Replay evidence

A local deterministic paper fixture exercises intent, atomic reservation, simulated fill, position update, balanced postings, external-style reconciliation and replay equality; the Phase 1B gate has not passed.

Pending approval

Identifier: CLM-011
Scope: Local, in-memory deterministic paper fixture.
Evidence date: 2026-08-09
Owner role: operations-owner
Review / expiry date: 2026-10-31
Residual risk: No live venue, broker, bank or custodian record is involved. Durable persistence, external statements, operational drills and independent approval remain outstanding.

Evidence sources

Limitations

Do not represent replay equality as investment performance or live-system resilience.
Do not describe the Phase 1B gate as passed.

Explanatory methodology

Within the scope of CLM-011, deterministic replay reconstructs paper trading sessions from recorded event streams. The replay engine consumes a canonical event log and produces exact position and financial state reconstructions. Injected breaks are detected by comparing replay output against expected state.

Events are recorded as an ordered, typed sequence.
Replay produces deterministic output given the same input and seed.
Injected-break detection compares replayed state to expected state and flags divergence.
Exact reconstruction equality is verified by hash comparison of final state.

Evidence metadata

Environment: Local, in-memory, paper simulation
Data source: Synthetic and recorded paper-session events
Venue connection: None
Production credentials: None
Replay determinism: Verified by exact state reconstruction equality
Break detection: Injected-break detection with divergence flagging

Limitations

Venue-calibrated latency, queue, market-impact, rejection and financing assumptions have not been independently validated.
External broker, venue and bank statements are not reconciled by a completed Phase 1B flow.
In-memory state is not durable production accounting.
Replay results are simulation evidence, not performance claims.