Skip to content

INDEX.md

Latest commit

 

History

History
80 lines (61 loc) · 7.2 KB

INDEX.md

File metadata and controls

80 lines (61 loc) · 7.2 KB
Raw

reports/ — Index

Classification of every file under reports/. Read this before picking a v1b_*.md or other dated report — most are historical evidence snapshots tied to a methodology (M5) that is no longer deployed.

For the current deployed state, read STATUS.md (root). For the dated decision log, read methodology_history.md.

Conventions

  • current — describes what is deployed or what an in-flight workstream is working on. Read these freely.
  • paper-evidence — frozen evidence pack for a paper section. Read when revising or citing that section.
  • historical — evidence snapshot tied to a superseded methodology. Read only when investigating why something is the way it is, not what is true today.
  • operational — recurring monitoring output. Read when on-call or auditing live coverage.

Current

File Role
methodology_history.md Append-only dated decision log. §0 = current state, §1 = chronological entries, §2 = open questions, §3 = artefact map. Source of truth for why the deployed methodology looks the way it does.
m6_validation.md Full M6 LWC robustness battery — per-symbol Kupiec, LOSO, GARCH baselines, Phase 7/8 evidence. Companion to Paper 1 §6 / §7 revision.
m6_sigma_ewma.md Phase 5 evidence for the σ̂ K=26 → EWMA HL=8 promotion (current canonical σ̂ rule).
m6_simulation_study.md 4-DGP Monte Carlo + Phase 6 sample-size sweep (newly-listed-symbol admission threshold).
active/ In-flight working docs (M6 refactor plan, Phase 7/8 results, validation backlog). See active/INDEX.md.

Operational (recurring monitoring)

File Role
m6_forward_tape_0weekends.md, m6_forward_tape_1weekends.md, m6_forward_tape_1weekends_variants.md Forward-tape harness output (auto-rolling N-weekends report; regenerated weekly Tuesday by launchd/com.adamnoonan.soothsayer.forward-tape.plist).
kamino_xstocks_weekend_20260417.md, 20260424.md, 20260501.md Per-weekend Kamino-xStocks oracle comparison snapshots.
backed_corp_actions_enriched.md, backed_corp_actions_summary.md Backed corporate-action audit snapshots.
kamino_liquidations_first_scan.md First-scan write-up of the kamino/liquidations/v1 panel (102 events, 2025-08 → 2026-04). Paper 3 evidence.

Paper drafts

Subdirectory Paper
paper1_coverage_inversion/ Paper 1 — calibration-transparent oracle. In revision against M6.
paper3_liquidation_policy/ Paper 3 — liquidation policy under calibrated uncertainty. Three-claim structure (Geometric / Structural / Empirical).
paper4_oracle_conditioned_amm/ Paper 4 — oracle-conditioned AMM. Forward data capture in flight (scryer item 51); Soothsayer consumers pending parquet rows.

Paper-evidence (frozen)

These are evidence snapshots cited by paper sections. Don't edit unless the paper section's claim changes.

File Cited from
v1b_paper1_robustness.md Paper 1 §6.4.1, §6.3, §9.3, §9.4 (per-symbol bimodality, split-date sensitivity, LOSO, per-class)
v1b_calibration.md, v1b_ablation.md, v1b_buffer_tune.md, v1b_decision.md Paper 1 §4 / §5 (M5 reference baseline lineage)
v1b_walkforward.md, v1b_window_sensitivity.md, v1b_leave_one_out.md Paper 1 §6.3 (walk-forward + window sensitivity + LOSO)
v1b_diagnostics.md, v1b_diagnostics_extended.md, v1b_density_rejection_localization.md Paper 1 §6.4 / §9 (density tests, Berkowitz / DQ disclosure, W2 lead-finding)
v1b_path_coverage.md, v1b_path_coverage_robustness.md Paper 1 §6.6 / §10.1 (path-fitted conformity)
v1b_pyth_comparison.md, v1b_chainlink_comparison.md, v1b_redstone_comparison.md, v1b_kamino_scope_comparison.md, v1b_incumbent_oracle_comparison.md Paper 1 §6.5 / §6.6 + Paper 3 §Structural (incumbent oracle benchmarks)
v1b_forward_coverage.md W5 (live forward-tape coverage; recurring)
v1b_w4_asymmetric_coverage_lending.md Paper 3 §Structural (one-sided lending-consumer table)
v1b_m6a_common_mode_partial_out.md, v1b_m6b_per_symbol_class_mondrian.md, v1b_m6c_combined.md, v1b_r_bar_forward_predictor.md Methodology log (W2-followup, W8 — dual-profile architecture and AMM-track gating). M6a/c are the "documented upper bound, not deployed" leads.
v3_bakeoff.md Methodology log 2026-05-03 (lead-up to the M6 LWC promotion)
bear_case.md Strategic gate log; updated when validation gates flip
demo_kamino_comparison.md Crate crates/soothsayer-demo-kamino evidence

Historical (do not consult for current state)

These predate M5 / M6 or were trial methodologies that did not progress. Read only when investigating a specific historical claim.

File Note
v1_chainlink_bias.md v1-era Chainlink weekend-bias post-mortem
v11_cadence_verification.md Chainlink v11 24/5 cadence verification (subsumed by current docs/sources/oracles/chainlink_v11.md)
v1b_extended_grid.md, v1b_macro_regressor.md, v1b_pooled_tail_trial.md, v1b_evt_pot_trial.md, v1b_conformal_comparison.md, v1b_hybrid_validation.md Trial methodologies; not adopted into M5/M6
phase1_week1.md, phase1_week2.md Phase-1 weekly scoping (pre-M5)

Subdirectories

Path Contents
active/ In-flight working docs. See active/INDEX.md.
archived/ Explicit archive (vault updates, retired v2/v3 trial notes).
figures/ Persisted plots cited by paper drafts and methodology entries.
tables/ CSV snapshots for every numerical claim in the reports above.