UK CGT: translate the 2024-25 gains outturn to the 2025 calibration year (forestalling year; OBR CGT +44.2% signed on the stack register)

[juaristi22]

Why

The UK national calibration binds hmrc.cgt.gains_total and hmrc.cgt.taxpayers_total at the 2025 calibration period under the doctrine "latest fact not after the period". Since the Chronicle re-pin to 6fb700e (microcosm#874, I0) the latest fact is HMRC's 2024-25 outturn — £127.3bn of chargeable gains and 584k taxpayers, against £69.9bn / 403k in 2023-24. 2024-25 is the forestalling year ahead of the 30 October 2024 rate rise (10/20% → 18/24%). The engine uprates stored gains by ~5% a year (£72.1bn 2024 → £75.8bn 2025 at design weights), so the weights lift the 2025 gains to the spike level, and 2025-26 rates then apply to all of it: £31.4bn of liability (24.7% effective) against OBR's £21.8bn FY2025-26 receipts and HMRC's realised 19.0% on the same gains. obr.capital_gains_tax@2025 reads +44.2% at the composed calibration (microcosm#874 receipts, Part C L4 run 4); with 2023-24 gains it read −25%.

tax year	chargeable gains	taxpayers	CGT liability	OBR cash receipts, following FY
2022-23	£82.0bn	377k	£14.7bn
2023-24	£69.9bn	403k	£12.8bn	FY2024-25 £13.7bn
2024-25	£127.3bn	584k	£24.2bn	FY2025-26 £21.8bn
2025-26	not published			FY2026-27 £20.8bn (forecast)

Facts: HMRC CGT statistics (2026 release, Table 1) and OBR EFO March 2026 receipts, both on the pinned Chronicle feed (6ae49d7d…).

Disposition now (ruling 2026-09-05)

Option 1 of the receipts: keep every fact bound as published and sign the OBR miss on the publication stack's target-fit register (uk/target_fit_reviewed_exclusions.json, entry obr.capital_gains_tax@2025, adjudication this issue). The target stays bound in the solve; only the release fence is deferred. While the entry is in force the calibrated 2025-26 base carries ~40% more CGT liability than OBR expects, and any CGT costing on it overstates the base by that much — the publication's known issues must say so.

Deliverable (retires the deferral)

A declared translation of the 2024-25 gains outturn to the 2025 calibration year, under the WS-C reconciliation layer (#736 "CGT 2025 level"):

1. The index ruling. The reference already carries uprating_from_period 2024 → uprating_to_period 2025 with uprating_index: None (the compile records the slot and applies the fact at face value). The only published series on the feed is OBR's CGT receipts, whose FY2026-27/FY2025-26 ratio (0.954) still leaves £121bn because it hides the rate rise. The honest translation is rate-adjusted (receipts ÷ an effective rate, ≈ £85–90bn), which is a derived series: it needs a declared computation and a Chronicle fact (facts belong in Chronicle), then the adjudication here.
2. Apply the index in the compile step. compile_ledger_target_references records uprating_index in metadata today and does not scale the value; the monetary-binding receipt already carries the three uprating fields.
3. Taxpayer count stays flat at 584k: the £3,000 annual exempt amount is the 2025-26 regime as well (uk-data also holds its count flat).
4. Cash-lag mapping for self-assessed receipts (obr.capital_gains_tax at FY2025-26 is mostly 2024-25 liabilities; the engine's 2025-26 liability lands in FY2026-27) — a doctrine question for every SA-lagged OBR line, to rule alongside.
5. Distribution follow-ups stay where they are: UK: banded CGT gains targets (by size of gain) to constrain the distribution, not just count + total #467 (banded gains), UK CGT top-tail age and geography joints inherit from incidence and look off — benchmark against HMRC marginals #725 (age/geography joints).

Not the deliverable

Re-vintaging the rows to 2023-24 (a stale £6,000 exempt-amount regime, OBR −25%) or signing the gains total out of the solve (dropping a published fact). Both were considered and rejected in the #874 receipts.

Refs: microcosm#874 (receipts experiments/834-childcare-tfc-receipts.md, Part C L4), #736, #796, #467, #725, #552 (closed), uk-data's hmrc_cgt.py (pins 2023-24, uprates gains, count flat).