ENERGYSCOPE · all reports agent-written from live data — how these reports earn trust

Regime — HORMUZ CRISIS

v2.9.2 · Apr 14 · IEA OMR
⚠ Named Regime: HORMUZ CRISIS
Strait of Hormuz closed since Feb 28. The largest supply disruption since 1973. Now confirmed by IEA April OMR. Supply crashed 10.1 mb/d to 97 mb/d. 13+ mb/d exports severed. Demand revised -730 kb/d (contracting). OPEC-IEA spread widened to 2.53 mb/d — fundamental disagreement on whether the world needs more or less oil.
Platform validation: The signal layer identified this regime 5 days before the cause was publicly known. From Apr 8-13, derived_signals correctly fired: freight_extreme (z=7.92-8.76), supply_shock_signature (VIX ρ +0.92), regime_curve_destruction (-$22.61 backwardation), forced_storage_build (Cushing +14.3%), freight_locked_arb (composite -0.71). All were downstream of Hormuz. The platform got the WHAT right without the WHY. The IEA OMR today added the causal label. This is the designed division of labor: deterministic signals catch regime changes; human/IEA context supplies the geopolitical cause.
All 9 Active Contradictions (ranked by score)
1.000diesel_clean_long — dominant_lens
0.965forced_storage_build — crude_supply × market_structure
0.922supply_shock_signature — macro × market_structure
0.920regime_curve_destruction — market_structure
0.797refiner_margin_trap — refined_products
0.725freight_locked_arb — market_structure
0.725fundamental_rally — market_structure × positioning
0.522gasoline_overhang — refined_products
0.263opec_revision_vs_price — market_structure × opec_compliance
Contradictions
9/15
Signals
19/45
OPEC-IEA
+2.53
IEA Supply
97
Methodology Appendix

Tools used: breakpoints (structural break detection on WTI, Brent, stocks), volatility (GARCH vol for WTI/Brent), derived_signals (full 45-signal snapshot with contradiction scoring), compare_runs (run-over-run diff for regime report), IEA.OMR_HEADLINE (supply, Hormuz loss, demand revision), OPEC MOMR Tables 5-7/5-8/11-1 (production, demand, reference basket).

Data sources: All 11 datasets — ICE (Brent futures curve), NYMEX/CME (WTI futures), EIA PET (US stocks), EIA STEO (short-term outlook), IEA OMR (global balances), OPEC MOMR (production/demand/compliance), CFTC (positioning), Baltic Exchange (freight), CBOE (VIX correlation), EIA WPSR (weekly supply), Baker Hughes (rig counts).

Signal layer: 45 predefined signals evaluated against the snapshot. 19 currently active (42%). Signals are deterministic predicates over metric thresholds — no ML, no fitting. 15 contradiction definitions test for internally inconsistent signal combinations; 9 are currently firing (60%). Contradictions are scored as the average of their constituent signal scores, ranked by score descending. The dominant_lens (diesel_clean_long, score 1.000) is a meta-signal identifying the single trade thesis best supported by the full contradiction set.

Platform validation: The signal layer detected the Hormuz regime on Apr 8 — 5 days before the IEA OMR confirmed the cause on Apr 13. Detection was based on freight_extreme, supply_shock_signature, regime_curve_destruction, forced_storage_build, and freight_locked_arb firing simultaneously. This demonstrates the designed separation: deterministic signals detect the WHAT; external context (IEA, OPEC, news) supplies the WHY.

Snapshot keys: wti_vol_garch, brent_vol_garch, vix_corr, bwet_z, crack_321_z, crack_diesel_z, brent_wti_zscore, m1_m12, roll_yield_pct, cushing_7d_pct, wti_price_pctl, iea_global_supply_mbd, iea_hormuz_loss_mbd, iea_demand_revision_kbd, opec_iea_spread_2026, regime, primary_dislocation, dominant_lens, contradictions_active_count, signals_active_count.

Contradiction themes (7): crude_supply, market_structure, macro, refined_products, positioning, opec_compliance, diesel_clean. The regime report is the meta-report — it synthesizes all themes rather than focusing on a single lens.

9/15 contradictions · 19/45 signals · 6/7 themes · HORMUZ CRISIS confirmed · Platform validation: regime detected 5 days before cause known
This pack was generated by one prompt against the EnergyScope platform.
The report presets, named signals, backtests and revision trail come built in — your Claude (or any MCP-capable agent) writes the words. The energy data here is the live demo; the same access layer points at your own data, in your own Snowflake or Iceberg lake.
Generate it yourself — free key → How these reports earn trust energyscope.io