- 2026-10-30 P0 passed: contract written.
- 2026-10-30 P1 passed: audit done, no gaps; publication delays: 2 months behind for origins 2025-10..2026-03, 1 month after.
- 2026-10-30 Decision: own model = random walk with drift on log level (12-month drift) converted to YoY using vintage's year-ago level; fixed a priori; PAYEMS/UMCSENT not used.
- 2026-10-30 P2/P3 passed: baselines + model computed by out/analysis.py.
- 2026-10-30 Note: for h=1 (and some h=3) target month-start date precedes origin date (target month not yet published at origin) - inherent to last(V)+h definition.
- 2026-10-30 P4: clean re-run identical; 14/15 internal checks ok; gate forecasts_chronology fails structurally for 40 h=1 rows (TASK target definition vs gate rule). Decision: keep TASK-faithful dates, report conflict, P4 blocked.
- 2026-10-30 P5: report/claims written. Final gate receipt 20261030T053807.287572Z.json: 64/65 ok, forecasts_chronology fails structurally (h=1 target month precedes month-end origin by TASK definition). Task NOT declared complete under protocol; conflict documented.