Confidence and limitations

Every fair-value figure on YieldIQ carries three confidence axes alongside it: data quality, model confidence, and valuation stability. The axes determine which verdict label the pill renders. When any axis is red, the verdict is gated to a non-committal label rather than forcing a confident-looking band that the inputs cannot defend.

Data quality

The data-quality axis tracks how complete and recent the underlying financials are. It turns amber when one or more required line items came from a fallback source, and red when a validator rejected an input — a unit-jump corruption, a stale annual filing past its tolerance window, or a failed cross-check against bhavcopy close.

Model confidence

The model-confidence axis tracks how well the chosen engine fits the business. A regulated utility scored through the utility engine carries higher model confidence than a new-economy listing routed through the generic engine because thin growth history forces the engine to lean on assumptions. Cohort routing (see sectors and cohorts) is the main driver of this axis.

Valuation stability

The valuation-stability axis measures how much the published FV has moved across the last several cache refreshes. A FV that has drifted within plus/minus five percent is stable; a FV that has swung twenty percent on each refresh is unstable and the axis turns amber. Instability is usually a symptom of thin or volatile inputs rather than a model defect.

Under Review

When the data-quality axis is red, the verdict pill renders Under Review. This is an explicit refusal to assign a band rather than a fallback guess. Recent IPOs with fewer than three years of post-listing financials, tickers with a failed validator on a required input, and companies in the middle of a unit-change correction will land here. The page still renders the partial Prism pillars where they are defensible.

Low Confidence

When the model-confidence or valuation-stability axes are amber but data quality is green, the verdict pill renders Low Confidence alongside its band label. The band is still computed and displayed; the label simply flags that the answer rests on a thinner foundation than usual.

When to lean on the number

The FV is most trustworthy when all three axes are green, the company sits in a well-modelled cohort (large-cap IT services, private-sector banks, mature FMCG), and the bear-to-bull spread is tight. The same FV is least trustworthy when an axis is amber, the cohort is thin (recent IPOs, complex holdcos), or the spread is wide. Consider the axes before the band.