An appraisal: $612,000. Confidence 0.91.
9.05 joules.

An appraiser is engaged under USPAP 2024-25, Standards Rule 1, with an intended use of "mortgage finance" and an effective date of 2026-05-07. The engagement letter pins the scope of work, the property rights appraised (fee simple), and the report option (Appraisal Report, SR 2-2(a)).

The 1,840 sqft structure is encoded as a UAD 2.0 instance (Fannie Mae / Freddie Mac dynamic appraisal data model, MISMO XML v3.6): 14 site fields, 38 dwelling fields, 6 site improvements, 4 utility connections. Every field maps to a UAD definitional ID.

Comp set NCRE-12839: five sales within 1.4 mi, all closed in the last 184 days, gross living area within 22% of subject. Each comp is itself a UAD 2.0 record. The selector logs every candidate it considered and why each accepted comp dominated the rejected ones.

The metric engine computes Mahalanobis distance over 11 normalized features (GLA, lot size, year built, baths, garage bays, view tier, condition C1–C6, quality Q1–Q6, etc.) using the inverse covariance of the local market sample (n=482, 90-day window). Distances: 0.31, 0.44, 0.62, 0.81, 1.07.

The inspector files 47 assertions under InterNACHI Standards of Practice 2024: 31 pass, 14 advisory, 2 material. Each assertion carries a photo hash (sha256), a confidence ∈ [0,1], and an InterNACHI SOP section reference. Average confidence: 0.87; median: 0.91.

Sale prices of the five comps adjust to the subject via published Fannie Mae adjustment guidance: +/-$48/sqft GLA, +$3,200 per bath, +$5,800 per garage bay. Reconciliation weights the five adjusted indications by 1 / Mahalanobis distance (normalized to sum to 1).

Weighted reconciliation lands at $611,840, rounded to $612,000 per Fannie Mae practice. Confidence 0.91 is the posterior probability the true value lies within ±5%, given the distance-weighted indications and the inspector's material findings.

The manifest records: UAD instance hash, comp-set ID NCRE-12839, market-sample window (2026-02-07–2026-05-07), inverse-covariance matrix hash, adjustment-guide version (SEL-2024-04), and the ExpertOS distance-engine git SHA (a3f9c2e). Anyone re-runs and gets $612,000 again.