SniperGold_ML/docs/PROJECT_STATE_TRANSITION_MAP.md

65 lines
2.9 KiB
Markdown

# PROJECT STATE TRANSITION MAP
```text
Document type : GOVERNANCE
DOCUMENT ROLE : HISTORICAL_TRANSITION
Created : 2026-08-28 (P3-GOV-1)
Revision : v2 (2026-08-28 P3-GOV-3)
AUTHORITY : HISTORICAL / ARCHITECTURAL TRANSITION DESCRIPTION ONLY. This map
MUST NOT authorize execution and MUST NOT independently define
the current ingestion cursor. It distinguishes LEGACY / REVISED /
CURRENT / PENDING for orientation.
```
## LEGACY / HISTORICAL (frozen reference; NOT current evidence)
- Legacy `SniperGold_ML.mqh` + old directional-drift MLP
- `repaired/post_purge/SniperGold_ML_postpurge.mqh`, `repaired/pre_purge/...`
- FROZEN LEGACY / REFERENCE
- AlgoForge engine estate (README legacy, `MQL5\AlgoForge\*`)
- FROZEN LEGACY / REFERENCE
- P3-S20 walk-forward pooled 0.5792 (OLD / faulty population)
- FROZEN HISTORICAL RESULT (do not erase; not the current conclusion)
- OLD research M30 construction (index-pair resample) — pre-P3-S22.3
- replaced / superseded by the corrected basis
## REVISED / CORRECTED
- P3-S22.3 corrected UTC-clock research M30 construction
- CORRECTED RESEARCH BASIS (frozen)
- P3-S22.4 corrected Candidate-Setup population (707 / 610 leads / 579 binary-fit)
- VERIFIED corrected population (frozen)
- P3-S23 corrected-population pre-registered ML baseline (pooled 0.5235)
- CORRECTED-POPULATION BASELINE (frozen result)
## CURRENT AUTHORITATIVE
- P3-S24 conclusion: D — CURRENT EDGE NOT DEMONSTRATED (current conclusion)
- Real-source ingestion run P3_S251_REAL_TICK_RUN_001 in progress, chunks 0-10
committed, next chunk 11 @ byte offset 163,578,081 (authoritative next offset)
- P3-S25.1C: SAFE_MAX_CHUNK_SIZE = 24 MiB; chunk-size invariance proven;
SAFE_CHUNKS_PER_TURN = NOT ESTABLISHED (P3-S25.1D multi-chunk benchmark OPEN)
- P3-S25.1D-R: interruption verifier defect isolated / fix verified (isolated
path only); P3-GOV-2 repository truth audit (B) reconciled by P3-GOV-2R
- P3-GOV-3: governance control-plane hardening (authority model, state vs
authorization separation, control-plane consistency suite) — 2026-08-28
- External Tickstory basis validation IN PROGRESS (NOT yet demonstrated)
## PENDING
- Full 34.5 GB tick -> M15/M30 oracle (continuing chunk-by-chunk)
- External Candidate Population parity / replay
- Dataset expansion to the P3-S24 power targets (~1,300-1,700 leads) — only
after a validated basis
- Post-validation ML (only with explicit authorization)
- Final certification of the full dataset (P3-s25-certification plan)
```
Transition notes:
LEGACY (old population / baseline) -> CORRECTED (UTC M30 / corrected population)
CORRECTED has NO demonstrated edge (P3-S24 D) -> PENDING external basis validation.
```
*End of P3-GOV-3 state transition map (HISTORICAL_TRANSITION). The P3-GOV-2R note
about normalization of "NOT SET" to "NOT ESTABLISHED" is a terminology clarification
applied during P3-GOV-3; science is unchanged.*