OCTAVIAN APEXGLOBAL DECISION INTELLIGENCE
RESEARCH · 2026-09-06

Confidence as a predictor

FALLEN   Fallen — broke its own criteria; kept on purpose

Hypothesis and result

The engine's confidence score does not separate winners from losers: AUC 0.498. Every confidence band and every threshold is net negative, and the gradient is not monotonic.

No threshold adjustment can help a score that carries no information; the fix has to be in the inputs, not the cut-off.

The evidence card

Sample465 trades · 57 days of data
EffectAUC 0.498 (p = 0.94), n = 465
Half-split testFails — the sign does not hold in both halves.
Multiple-testing statusp = 0.944. Judged against the whole family of 276 hypotheses Apex has logged (Bonferroni threshold 0.00018): does not clear it on its own — which is why forward shadow evidence is required before promotion.
StageFallen — broke its own criteria; kept on purpose
Re-measuredRe-measured when its source script is run — no schedule · last 2026-09-06

What would invalidate it

How to read this page

Every number describes the past under the stated conditions. A finding that passes today can fall tomorrow; Apex keeps fallen findings visible on purpose. Stages, tests and the family-wise correction are described in the methodology.

All research · Learn · The Apex day

Octopod Holding AS · Octavian Apex is a research and decision-support tool. Nothing on this site is investment advice, and no result shown is a promise of future performance. Understanding markets · Markets · Deep dive · Market intelligence · About · Methodology · Research · Learn · The Apex read · The Apex day · Terms · Privacy · invest@apexgdi.com