Research archive

Research documents record experiments and evidence. Their presence does not mean that a method affects production forecasts. Production changes require explicit implementation and review.

• Status reports are preserved in the repository research directory.
• Roadmap v8 final status records that the implemented accuracy experiments remain empirically blocked by unavailable frozen BTC/USD history; no skill or promotion claim was made.
• Roadmap v10 forecast evidence closeout summarizes the completed #397–#400 workstream outcomes, their dependencies, evidence artifacts, and remaining gates; no empirical accuracy conclusion or production change is claimed.
• Roadmap v11 forecast accuracy closeout summarizes all 16 completed work items, the blocked canonical corpus gate, limits of prospective diagnostics, and backup recovery verification; no accuracy-improvement claim or production change is authorized.
• Issue #397 corpus acquisition audit records the source evidence, exact admissibility constraints, acquisition/storage procedure, and blocked canonical audit.
• Issue #409 point-in-time corpus comparison records a documentation-based institutional/archive channel comparison and blocked procurement decision; machine-readable details are in JSON.
• Issue #399 production-parity replay status documents the blocked readiness gates, machine-readable report, and reproducible rerun procedure.
• Issue #400 bounded candidate confirmation records the blocked, no-scoring disposition, freeze-before-scoring protocol, and disjoint prospective D3 shadow prerequisites.
• Issue #416 blocked family closeout accounts for the #411–#415 family as blocked/not attempted, with zero attempts, no winner, and hash-checked evidence.
• Issue #412 TimesFM context/target evaluation records the diagnosis-gated blocked outcome, source hashes, and exact reopening requirements.
• Experiment catalog
• Statistical evidence