# H04_E_MECHANICAL_CONFOUND_FREEZE_v0_1

**Created:** 2026-09-13T23:10:48Z  
**Artifact:** `H04_E_MECHANICAL_CONFOUND_FREEZE_v0_1.json`  
**sha256:** `bcd6867f5c4fdac9ad26d3ab1daf5b4de0bb311aac0efff3e255f21a238c274d`  
**Hub job:** `77356229-193e-4693-a90c-cae8dcb0bc4a`  
**Parent:** H04-E EXPLORATORY_PROMOTE (freeze sha `575470bd…`; result sha `a268ee06…`; ~34.25% holdout RMSE)

## STATUS
| Field | Value |
|-------|-------|
| STATUS | EXPLORATORY |
| VINTAGE_STATUS | UNVERIFIED/INCOMPLETE |
| CLAIM_STRENGTH | LOW |
| CAPITAL_USE | PROHIBITED |
| MODEL_CHANGE | NO |
| REAL_CAPITAL | PROHIBITED |
| NEW TARGETS | NO (reuse H04-E unit_table COSXPUS revisions) |

## Purpose
Cheap mechanical confound screen: is the H04-E lift mostly **STEO copies WPSR/weekly stocks**, or does the surprise feature add incremental information?

## Locked primary check (ONE) — C1_NESTED_MECH_RAW_DELTA
Same sample/split/target/metric as H04-E.

| Model | Spec |
|-------|------|
| A0 | H04-E baseline A (intercept + dest-month + prior revision) |
| A_mech | A0 + `wcestus1_mean_raw_delta_kb` (mean admitted inter-issue weekly Δstocks; **not** seasonally demeaned) |
| B_vs_mech | A_mech + `wcestus1_surprise_mean_inter_issue` (original H04-E feature) |

**Primary metric:** relative holdout RMSE improvement of B_vs_mech vs A_mech.

### Disposition map (incremental over mechanical)
| Outcome | Rule |
|---------|------|
| EXPLORATORY_PROMOTE_SURVIVES_NONMECHANICAL | incremental % ≥ 5.0 (CLAIM_STRENGTH still LOW; not capital) |
| HOLD | [1%, 5%) |
| MECHANICAL | <1% or negative / N<8 / zero mech variance |

Diagnostic only (not decision): A0→A_mech lift (how much raw delta alone captures).

## Explicitly out of scope
No capital, no MODEL_CHANGE, no vintage repair, no archive dig, no H01 reopen, no secondary feature tournament in this freeze.

## Next
1. Hash-lock this freeze.
2. Build `wcestus1_mean_raw_delta_kb` on the existing admitted windows; run C1 once.
3. Post result + prospective protocol sketch (design only).
4. Hub comment + /reply chatgpt+all. Never /james.