Reading summary
The result page surfaces one of three readings. The reading is one input to the closed five-tier verdict; the verdict layer combines the reading with the direction of the better-fitting model. The lookup below records the closed vocabulary in declaration order.
| Reading |
Reading key |
What it says |
| Below band |
below_band |
Additive model fits better by a margin clear of the public lower threshold. |
| Inside tie band |
inside_tie_band |
Multiplicative and additive fits sit at similar quality; no structural winner on this signal alone. |
| Above band |
above_band |
Multiplicative model fits better by a margin clear of the public upper threshold. |
Below band
If your result page reads below band, the diagnostic found that the additive model fits your data better than the multiplicative one by a margin large enough to clear the public lower threshold.
How it relates to the verdict
The reading is one input to the verdict, not the verdict itself. The verdict layer combines the reading with the direction of the better-fitting model to produce one of the closed five-tier verdicts.
What this means in practice
An additive composition is a clean structural signal. The diagnostic is fit for review under SR 11-7, PRA SS1/23 Principle 3.4, EU AI Act Article 9, ECB IM Guide, and APRA CPG 220 with this reading; the validation report records it explicitly.
Inside tie band
If your result page reads inside tie band, the multiplicative and additive fits sit between the public lower and upper thresholds. The two competing models fit your data at similar quality; the diagnostic does not declare a structural winner on this signal alone.
How it relates to the verdict
The reading is one input to the verdict, not the verdict itself. The verdict layer combines the reading with the direction of the better-fitting model to produce one of the closed five-tier verdicts.
What this means in practice
A tie reading is a directional finding in itself. It tells the buyer that neither composition dominates on this dataset, and the next step is the lens stack on the result page rather than a strong structural claim.
Above band
If your result page reads above band, the diagnostic found that the multiplicative model fits your data better than the additive one by a margin large enough to clear the public upper threshold.
How it relates to the verdict
The reading is one input to the verdict, not the verdict itself. The verdict layer combines the reading with the direction of the better-fitting model to produce one of the closed five-tier verdicts.
What this means in practice
A multiplicative composition is the canonical IADTX signal. The diagnostic is fit for review under SR 11-7, PRA SS1/23 Principle 3.4, EU AI Act Article 9, ECB IM Guide, and APRA CPG 220 with this reading; the validation report records it explicitly.
What this page does not disclose
The numerical band edges are part of the locked private calibration and are not disclosed on this page. The reading vocabulary is closed and exhaustive on the public surface; no other readings exist.
