AV2-D-01 — A2.10 Validation-vs-Holdout divergence

av2 / AV2-D-01

preregistered challengerPRE-REGISTERED

Pre-registered DIAGNOSTIC only: it may generate a future hypothesis and can never modify a champion. No result exists.

Lifecycle

current

Outcome

—

Phase

none

Research gate

—

Engine

preregistration only

Market

—

Timeframe

—

Side

—

Tracking

Family, lineage and what this record needs next

Mechanism family

Unclassified

Mechanism parent

None declared.

Phase detail

Pre-registered only — no phase entered.

Next required test

Its pre-registered ladder's first stage (no stage has run).

Integrity & certification

Signals as recorded; NOT_PASSED means not passed or not applicable

Engine certified

unknown

Execution certified

unknown

Integrity audit

unknown

Shadow verifier

unknown

Technical invalidation

None recorded.

Artifacts

Everything committed for this record

AV2 pre-registration

research-output/av2-planning/AV2_1_PREREGISTRATION.md

Evidence recorded

Nothing has been run — no metric may be shown for a pre-registered item

No stage evidence exists for this entry. It is pre-registered only: nothing has been run, and no metric may be shown for it.

Read from research-output/av2-planning/AV2_1_PREREGISTRATION.md

Lineage

Declared in the pre-registration

Relation to prior work

preregistered challenger

Note

why does Holdout expectancy (+0.260R, n=100) diverge materially from Validation (+0.409R, n=88)? This is the exact unresolved risk the Final Research Audit cited when it withheld A2.10 from Phase B, and it is the highest-value open question across both champions

Pre-registration

Declared before any result was seen; no result exists yet

Declared change

Alters

—

Declared overfitting risk

—

Constraint

DIAGNOSTIC ONLY: it may generate a future hypothesis and can never modify a champion without its own separate pre-registration.

Read from research-output/av2-planning/AV2_1_PREREGISTRATION.md