Evidence · Research

How we reject weak ideas

Failed and unfinished hypotheses stay visible. A historical research PASS is not a production return claim. Forward tests remain in progress until sample thresholds are met.

Broad campaign portfolio

Does capacity-aware allocator restore production claim?

Did not survive production-like validation

Capacity-aware allocator did not restore a commercial production claim.

Advanced / auditor disclosure

Internal id: CW_ALLOCATOR
Report date: 2026-08
Holdout: closed
Raw verdict: NOT REPLICATED

Exact production T+1

Does exact T+1 replicate claimed monthly expectancy?

Hypothesis rejected

Exact T+1 portfolio path did not replicate claimed monthly expectancy.

Advanced / auditor disclosure

Internal id: CT_EXACT_T1
Report date: 2026-08
Holdout: closed
Raw verdict: FAIL

Actor persistence

Historical actor research finding after PIT?

Hypothesis rejected

No monetizable historical actor research finding after PIT controls.

Advanced / auditor disclosure

Internal id: DE_ACTOR_EDGE
Report date: 2026-08-05
Holdout: closed
Raw verdict: FAIL

Volume anomaly

Insider × volume anomaly?

Hypothesis rejected

No volume-anomaly edge on chronological holdout.

Advanced / auditor disclosure

Internal id: DG_VOLUME
Report date: 2026-08-05
Holdout: closed
Raw verdict: FAIL

Selective ML

Selective quality gate on PIT campaigns?

Hypothesis rejected

Selective quality gate was not a monetizable holdout product change.

Advanced / auditor disclosure

Internal id: CZ_SELECTIVE_PIT
Report date: 2026-08
Holdout: closed
Raw verdict: FAIL

Momentum M3

Insider × momentum interaction?

Historical research result (not a live return claim)

M3 chronological holdout validated historically; not a production-return claim.

Advanced / auditor disclosure

Internal id: DF_MOMENTUM
Report date: 2026-08-05
Holdout: unopened until freeze
Raw verdict: HISTORICAL HOLDOUT PASS

Historical replay factory

Can historical campaigns be replayed as-of from recovered raw sources with unified proof?

Historical V1 — INVALID FOR PERFORMANCE CLAIMS

Replay architecture proven with source limitations; finalized in DK.

Preserved legacy research artifact. Superseded by Historical V2 RC2. Do not treat as a currently valid performance PASS.

Advanced / auditor disclosure

Internal id: DJ_HISTORICAL_EVIDENCE_FACTORY
Report date: 2026-08-05
Holdout: reuse DF holdout labels only
Raw verdict: WATCH

Historical Evidence V1 freeze

Freeze canonical historical universe and reconcile CX/DF

Historical V1 — INVALID FOR PERFORMANCE CLAIMS

strict historical replay universe and metrics frozen as historical-v1.

Preserved legacy research artifact. Superseded by Historical V2 RC2. Do not treat as a currently valid performance PASS.

Advanced / auditor disclosure

Internal id: DK_HISTORICAL_EVIDENCE_FINALIZATION
Report date: 2026-08-05
Holdout: frozen
Raw verdict: FROZEN

Momentum forward

Does M3 retain edge under Grade A forward capture?

Forward test in progress

Forward Grade A capture and Momentum Shadow are collecting evidence. No forward performance claim.

Advanced / auditor disclosure

Internal id: FORWARD_GRADE_A
Report date: 2026-08-05
Holdout: collecting
Raw verdict: COLLECTING

Evidence overview · Methodology · Archive