test(research): add Fib to Genesis V2 Phase 2 dummy contract

Mechanical contract/dummy test inside the Fib repo only — not a real export, not
Genesis integration. New stdlib-only research/feature_contract.py validates two
committed synthetic dummy CSVs against the Phase 1 schema: exact-header schema,
join keys (symbol,timeframe,timestamp) non-null + unique, per-row causality
known_after_ts <= timestamp over the whole reference set, knowability floor
known_after_ts >= max(anchor_b)+buffer, 1H fail-closed, feature/metadata boundary
asserted at import. No fib computation, no Genesis import, no fibengine runtime
import, no pipeline/ML/backtest/signal/edge. confirmation_buffer unit pinned to
hours (confirmation_buffer_hours). 20 tests; ruff + 426 passed + repo-bounds green.
Wiki: log/handoff updated; 06-11..06-12 entries archived to post-reset part 1.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: JohnCCarter

docs/research_wiki/handoff.md

@@ -22,6 +22,12 @@ append-only trail lives in [log.md](log.md).
 
 ## Recent Changes
 
+- **2026-06-16 Fib → Genesis V2 Phase 2 dummy contract test (narrow slice, DONE)** —
+  mechanical contract/dummy test **inside Fib only** (not export, not Genesis integration).
+  `research/feature_contract.py` (stdlib) validates two committed dummy CSVs vs the Phase 1
+  schema (join keys, causality, knowability floor, 1H + feature/metadata fail-closed). No fib
+  computation, no Genesis touch. 20 tests; gates green. **Stopped as scoped.**
+  [Report](reviews/btc-fib-to-genesis-v2-phase2-dummy-contract-20260616.md).
 - **2026-06-15 MTF confluence CP1–CP3 CLOSED + interpretation/decision note** — first atlas
   pack done (5 cards, 3 archetypes, all **human-approved**): c001 robust fixed-band 4-TF; c002
   chaining-dependent contrast (NOT tight); c004/c006/c007 zero-span exact-price 3-TF.
@@ -62,41 +68,10 @@ append-only trail lives in [log.md](log.md).
   in [2021-01-01, 2021-04-01), ~37 fibs). Threshold rule confirmed: local density per
   zone, not total fib count. Y-axis log confirmed (line 246 `monthly_fib_map.py`).
   Full review: [`reviews/btc-4h-tier1-map-review-20260615.md`](reviews/btc-4h-tier1-map-review-20260615.md).
-- **2026-06-12 4H visual confirmation Tier 1 built** — `research/fourh_source_fib_map.py`
-  renders annual combined 4H maps (366/366 drawn, 11 groups; dense 2017_h2 flagged for
-  Tier 2). Reuses `monthly_fib_map` primitives; output gitignored. Full detail in log.md.
-- **2026-06-12 4H source-fib phase complete** — **366** manual BTC/USD 4H source fibs
-  drawn and verified (timeframe `4h`, log scale, `tradingview_log_chamoun`, levels
-  `[0, 0.382, 0.5, 0.618, 0.786, 1]`, no 0.236, endpoint mapping ratio 0.0=anchor_b /
-  1.0=anchor_a, anchor direction, log-spacing, human/manual only). Coverage
-  **2017-01-05 → 2026-06-05**; **up=169 / down=197**. This is **source-labeling
-  completion, not reaction-review** — visual confirmation / reaction-review is a later,
-  separate decision. No auto-fib, no trading conclusions. 366/366 schema verification PASS.
-- **2026-06-12 1D reaction-review complete** — all **67** BTC/USD 1D source fibs
-  reviewed on **4H** (expansion config, 4H history to 2017-01-01). Review window:
-  anchor_b + 90 days (fixed horizon). **1 816** total 4H interactions.
-  `review_windows.yaml` written; artifacts in `experiments/review/source_fib_projection/`.
-  Summary: [`reviews/btc-1d-reaction-review-cycle-20260612.md`](reviews/btc-1d-reaction-review-cycle-20260612.md).
-  Outliers: Jan–Feb 2022 cluster (52–65 events); Dec 2020 breakout (2 events).
-  Batch script: `scripts/run_btc_1d_reaction_review.py`. 1H deferred (cache missing).
-- **2026-06-11 1D source-fib labeling complete (source-facit only)** — **67** manual
-  BTC/USD 1D source fibs drawn and verified (timeframe `1d`, log scale,
-  `tradingview_log_chamoun`, levels `[0, 0.382, 0.5, 0.618, 0.786, 1]`, no 0.236,
-  endpoint mapping ratio 0.0=anchor_b / 1.0=anchor_a, anchor direction, log-spacing,
-  human/manual only). Coverage **2017-01-05 → 2024-12-20**; **34 down / 33 up**.
-- **2026-06-11 1W source-fib phase complete** — **21/21** BTC/USD manual 1W source
-  fibs drawn (log scale, `tradingview_log_chamoun`) and verified (profile, scale,
-  levels `[0, 0.382, 0.5, 0.618, 0.786, 1]`, no 0.236, anchor direction,
-  human/manual only). Two new research modules:
-  - `research/weekly_source_fib_map` — combined 1W/1D/4H map (all fibs per TF).
-    **1W/1D usable; combined 4H too compressed** for candle-level confirmation.
-  - `research/weekly_source_fib_zoom` — per-fib windowed 4H confirmation
-    (one chart per fib, A→B + bounded context). **Per-fib 4H usable.**
-  Strict separation kept: **1M source**, **1M→1W projection** (`weekly_projection_map`),
-  and **true 1W source** fibs are distinct flows; fail-closed guards reject any
-  non-1W / non-log / wrong-profile / 0.236 / non-human fib. No auto-fib, no
-  trading conclusions. Commits `4eb2f4b` (map + facit), `939de97` (zoom),
-  `e379fae` (CI: format + bounds fix).
+- **2026-06-11→06-12 source-fib milestones (archived)** — 1W (21), 1D (67) + 4H
+  (366→365) source phases complete, 1D 4H reaction-review (1816 interactions), 4H Tier 1
+  maps. Full detail: [log post-reset part 1](log-archive-btc-postreset-part1.md); current
+  counts in Verification Snapshot below.
 - **2026-06-08…11 (earlier milestones)** — 1M reaction-review cycle (9 fibs, 1D+4H),
   Addendum 2 golden-zone retirement (issue #30, added `human_highlights`), log-scale +
   profile fix, 1M re-label, events log-scale fix, and the BTC monthly-first reset. Full
@@ -142,7 +117,9 @@ confluence is **geometry, not edge proof** → **stop the MTF track**. **Fork wr
 (docs-only):** [Phase 0 prereg](reviews/btc-fib-to-genesis-v2-phase0-prereg-20260615.md) registers
 one falsifiable behaviour question (causal confluence zones vs placebo/naïve levels, OOS) +
 leakage manifest, baselines, time-split holdout, stop/go; [Phase 1 spec](reviews/btc-fib-to-genesis-v2-phase1-feature-export-spec-20260615.md)
-(zone+bar tables, `known_after_ts` rule) **reviewed PASS — Phase 1 closed as docs-only contract**. **Phase 2 (dummy-file test) NOT authorised** without an explicit go.
+(zone+bar tables, `known_after_ts` rule) **reviewed PASS — Phase 1 closed as docs-only contract**. **Phase 2 (dummy-file test) — DONE 2026-06-16** (authorised narrow slice):
+`research/feature_contract.py` + committed dummy CSVs prove the contract validates mechanically
+with no real export and no Genesis touch ([Phase 2 report](reviews/btc-fib-to-genesis-v2-phase2-dummy-contract-20260616.md)). **Stopped** — any real export or Genesis touch needs a fresh go.
 
 **Deferred:** 1H source labeling — 4H is the lowest active timeframe; fetch 1H cache first
 (`data.fetch --timeframes 1h`). Separate decision before starting.

docs/research_wiki/log-archive-btc-postreset-part1.md

@@ -0,0 +1,88 @@
+# Research Wiki Log — BTC post-reset archive (part 1)
+
+Archived entries from [log.md](log.md) to keep the active log within its size bound.
+Covers the **2026-06-11 → 2026-06-12** source-fib completion milestones (1M reaction-review,
+1W, 1D, 4H source phases + 4H Tier 1 design/maps). Append-only; do not edit.
+
+## [2026-06-12] review | BTC/USD 4H visual confirmation Tier 1 — annual source-fib maps built
+
+Implemented `research/fourh_source_fib_map.py` (Tier 1 of the locked design): annual
+combined 4H source-fib maps, fibs grouped by `anchor_a` year, dense years (>60 fibs)
+split into calendar half-years. Reuses `_draw_map` / `_load_fibs` / `_nearest_pos` /
+`_short_id` from `monthly_fib_map` unchanged; no snap (source TF == chart TF == 4h);
+each group windowed by its fib span `[min(anchor_a) − pad, max(anchor_b) + pad]` (not
+calendar boundaries, so a December fib whose anchor_b crosses into the next year still
+renders). Fail-closed guard adapted to `SOURCE_TF="4h"` (timeframe/profile/scale/0.236/
+human-manual/no candidate-auto-inferred). 14 tests in `tests/research/test_fourh_source_fib_map.py`;
+ruff + repo-bounds + full suite pass (one pre-existing flaky `test_synthetic_ohlcv_high_ge_low`,
+untouched).
+
+**Run finding (real facit + cache, expansion config):** all **366/366 fibs drawn, 0
+skipped**, across **11 groups** — 2017 split 13 (h1) / **103 (h2)**, 2018=33, 2019=26,
+2020=31, 2021=55, 2022=24, 2023=17, 2024=22, 2025=34, 2026=8. Mid-density maps (e.g.
+2022, 2019) are cleanly scannable. **2017_h2 (~103 fibs, the Sep–Dec parabolic run)
+exceeds map-reviewable density → flags 2017 for Tier 2 `fourh_source_fib_zoom`.** That
+is the Tier-1 deliverable's signal (per design: build Tier 2 only where Tier 1 shows
+per-fib zoom is needed). No reaction-review, no events, no trading conclusions. Output
+under `experiments/review/fourh_source_fib_map/` (gitignored).
+
+## [2026-06-12] decision | BTC/USD 4H visual confirmation / source-quality review — design locked
+
+4H is the lowest active timeframe (1H paused). 4H source-facit locked (366 fibs,
+2017-01-05 → 2026-06-05, up=169/down=197, log scale, `tradingview_log_chamoun`,
+`[0, 0.382, 0.5, 0.618, 0.786, 1]`, no 0.236, human/manual). Next phase: **4H visual
+confirmation / source-quality review** — not reaction-review, not 1H.
+
+**Tier 1 (first implementation):** `fourh_source_fib_map.py` — annual combined 4H candle
+maps, fibs grouped by `anchor_a` year; ~10 charts (~20 PNGs); fast source-quality scan
+over 366 fibs without per-fib overhead. Fail-closed: `timeframe==4h`, log,
+`tradingview_log_chamoun`, no 0.236, human/manual, no candidate/auto/inferred.
+**Tier 2 (on-demand, after Tier 1):** `fourh_source_fib_zoom.py` — per-fib windowed 4H
+charts. Build only if Tier 1 shows per-fib zoom is needed.
+
+Reactive modules (`source_fib_projection_review`, `source_fib_projection_chart`) are NOT
+used — no events, no `review_sample.csv`, no interactions, no trading conclusions. Full
+design:
+[btc-4h-visual-confirmation-design-20260612.md](reviews/btc-4h-visual-confirmation-design-20260612.md).
+
+## [2026-06-12] review | BTC/USD 4H source-fib phase complete — 366 fibs
+
+366 manual 4H source fibs drawn and verified: timeframe `4h`, log scale,
+`tradingview_log_chamoun`, levels `[0, 0.382, 0.5, 0.618, 0.786, 1]`, no 0.236,
+endpoint mapping (ratio 0.0=anchor_b / 1.0=anchor_a), anchor direction, log-spacing,
+human/manual only. Coverage **2017-01-05 → 2026-06-05**; **up=169 / down=197**.
+366/366 schema verification PASS (0 failures). This is **source-labeling completion,
+not reaction-review** — visual confirmation / reaction-review is a later, separate
+decision. No auto-fib, no trading conclusions. Separation preserved across 1M source /
+1M→1W projection / true 1W source / true 1D source / true 4H source fibs.
+
+## [2026-06-11] review | BTC/USD 1D source-fib labeling complete (source-facit) — 67 fibs
+
+67 manual 1D source fibs drawn and verified: timeframe `1d`, log scale,
+`tradingview_log_chamoun`, levels `[0, 0.382, 0.5, 0.618, 0.786, 1]`, no 0.236,
+endpoint mapping (ratio 0.0=anchor_b / 1.0=anchor_a), anchor direction, log-spacing,
+human/manual only. Coverage **2017-01-05 → 2024-12-20**; **34 down / 33 up**. This is
+**source-labeling completion, not reaction-review** — reaction-review / visual
+confirmation is a later, separate phase (not required for source completion). No
+auto-fib, no trading conclusions. Separation preserved across 1M source / 1M→1W
+projection / true 1W source / true 1D source fibs.
+
+## [2026-06-11] review | BTC/USD 1W source-fib phase complete — 21 fibs, map + per-fib 4H zoom
+
+21/21 manual 1W source fibs drawn (log scale, `tradingview_log_chamoun`) and verified
+(profile, scale, levels `[0, 0.382, 0.5, 0.618, 0.786, 1]`, no 0.236, anchor direction,
+human/manual only). Added `research/weekly_source_fib_map` (combined 1W/1D/4H — 1W/1D
+usable, combined 4H too compressed) and `research/weekly_source_fib_zoom` (per-fib
+windowed 4H — usable). Strict separation kept between 1M source, 1M→1W projection
+(`weekly_projection_map`), and true 1W source fibs; fail-closed guards reject non-1W /
+non-log / wrong-profile / 0.236 / non-human fibs. No auto-fib, no trading conclusions.
+Commits `4eb2f4b`, `939de97`, `e379fae`.
+
+## [2026-06-11] review | BTC/USD 1M reaction-review cycle complete — all 9 source fibs
+
+All 9 human-drawn 1M source fibs reviewed through 1D + 4H using
+`source_fib_projection_review` + `source_fib_projection_chart` (log scale,
+`tradingview_log_chamoun`). Review windows confirmed in `review_windows.yaml`
+(anchor_b → next macro boundary; 20260101 window extends to latest cache 2026-06-08).
+Total: 62 1D events, 127 4H events across the full set.
+Summary artifact: [reviews/btc-1m-reaction-review-cycle-20260611.md](reviews/btc-1m-reaction-review-cycle-20260611.md).

docs/research_wiki/log.md

@@ -11,9 +11,23 @@ Use headings like:
 Types: `ingest`, `decision`, `review`, `question`, `maintenance`.
 
 
-> Older entries (2026-06-10 and earlier): [part 3](log-archive-pre-btc-reset-part3.md) →
+> Older entries (2026-06-11→06-12 source-fib milestones):
+> [post-reset part 1](log-archive-btc-postreset-part1.md).
+> Pre-reset (2026-06-10 and earlier): [part 3](log-archive-pre-btc-reset-part3.md) →
 > [part 2](log-archive-pre-btc-reset-part2.md) → [part 1](log-archive-pre-btc-reset-part1.md)
 
+## [2026-06-16] feat | Fib → Genesis V2 — Phase 2 dummy contract test (narrow slice)
+
+Mechanical contract/dummy test **inside the Fib repo only** — not export, not Genesis
+integration. New stdlib-only `research/feature_contract.py` validates two synthetic dummy CSVs
+(committed under `reviews/contracts/phase2_dummy/`: 3 zones, 4 bars incl. a multi-zone row)
+against the Phase 1 schema: exact-header schema, join keys `(symbol,timeframe,timestamp)`
+non-null + unique, causality `known_after_ts <= timestamp` over the whole reference set,
+knowability floor `known_after_ts >= max(anchor_b)+buffer`, 1H fail-closed, feature/metadata
+boundary asserted at import. No fib computation, no Genesis import, no pipeline/ML/backtest/
+signal/edge. 20 tests; ruff + 426 passed (76%) + repo-bounds green; CLI smoke OK. **Stop after
+this.** [Report](reviews/btc-fib-to-genesis-v2-phase2-dummy-contract-20260616.md).
+
 ## [2026-06-15] review | Fib → Genesis V2 — Phase 1 closed (PASS)
 
 Phase 1 feature-export spec reviewed — **verdict: PASS**; **closed as a docs-only contract**.
@@ -340,86 +354,6 @@ maps; Tier 2 per-fib zoom windows will be narrower and more readable.
 
 **Next:** implement Tier 2 `fourh_source_fib_zoom.py`.
 
-## [2026-06-12] review | BTC/USD 4H visual confirmation Tier 1 — annual source-fib maps built
-
-Implemented `research/fourh_source_fib_map.py` (Tier 1 of the locked design): annual
-combined 4H source-fib maps, fibs grouped by `anchor_a` year, dense years (>60 fibs)
-split into calendar half-years. Reuses `_draw_map` / `_load_fibs` / `_nearest_pos` /
-`_short_id` from `monthly_fib_map` unchanged; no snap (source TF == chart TF == 4h);
-each group windowed by its fib span `[min(anchor_a) − pad, max(anchor_b) + pad]` (not
-calendar boundaries, so a December fib whose anchor_b crosses into the next year still
-renders). Fail-closed guard adapted to `SOURCE_TF="4h"` (timeframe/profile/scale/0.236/
-human-manual/no candidate-auto-inferred). 14 tests in `tests/research/test_fourh_source_fib_map.py`;
-ruff + repo-bounds + full suite pass (one pre-existing flaky `test_synthetic_ohlcv_high_ge_low`,
-untouched).
-
-**Run finding (real facit + cache, expansion config):** all **366/366 fibs drawn, 0
-skipped**, across **11 groups** — 2017 split 13 (h1) / **103 (h2)**, 2018=33, 2019=26,
-2020=31, 2021=55, 2022=24, 2023=17, 2024=22, 2025=34, 2026=8. Mid-density maps (e.g.
-2022, 2019) are cleanly scannable. **2017_h2 (~103 fibs, the Sep–Dec parabolic run)
-exceeds map-reviewable density → flags 2017 for Tier 2 `fourh_source_fib_zoom`.** That
-is the Tier-1 deliverable's signal (per design: build Tier 2 only where Tier 1 shows
-per-fib zoom is needed). No reaction-review, no events, no trading conclusions. Output
-under `experiments/review/fourh_source_fib_map/` (gitignored).
-
-## [2026-06-12] decision | BTC/USD 4H visual confirmation / source-quality review — design locked
-
-4H is the lowest active timeframe (1H paused). 4H source-facit locked (366 fibs,
-2017-01-05 → 2026-06-05, up=169/down=197, log scale, `tradingview_log_chamoun`,
-`[0, 0.382, 0.5, 0.618, 0.786, 1]`, no 0.236, human/manual). Next phase: **4H visual
-confirmation / source-quality review** — not reaction-review, not 1H.
-
-**Tier 1 (first implementation):** `fourh_source_fib_map.py` — annual combined 4H candle
-maps, fibs grouped by `anchor_a` year; ~10 charts (~20 PNGs); fast source-quality scan
-over 366 fibs without per-fib overhead. Fail-closed: `timeframe==4h`, log,
-`tradingview_log_chamoun`, no 0.236, human/manual, no candidate/auto/inferred.
-**Tier 2 (on-demand, after Tier 1):** `fourh_source_fib_zoom.py` — per-fib windowed 4H
-charts. Build only if Tier 1 shows per-fib zoom is needed.
-
-Reactive modules (`source_fib_projection_review`, `source_fib_projection_chart`) are NOT
-used — no events, no `review_sample.csv`, no interactions, no trading conclusions. Full
-design:
-[btc-4h-visual-confirmation-design-20260612.md](reviews/btc-4h-visual-confirmation-design-20260612.md).
-
-## [2026-06-12] review | BTC/USD 4H source-fib phase complete — 366 fibs
-
-366 manual 4H source fibs drawn and verified: timeframe `4h`, log scale,
-`tradingview_log_chamoun`, levels `[0, 0.382, 0.5, 0.618, 0.786, 1]`, no 0.236,
-endpoint mapping (ratio 0.0=anchor_b / 1.0=anchor_a), anchor direction, log-spacing,
-human/manual only. Coverage **2017-01-05 → 2026-06-05**; **up=169 / down=197**.
-366/366 schema verification PASS (0 failures). This is **source-labeling completion,
-not reaction-review** — visual confirmation / reaction-review is a later, separate
-decision. No auto-fib, no trading conclusions. Separation preserved across 1M source /
-1M→1W projection / true 1W source / true 1D source / true 4H source fibs.
-
-## [2026-06-11] review | BTC/USD 1D source-fib labeling complete (source-facit) — 67 fibs
-
-67 manual 1D source fibs drawn and verified: timeframe `1d`, log scale,
-`tradingview_log_chamoun`, levels `[0, 0.382, 0.5, 0.618, 0.786, 1]`, no 0.236,
-endpoint mapping (ratio 0.0=anchor_b / 1.0=anchor_a), anchor direction, log-spacing,
-human/manual only. Coverage **2017-01-05 → 2024-12-20**; **34 down / 33 up**. This is
-**source-labeling completion, not reaction-review** — reaction-review / visual
-confirmation is a later, separate phase (not required for source completion). No
-auto-fib, no trading conclusions. Separation preserved across 1M source / 1M→1W
-projection / true 1W source / true 1D source fibs.
-
-## [2026-06-11] review | BTC/USD 1W source-fib phase complete — 21 fibs, map + per-fib 4H zoom
-
-21/21 manual 1W source fibs drawn (log scale, `tradingview_log_chamoun`) and verified
-(profile, scale, levels `[0, 0.382, 0.5, 0.618, 0.786, 1]`, no 0.236, anchor direction,
-human/manual only). Added `research/weekly_source_fib_map` (combined 1W/1D/4H — 1W/1D
-usable, combined 4H too compressed) and `research/weekly_source_fib_zoom` (per-fib
-windowed 4H — usable). Strict separation kept between 1M source, 1M→1W projection
-(`weekly_projection_map`), and true 1W source fibs; fail-closed guards reject non-1W /
-non-log / wrong-profile / 0.236 / non-human fibs. No auto-fib, no trading conclusions.
-Commits `4eb2f4b`, `939de97`, `e379fae`.
-
-## [2026-06-11] review | BTC/USD 1M reaction-review cycle complete — all 9 source fibs
-
-All 9 human-drawn 1M source fibs reviewed through 1D + 4H using
-`source_fib_projection_review` + `source_fib_projection_chart` (log scale,
-`tradingview_log_chamoun`). Review windows confirmed in `review_windows.yaml`
-(anchor_b → next macro boundary; 20260101 window extends to latest cache 2026-06-08).
-Total: 62 1D events, 127 4H events across the full set.
-Summary artifact: [reviews/btc-1m-reaction-review-cycle-20260611.md](reviews/btc-1m-reaction-review-cycle-20260611.md).
+> **2026-06-11→06-12 entries** (1M reaction-review, 1W/1D/4H source phases, 4H Tier 1
+> design/maps) archived to [post-reset part 1](log-archive-btc-postreset-part1.md).