Deep dive — chart 3

Chart 3 — lane rate per applicant group, baseline vs recent

The chart with the operational + regulatory bite. Where score-distribution shifts translate into actual application routing. H3 decline / non-QM rate rose 14% → 61%. Other applicant groups stable. That is the exact disparate-impact signature regulators look for.

Underwriting-lane rate per applicant group, baseline vs recent

Underwriting-lane rate per applicant group — baseline (left) vs recent (right). H3 decline / non-QM rate rose sharply while other applicant groups stayed stable.

What you are looking at

Grouped bars per applicant group (H1 / H2 / H3 / H4 / H5)
Three colors per applicant group: Auto-Approve Conforming (green) / Manual Review + Upcharge (blue) / Decline / non-QM (red)
Left cluster: baseline (first 30 days) — the "normal" routing per applicant group
Right cluster: recent (last 30 days) — the current routing per applicant group

What the chart shows

H1, H2, H4 — lane rates unchanged between baseline and recent.
H5 — modest shift; secondary signal from the self-employed distribution divergence.
H3 — decline / non-QM rate rises sharply (14% baseline → 61% recent). Auto-approve rate shrinks. Manual-review absorbs part of the shift.
The concentrated differential is H3. Other applicant groups are largely stable.

47-percentage-point differential in decline / non-QM routing. One applicant group. Silently. Over 45+ days. This is the exact operational pattern the Fair Housing Act disparate-impact analysis, ECOA effects-testing, and CFPB fair-lending exam are calibrated to detect.

Why aggregate metrics missed it

Because H3 is one applicant group among five, the aggregate approval rate barely moved: the H3 shift rounded to noise across the whole pipeline. That is why the origination dashboard stayed green. Aggregate metrics are inherently blind to group-differential patterns, and group-differential patterns are inherently what disparate-impact analysis is built to surface.

The signal that fired the A/B/C determination

Once any group-differential signal fires OR ≥ 3 high-severity events land, the Snapshot lands on Category C by rule — no matter what any other metric shows. This sample has 10 high-severity events and a clear H3 pattern. Category C is not marginal.

See the A/B/C decision logic →

The exhibit-quality read

This chart is what a plaintiff's expert witness, state banking-department examiner, or DOJ fair-lending investigator asks the operator for first. It is also what the operator's GC should have on file before either lands — because the same chart, in the operator's hand, tells a different story:

Same chart, operator stance: "we identified this pattern through independent verification on Day X, initiated the following remediation on Day X+N, and here is the timeline of that remediation." The chart becomes evidence of good-faith investigation, not evidence of missed monitoring.

$499 Snapshot. 3 business days.

Same lane-shift chart for your lender's actual AI surface + full determination + 3 fix-first items.

Buy $499

Back to housing sample hub.