LL-4: multi-strategy shadow-equity ledgers and capital priority

Date: 29 August 2026 Status: DESIGN/PENDING OWNER · FIX-FIRST · no authority changed Source status: VERIFIED · Date: 29 August 2026 · checked-out docs commit 9e5a130; external status reports #347 merged and #348 merged pending main CI, but those merges are not present in this branch history.

1 — Actual boundary

• STATUS: VERIFIED · DATE: 29 August 2026. 1 global hash chain already carries sorted strategyIds, tickerOwners, and per-order strategy joins. The gap is economic—execution/plan.ts collapses target sets into 1 approval, 1 overlay, 1 StrategySpec hash and account-level whole shares; reconciliation, daily NAV, daemon HWM, rehearsal, graduation, and both scripts remain single-strategy. Joined IDs are labels, never ownership.

2 — One account, unambiguous shadows

• STATUS: PROPOSED · DATE: 29 August 2026. Every post-cutover account receipt carries sorted per-strategy checkpoints: prior checkpoint hash, cash, shares by ticker, receivables, NAV, HWM, StrategySpec/approval/stage hashes, and contribution hash. Prove on every receipt that strategy shares sum to each broker position, strategy cash plus explicit reserve equals broker cash, and strategy NAVs plus reserve equal account equity; any residual is H3. The existing global previousHash remains the only ordering authority.
• STATUS: PROPOSED · DATE: 29 August 2026. Each StrategySpec keeps its own approval, overlay, and 10→25→50 ladder; only the £10k cap is account-wide. H1 day loss and H2 drawdown use each strategy’s shadow NAV/HWM. A breach disarms that strategy and the common coordinator moves the account to the surviving aggregate target; inability to prove selective cleanup escalates to H3. H3 reconciliation/data failure and H4 short/buying-power/trading-blocked remain account-wide, disarm all strategies, and use the existing cancel→snapshot→incident→flatten→two-zero sequence.

3 — Capital priority and implementability

• STATUS: OWNER-SEAL REQUIRED · DATE: 29 August 2026. Seal capital-priority/v1 before candidate #2 graduates. Proposed deterministic waterfall: safety reductions first; then complete approved strategy reservations in ascending sealed rank; ties by strategyId, then ticker. Compute and floor each strategy independently before aggregation, so overlapping fractional dust cannot manufacture a share. Never fund from expected same-open sale proceeds. A lower-ranked book that cannot receive its complete reservation is deferred, not silently haircut or reweighted.
• STATUS: ROUND-4 CORRECTION / OWNER AUTHORITY REQUIRED · DATE: 29 August 2026. Apply prospective policy-v2 implementability only to each strategy's positive intended target book after priority/overlay/ADV cap and before current-position deltas or netting: ≥90% of managed names with positive quantity, ≥90% deployed notional, and per-name one-share headroom. Equality passes. Flat plans are applicable:false with zero quantities and null fraction/headroom metrics; emergency kill/flatten bypasses the planner. Failure seals submitEligible:false / IMPLEMENTABILITY_REFUSED and reaches no POST. At $12,700 NAV ×0.428, 0/277 months pass all gates; worst names are 18.0488%, deployment 13.9534%, and the latest month has 45/217 names, 16.0794% deployed, and 0.007789× headroom. The former $881,948.56 result covered only the name/deployment gates. The all-gate common minimum is $2,521,437,128,668,714.50 because a June-2013 raw-price outlier selected by the frozen rule dominates all-name headroom. Do not filter it away: make headroom a per-name operational test, not an asserted practical capital requirement.

4 — Net once, attribute exactly

• STATUS: PROPOSED · DATE: 29 August 2026. A new multi-strategy execution-plan v4 seals each strategy's current shares, target shares, signed delta, rank, and contribution hash. Round-4 v3 remains byte/schema-specific to the shipped single-strategy planner and is never reused for an incompatible payload. Same-direction claims sum; for each ticker, internally cross opposing deltas and POST exactly 1 residual account order. Net zero means no broker order.
• STATUS: PROPOSED · DATE: 29 August 2026. Reconciliation prices internal crosses at the verified official open, allocates external/partial fills by sealed rank, and proves allocated quantity, cash and actual costs equal broker evidence. Splits transform every shadow lot; dividends and CIL use ex-date shadow entitlement. Consume each broker activity once globally, then allocate exact cents/rational residuals by the sealed order.
• STATUS: REPORT-ONLY · DATE: 29 August 2026 · trial:null · variants_allowed:0. The survivor netting lens reproduces 5,767 daily rows and 275 GP/A intervals without post-2021 market-return rows. A vol-scaled 50/50 blend reports component-net arithmetic mean ×12 of 9.97%/yr, CAGR 9.34%, Sharpe 0.70, and drawdown −34.41%; equal-risk weights are 37.14% GP/A / 62.86% momentum. This does not validate multi-strategy execution: the report has no holdings/trades, opposite-trade overlap, netted costs, or exact shadow attribution. Its approximately-zero active-GP/A-versus-total-momentum correlation is not a total-book diversification estimate.

5 — Receipts, migration, files

• STATUS: PARTLY DELIVERED / MIGRATION PROPOSED · DATE: 29 August 2026. Round 4 makes live-lane-execution-plan/v3 canonical for the single-strategy lane: producers, CLI loading, submit/reconcile validation, and rehearsal replay reject v2/unknown receipts, and the diagnostic calls the same planner kernel. The 4–11 September rehearsal chains start fresh with v3. The later multi-strategy cutover requires execution-plan v4 (or a separately versioned envelope that leaves v3 bytes untouched), sizing aggregate v2, submit v2, reconciliation v5, daily-NAV v2, rehearsal v3, and graduation v2.
• STATUS: MIGRATION REQUIRED · DATE: 29 August 2026. Append—not rewrite—shadow-ledger-migration/v1 after verifying the full chain, latest matched v4 receipt, zero open orders/incidents, and current broker snapshot. It wraps momentum’s v1 graduation/rehearsal, assigns opening shares/cash/reserve, and becomes the sole v5 cutover hash. Reject a second migration or any post-cutover legacy/mixed receipt.
• STATUS: BOUNDED · DATE: 29 August 2026. Add lib/live-lane/shadow-equity/index.ts; change registry/index.ts, execution/{plan,alpaca}.ts, reconcile/{index,receipt}.ts, corporate-actions/index.ts, risk/{index,hard-action}.ts, daemon/index.ts, rehearsal/index.ts, graduation/index.ts, monitoring/index.ts, lib/live-lane/index.ts, and both live-lane scripts, plus scoped tests. Freeze lib/live-lane/strategy/**, lib/paper/intraday/tournament/historical-corpus.ts, both preregistrations, and live-risk-policy/v1 Markdown/event bytes.
• STATUS: NON-GOALS · DATE: 29 August 2026. No broker subaccounts, margin, shorts, fractional broker shares, tax lots, policy recalibration, StrategySpec edits, or second simulator/ledger/executor.

6 — Five boundary tests

1. STATUS: REQUIRED · DATE: 29 August 2026. Opposing same-name deltas produce zero/1 POST and conserve shadow cash/shares.
2. STATUS: REQUIRED · DATE: 29 August 2026. Scarce capital follows sealed rank; an unimplementable book refuses without reweighting.
3. STATUS: REQUIRED · DATE: 29 August 2026. 1 strategy H1/H2 removes only its shadow; H3/H4 flatten all.
4. STATUS: REQUIRED · DATE: 29 August 2026. Shared split/dividend/CIL allocations sum exactly; duplicate activity refuses.
5. STATUS: REQUIRED · DATE: 29 August 2026. Legacy v4→migration→v5 succeeds once; mixed, reordered or repeated cutover refuses.

7 — Falsifiers and risks

• STATUS: OPEN · DATE: 29 August 2026. Missing official open, allocation residual, partial-fill mismatch, stale approval, unsealed priority, or broken migration lineage refuses. Candidate #2’s production StrategySpec/data path is absent from current code and must arrive in a separate frozen-spec packet.

Review amendments — controlling

STATUS: FIX-FIRST · DATE: 29 August 2026. This section supersedes any conflicting proposal above. The design remains pending owner authorization and engineering validation.

1. STATUS: NEW GOVERNANCE REQUIRED · DATE: before implementation. The 90% managed-name / 90% deployed-notional / per-name headroom gate and IMPLEMENTABILITY_REFUSED are not in sealed policy v1. Add a newly owner-authorized, hash-bound sibling policy v2; do not describe this as mere ratification. The gate applies independently to each strategy's target book after priority, never its net broker-order residual, net book, or union-name coverage. Flat plans are explicitly exempt and kill/flatten authority remains available.
2. STATUS: REQUIRED · DATE: before multi-strategy plan-v4 implementation. Define canonical available cash, explicit reserve, expected costs, and same-open-proceeds treatment. Bind the capital-priority/v1 hash through plan, submit, fill allocation, and reconciliation. Hash construction, deterministic tie-breaking, and refusal behavior are engineering invariants, not owner choices; shipped v3 evidence remains immutable.
3. STATUS: REQUIRED · DATE: before reconciliation-v5 implementation. Specify schemas for ex-date shadow entitlement; gross dividend, withholding, and reversals; split fractions when the broker supplies no CIL; exact-cent residual ownership; partial fills; ambiguous POST recovery; and deterministic external-fill allocation. Account activity is consumed once globally and every allocation must sum exactly.
4. STATUS: REQUIRED · DATE: before migration. The cutover event transfers HWM, arm age, consumed-activity set, approval/graduation lineage, reserve, and interrupted-state recovery. Validate a partial fill/timeout, split without CIL, dividend reversal, shared-ticker selective H1, interrupted migration, and post-cutover legacy write.
5. STATUS: REQUIRED · DATE: before policy binding. Reconcile the policy’s strict < language with code’s <= / >= comparisons and pin one boundary convention. A mismatch refuses migration and live authority.
6. STATUS: FALSIFIER · DATE: before candidate #2. If per-strategy shares/cash/NAV cannot sum exactly to broker positions/cash/equity after the above event classes, or selective cleanup cannot preserve a surviving strategy, the multi-strategy design is refused.

Pending owner

1. STATUS: OWNER DECISION · DEADLINE: 3 September 2026. Choose policy-v2 A: 0.90/0.90 plus all-name one-share headroom; B: 0.80/0.80 or 0.95/0.95 plus an explicit headroom choice; or C: reject the gate and accept distorted books. Prefer policy-read thresholds over a second code-constant authority; if all-name headroom remains, acknowledge that the name-fraction threshold is redundant.
2. STATUS: OPEN · DATE: 29 August 2026. Seal strategy ranks.
3. STATUS: OPEN · DATE: 29 August 2026. Choose momentum’s opening shadow cash/NAV versus reserve.
4. STATUS: OPEN · DATE: 29 August 2026. Forbid or allow same-open sale proceeds in available cash.
5. STATUS: OPEN · DATE: 29 August 2026. Confirm that H1/H2 preserves surviving strategies after selective cleanup.
6. STATUS: OPEN · DATE: 29 August 2026. Decide whether actual broker fees enter shadow NAV.

Critic round 1 — applied

• STATUS: APPLIED · DATE: 29 August 2026. Removed the unsupported 68% claim, distinguished the $881,948.56 pre-headroom result from the $2.521 quadrillion all-gate outlier, and made per-name headroom the operational nuance.
• STATUS: APPLIED · DATE: 29 August 2026. Recorded flat-plan/kill exemptions, plan-receipt v3, shared planner/diagnostic kernel, fresh rehearsal cutover, and prospective policy-v2 authority.
• STATUS: APPLIED · DATE: 29 August 2026. Retained every independent design-review finding as a controlling correction or explicit owner/engineering gate.

Round 3 — applied

• STATUS: APPLIED · DATE: 29 August 2026. Linked the report-only two-survivor netting lens and retained its missing holdings/trades/netted-cost evidence as a falsifier, not an allocation claim.
• STATUS: OPEN · DATE: 29 August 2026. Multi-strategy authority still requires the review’s exact-cent allocation schemas, migration lineage, HWM/arm-age transfer, policy boundary convention, and owner decisions above.