A redlining scan is a map and a comparison. Where your lending lands, against who lives there, against what every other lender in the same market managed. HMDAVision puts that in front of you in an afternoon, with the filters recorded so the analysis can be reproduced and defended.
Ask it free at mortgagedata.aiSee plansFree tier is three answers a day, fifteen a month, no card.
Redlining is unequal access to credit based on where an applicant lives or where the property sits, rather than on the applicant.
That definition has a useful property. It is geographic, which means it is measurable from data you already report. Your loan application register records the census tract of every application. Census demographics record who lives in each tract. Every other lender in your market reports the same thing.
Put those three together and you can see your own pattern the way an examiner, a journalist, or a plaintiff's expert would see it. The analysis is not exotic. What has been missing is a way to run it yourself, repeatedly, without commissioning it.
Tracts inside your lending area where you took no applications at all, while others did.
A lower share of lending in majority-minority tracts than lenders of comparable size achieved in the same market.
A footprint that stops at a line matching racial composition rather than county, licensing, or branch geography.
Higher denial rates in particular tracts, especially denials citing insufficient collateral.
Rate-spread loans landing disproportionately in minority tracts, which is where reverse redlining appears.
A distance from peers that grows across HMDA releases rather than holding steady or closing.
One metro. Every census tract shaded by minority population share. Your originations as points, then the same market with peer originations. This single comparison answers most of a redlining question before any statistics are run.
1,847 loans, 11.6% in majority-minority tracts
Comparable lenders, same market, 19.4% in the same tracts
The left map has a visible edge. Lending thins as the shading darkens and stops almost entirely in the southwest quadrant. The right map covers the same ground evenly. Neither lender did anything different in the pale tracts.
This is what an examiner sees first, and it is what a reporter sees first. The difference between the two maps is the finding. Everything after it is quantification.
Illustrative figures shown to make the output legible. Your version is built from your own loan application register against tract demographics for the markets you actually lend in, with every filter recorded.
Comparison sets are built by volume, not by affinity. The convention is every lender originating between 50% and 200% of your HMDA volume in the same market. A peer group you assembled yourself, however reasonable, has drawn criticism in preliminary findings letters.
Two rules catch people out. Purchased loans, reported under action taken code 6, are excluded both from majority-minority tract counts and from the volume formula that builds the set. And strong numbers across majority-minority tracts do not settle the question if few of those tracts are also majority-Black or majority-Hispanic, which recent analyses have looked at separately.
This is the whole scan. Each step is a set of filters in HMDAVision, and each produces something you can put in front of a board, a regulator, or a production meeting.
Your assessment area, field of membership, or the geography your lending and marketing genuinely reach. This is the step that decides everything downstream, and the one lenders most often get wrong by drawing it too narrowly.
Output: a defined market with a documented rationale.
Filter to lenders at 50% to 200% of your originations in that market, excluding purchased loans. Not the competitors you think about. The ones the arithmetic selects.
Output: a defensible comparison group you did not curate.
Both maps, side by side, at tract level. Look for edges, holes, and boundaries that track demographics rather than geography or licensing.
Output: the visual that starts every conversation.
Your share of lending in majority-minority, majority-Black and majority-Hispanic tracts against peers and against the market aggregate. Then convert the percentage gap into a loan count.
Output: a number a board understands without translation.
Denial rates, denials citing insufficient collateral, and higher-priced originations by area. Lending there is not sufficient if the terms differ once you do.
Output: whether the issue is access, outcome, or price.
Year, geography, loan type, metric, comparison group. Keep the selections so the analysis can be reproduced and reviewed by someone who was not in the room.
Output: a trend line, and an audit trail.
Fair lending analysis has traditionally arrived as a deliverable. Someone runs it, a report lands, and the findings are true as of a date that has already passed. That model made sense when the data was hard to reach. It is not any more.
Runs once or twice a year
Findings arrive after the quarter closes
Methodology sits with the vendor
Answering a follow-up question means another engagement
Compliance receives it. Production rarely sees it.
Run it whenever you want to know
Check a market before you enter it, not after
Filters and selections are yours and recorded
Follow-up questions take minutes
The same view serves compliance and production
None of this replaces counsel or a formal program. It means arriving at those conversations already knowing what your data says.
Reaching peer parity in those tracts is 144 originations. A compliance officer sees exposure in that number. A production leader sees volume no competitor is contesting. Same arithmetic, same tracts, same afternoon of work. What changes is who is looking, and what they do next.
The tracts with the widest peer gap tell you where an originator, a partnership, or an office would find demand that is already being served by someone else.
Marketing that reaches the whole market is both a fair lending expectation and the cheapest source of applications you are currently not receiving.
Re-run the same filters. A closing gap is evidence of a working program, and it is evidence you produced yourself rather than purchased.
HMDAVision is so much more than a fair lending tool. It will deeply impact your marketing team and its efforts as well. No serious lender should be without it.
Debra Leone, Deputy General Counsel, Hudson Valley Credit Union
Free, no card
Ask in plain English how your lending in majority-minority tracts compares to your market, and get a governed answer from loan-level HMDA. Three a day, fifteen a month.
Start askingIncludes Vision Pro
The full HMDAVision interface to run the six steps yourself: peer sets by volume, tract demographics, mapping, and comparison across every market you lend in.
See ProfessionalIncludes Vision Premium
Deeper history and data export, for teams that need the full time series and want the analysis to live inside their own reporting and board materials.
See ExecutivePolygon Research is a research and analytics firm, not a law firm. Nothing here is legal advice, and a fair lending program should be built with counsel.