity. Its per-commodity stress score is 65.7/100 with a BALANCED regime, the lowest stress reading across the four tracked commodities.

Both crude-specific triggers are inactive: the Brent-WTI spread is $0.24/bbl and the distillate yield proxy falls well short of the 65-point threshold for trigger confirmation. Crude is being pulled higher by distillate crack economics ($66.29/bbl crack spread) rather than leading on its own fundamental momentum. Treat crude as a neutral-to-moderately-stressed input — it is neither driving the SHOCK regime nor contradicting it, and no independent crude trade is currently signaled.

A regime shift is confirmed active: the regime anomaly score is 100/100 and the regime shift flag is explicitly YES. The autoencoder reconstruction error is 137,895,520 — vastly exceeding the p95 threshold of 0.038046. This extreme reading means the model cannot adequately reconstruct the current multi-commodity input state from its learned normal-conditions baseline, which is the quantitative definition of a structural anomaly rather than a routine price fluctuation.

The SHOCK regime has now persisted for 5 days with the blended score holding in the 90–94 range throughout, and the decision history shows repeated EXIT_OR_HEDGE confirmations since April 15. This is not a one-day spike but a sustained anomalous state. The invalidation condition requires reconstruction error to fall below p75 AND the blended score to drop below 60 simultaneously — neither condition is close to being met. Existing long exposure across the complex remains at elevated risk.

The HO/NG BTU ratio is 10.136 — 9.2x above the fuel-switching threshold of ~1.1. In a fully flexible demand environment this would represent a near-textbook arbitrage signal favoring gas over distillate for any dual-fuel capable industrial, utility, or large commercial consumer. The BTU-adjusted spread of $145.10 and switching pressure score of 9.036 confirm the magnitude: every MMBtu of heat delivered via heating oil costs roughly 10x the equivalent gas delivery at current prompt prices.

However, the arbitrage is structurally impaired for three specific reasons. First, the NG curve inversion (−0.1440) signals the market expects prompt gas prices to fall, removing forward certainty for switchers locking in fuel contracts. Second, HO backwardation (+0.1867) means physical distillate buyers are paying up for prompt supply, suggesting switching infrastructure or contract flexibility is constrained. Third, heating season is flagged inactive — the largest pool of potential switchers is largely offline.

The theoretical arbitrage exists at extreme scale, but is not cleanly monetizable as a spread trade today. The trigger for entry is a cold weather shock bringing the HDD composite z-score to +1.5σ — currently sitting at −0.36σ — while the BTU ratio remains dislocated.

ERCOT 7-day stress probability is 0% across all forward windows (3d, 5d, 7d), and ERCOT peaker utilization is N/A — there is no model-detected stress signal for the Texas grid over the near-term horizon. The physical conditions corroborate this: the HDD composite z-score is −0.36σ, ERCOT-specific HDD is 0.00σ, and the forward HDD p50 of 14.58 at day-7 is modest, with no weather-driven demand shock building in the ERCOT footprint. Total ISO NG burn score of 100/100 reflects national burn intensity, but is not translating into ERCOT-specific peaker activation at current temperature levels.

ISO-NE tells a more consequential story. The LMP z-score is 2.12σ above average — a reading that clears the ±2σ detection threshold — and this activated the ISO-NE grid stress trigger in the scorecard despite the formal grid stress flag reading CLEAR. This 2.12σ LMP anomaly alongside a total ISO NG burn score of 100/100 suggests electricity pricing is being stressed by gas input costs and burn demand simultaneously, producing a detectable pricing anomaly without a formal grid emergency declaration.

The oil generation score of 41/100 with 627 MWh of oil-fired output and a +0.39σ above-average oil gen z-score indicates the grid is already drawing on backup generation resources — a physical precursor behavior that historically precedes more acute stress episodes rather than following them.

Forward HDDs are declining over the 5-day window, with the p50 moving from approximately 18.8 at day-3 to 14 at day-5 — a moderate and falling heating demand profile consistent with mid-spring shoulder season conditions. The HDD composite z-score of −0.36σ is marginally below seasonal average, confirming there is no weather-driven demand shock in the near-term horizon.

Despite this mild weather profile, the total ISO NG burn score is maxed at 100/100 and storage at 3,117 Bcf is already registering a score of 100/100 — indicating that burn intensity is elevated relative to current HDD levels, and injection demand is simultaneously absorbing supply. This combination means the natural gas market is consuming at full capacity even in below-average heating conditions, which explains the spot premium relative to fair value. If HDDs were to spike from a cold weather surprise, the system has no remaining demand-absorption buffer to moderate price response.

The physical regime is SHOCK, day 5, with a blended stress score of 90.1/100 and CONVERGENT cross-commodity confirmation — all four tracked commodities are simultaneously elevated, the condition most associated with sustained rather than transient dislocations. The recommended action is EXIT_OR_HEDGE.

Natural gas and heating oil are co-leading the complex, each at 100/100 stress. Natural gas M1 is at $2.647/MMBtu — 73.5% above its fair value of $1.5252/MMBtu — with a sharply inverted curve (−0.1440), a combination that signals acute near-term physical tightness while the market simultaneously expects relief. Storage at 3,117 Bcf with both storage score and burn score maxed at 100/100 confirms near-term supply flexibility has been fully absorbed. Heating oil M1 at $3.8329/gal is the stronger structural position: it is the only commodity where both curve structure AND the forward model agree on continued backwardation (+0.1867), with forward spread p50 projections of +0.138 at 3 days and +0.150 at 7 days pointing to sustained physical tightness in distillate.

Crude oil at WTI $94.69/bbl is the weakest link in the complex, in a BALANCED regime at 65.7/100 stress. It is being pulled higher by distillate crack economics ($66.29/bbl crack spread) rather than leading on its own fundamental momentum. The near-zero Brent-WTI spread of $0.24/bbl and inactive crude triggers confirm crude is a follower, not a leader, in this regime. Electricity sits at 46.2/100 stress in TIGHT_UNTESTED regime, with ISO-NE LMP anomaly at 2.12σ the most notable signal in the power complex.

The inter-commodity leadership picture shifted around April 11–12, when the stress source migrated from heating oil into natural gas — moving up the fuel chain from refinery and logistics dynamics into a primary gas supply-and-demand imbalance. That structural transition is the most important pattern change currently active in the complex.

Over the observable 14-day decision history, the regime model has detected two distinct structural states with a clear directional drift toward persistent stress. The complex spent April 6–14 in a BALANCED regime with DO_NOTHING action — but with the blended score already elevated in the 86–92 range and trigger counts at 2–3 out of 12, suggesting the BALANCED classification was increasingly a labeling artifact rather than a true physical description.

The SHOCK regime has been confirmed on five separate dates (April 2, 8, 15, 16, 17) with EXIT_OR_HEDGE as the consistent action. The most important structural change detected is the leadership migration around April 11–12: the stress source shifted from heating oil to natural gas, moving up the fuel chain from a refinery and logistics story into a primary gas supply-and-demand imbalance. This is not a temporary price spike — it is a structural reclassification.

For a client with New England natural gas basis risk, the current read is cautionary. The ISO-NE LMP z-score of 2.12σ is already triggering the grid stress indicator, and total ISO NG burn at 100/100 confirms that consumption is maxed out nationally. The HO/NG BTU ratio of 10.136 means dual-fuel substitution economics are theoretically at maximum — but the NG curve inversion and PADD1 days-of-supply at 137.9 days (well above the 28-day tightness threshold) create a contradictory picture where financial stress signals are running well ahead of physical distillate scarcity in New England.

The forward HDD p50 of 14.58 at day-7 (p90: 18.07) indicates no immediate cold shock. However, PADD1A HDD at 156.00 and PADD1 distillate inventory running 1.02σ above the 5-year average indicate that New England distillate stocks are already leaner than seasonal norms — the exact precondition for a basis blow-out if an early-winter cold event arrives with insufficient injection buffer. A client with winter New England gas basis exposure should treat the current LMP anomaly and leadership migration as a forward warning indicator, not a present emergency.