Burke County Acceleration — End-to-End Verification

Verified: 2026-03-24 Data: Live Snowflake + RAW layer cross-check Analyst: Claude Code
Verdict: CAVEAT

The Burke County acceleration is real signal. Direction confirmed independently at three layers (FCT analytics, RAW docket, zero bucket overlap). Specific numbers cited on the ML page (+573%, 175 recent wells, 26 baseline) are from a prior run — current data shows +444%, 136 recent, 25 baseline. Four specific caveats documented below.

01

Computation Trace

The chain is clean and correct end-to-end:

RAW_DOCKET_INTELLIGENCE (raw docket JSON, one row per docket file/hearing_date) └─ fct_docket_cases_refresh.sql ├─ LATERAL FLATTEN → one row per case per hearing_date ├─ first_hearings CTE → MIN(hearing_date) per case_no = FIRST_HEARING_DATE ├─ QUALIFY ROW_NUMBER() PARTITION BY case_no → ONE row per case (most recent hearing) └─ Filters: commercial_relevance IN ('high','medium'), hearing_date >= 2024-01-01 └─ ML_MOMENTUM_INPUT view └─ is_recent = FIRST_HEARING_DATE >= -90d window = FIRST_HEARING_DATE >= -180d PIPELINE_STATUS NOT IN ('dismissed') └─ county_momentum.json (generate_applied_intelligence.py → county_sql) └─ SUM(COALESCE(WELLS_AUTHORIZED, 1)) bucketed by FIRST_HEARING_DATE Groups by exact COUNTIES string value

The FIRST_HEARING_DATE fix is correctly implemented. The FCT QUALIFY dedup guarantees each case appears exactly once. Each case_no maps to exactly one time bucket — no case can be counted in both recent and baseline.


02

Case-Level Detail for Burke County

Recent wells (0–90d)
136
48 cases — county_momentum.json
Baseline wells (91–180d)
25
~12 cases
Growth (published)
+444%
generated 2026-03-24
Double-count overlap
0
zero case_nos in both buckets

Live FCT query (non-dismissed only): 49 cases, 116 wells authorized. The 136 vs. 116 difference is COALESCE(WELLS_AUTHORIZED, 1) — approximately 20 Formentera Operations LLC cases in Burke with 0 specified wells each get counted as 1 in the published JSON. These are infrastructure/spacing filings, not drilling authorizations.

Phoenix Operating — Burke cases (non-dismissed, 0–90d):

CasesWells eachFirst HeardLast HeardStatus
32518–32521 (4 cases)52026-01-222026-03-26progressed_to_order
32528–32533 (6 cases)52026-01-222026-02-26progressed_to_order
32610, 32621, 32622 (3)52026-02-262026-02-26progressed_to_order
32587, 32588 (2 cases)12026-02-252026-03-25progressed_to_order

Phoenix non-dismissed recent: 15 cases, 72 wells — 62% of Burke's 116 recent non-dismissed wells. 6 additional Phoenix/Burke cases (32522, 32523, 32534–32538, 5 wells each) were dismissed and correctly excluded from counts.

FIRST_HEARING_DATE vs. HEARING_DATE for Phoenix/Burke continued cases: Cases 32518–32521 were first heard 2026-01-22 and last appeared on docket 2026-03-26 (3 raw appearances each). FCT correctly records first_hearing_date = 2026-01-22, hearing_date = 2026-03-26. They fall in the recent bucket (Jan 22 is 62 days ago). With the old HEARING_DATE logic, continued cases could appear in multiple time buckets across pipeline runs; FIRST_HEARING_DATE locks each case to one bucket permanently.


03

RAW Docket Cross-Check (Bypasses FCT Layer)

Queried RAW.RAW_DOCKET_INTELLIGENCE directly — no dedup, no commercial_relevance filter, no dismissed filter. All 74 distinct case_nos with Burke in the 180-day window:

Hearing DateDistinct CasesWells AuthContinued from Prior
2026-03-26213614
2026-03-255192
2026-02-26358321
2026-02-25481
2026-01-2229963
2026-01-215160
2025-12-18353
2025-12-17351
2025-11-20461
2025-10-233111
2025-10-22200
2025-09-26200
2025-09-25531

Highlighted rows fall in the recent 0–90d window (on or after 2025-12-24).

RAW recent (0–90d)
+504%
48 cases · 133 wells
RAW baseline (91–180d)
22 wells
16 cases
RAW bucket overlap
0
no double-counting in raw either

The raw hearing data independently confirms Burke's acceleration at +504.5% without relying on any FCT transformation. The pattern of high continued-case counts on Jan 22 / Feb 26 / Mar 26 is consistent with a real regulatory filing push — continued cases indicate hearings where commissioners requested additional documentation or split large case batches.


04

Independent Data Confirmation

nd_daily.json (production/drilling status): 0 Burke wells in this dataset. nd_daily.json tracks active wells by drilling/production status; Burke's docket filings are at the authorization stage and have not yet propagated to spud or production records. This is expected — docket filings precede spud by weeks to months. No contradiction, but no confirmation from this source.

operator_momentum_report.json (current pending pipeline): Phoenix Operating appears 4 times in the top-10 pending pipeline: Williams/Big Stone-Bakken (25 wells, 99.97% progression probability), Divide/Skabo-Bakken (13 wells), Divide/Burg-Bakken (10 wells), Dunn/Murphy Creek-Bakken (12 wells). Phoenix's Burke cases are absent from the pending pipeline because they have already progressed_to_order — approved and complete. The pending pipeline correctly reflects what comes next. The Burke completion and the Williams/Divide/Dunn pipeline are two phases of the same Phoenix expansion.

No permit/scout ticket data: No wells.csv, permit CSV, or spud CSV was found locally. Cross-referencing NDIC well data would be the next independent confirmation step.


05

Specific Flags and Caveats


06

Summary

SourceRecent WellsBaseline WellsGrowthOverlap
Claim (prior run)17526+573%
county_momentum.json (2026-03-24)13625+444%0
Live FCT (non-dismissed)116~22+427%0
RAW (independent, no filters)13322+504%0

Every layer produces growth of roughly +430%–+505%, independently, with zero double-counting. The Burke County acceleration is real. The specific headline figure should be updated from +573% to +444% on next page refresh.