# AvLu Automated Regression QA v4.0.0

## Audited coverage

| Metric | Result |
| --- | ---: |
| Synthetic scenario packets audited | 2,000 |
| Assertions executed per packet | 24 |
| Automated QA assertions executed | 48,000 |
| Assertions passed | 48,000 |
| Regression pass rate | 100% |
| Scenario types covered | 6 |

Every packet passed the 24 named checks. The checks cover synthetic identity labeling, source classification, VIN-token exclusion, required dates and scope, positive cost fields, estimate arithmetic, simulated closeout arithmetic, variance arithmetic, formula impact, protected-maximum reconciliation, packet completeness, and closed dealer-outcome flags.

## Formula coverage

The matrix runs one check per packet in each of these five formula categories:

- AvLu estimate reconciliation
- Simulated closeout reconciliation
- Variance reconciliation
- Formula-impact reconciliation
- Revised protected-maximum reconciliation

## Evidence boundary

Automated regression QA validates deterministic synthetic data integrity and formula behavior. It does not establish predictive accuracy, dealer acceptance, repair invoices, actual repair outcomes, purchase approval, or realized profit.

The dealer-validation campaign is separate and still requires all nine evidence gates. No synthetic packet is a dealer-verified outcome.

## Machine-readable results

The complete assertion catalog, dataset fingerprint, scenario coverage, and 2,000 record-level results are available in `qa-regression-matrix-v4.0.0.json`.

© 2026 AVLU. All rights reserved.