By subscribing you agree to our Privacy Policy and provide consent to receive updates from our company. You can unsubscribe at any time.
© 2026 Polygon Research. All right reserved.
Fair lending use case · Redlining, marketing, underwriting, pricing, steering
Five scans on loan-level HMDA data, run in one session on your institution, the peers you choose and the market. Each scan ends in a number you can act on.
Free tier is three answers a day, fifteen a month, no card.
This page helps answer
Where does our footprint stop, and who is lending there instead?
Is our application mix keeping pace with the market we serve?
Which lever closes a denial gap: down payment assistance, credit policy or product mix?
Five scans, one loan file
HMDA is public and loan-level. HMDAVision runs all five fair lending scans against it, for your institution, for the peers you pick and for the whole market, with the same filters carried across every view. The scans are the ones examiners run. The difference is who runs them first.
Scan 01
Where your applications and originations land against tract demographics, and who lends in the tracts you do not.
Output · Tract gap by minority share
Scan 02
Your application mix against the population of the market, by race and ethnicity together.
Output · Applications to close the gap
Scan 03
Denial rates by borrower group and loan purpose, with denial reasons and odds of denial beside them.
Output · The Minority Delta
Scan 04
Higher-priced loan incidence for your institution against the market, by loan purpose and tract type.
Output · Odds of HPL
Scan 05
Loan type, purpose, term and channel distribution by borrower group, benchmarked against peers.
Output · Product mix by group
Scan 01 · Redlining
One metro, every census tract shaded by minority share, each circle sized by your applications or your originations. Toggle between the two. Then press Select Excluded Tracts to keep only the tracts where you took nothing, and see who originated there instead, by loan purpose.
Illustrative figures from public HMDA and Census data for one county, 2025. Lender identity removed. Your version runs on your own loan application register with every filter recorded.
What the scan says
Tracts above 50 percent minority are 19 percent of this county and 5 percent of this lender's originations. Sixteen loans closed in one of the excluded tracts that year, most of them purchase. Other lenders made them.
The excluded-tract list is a branch, marketing and referral-partner plan before it is a finding. The lenders succeeding there show which products the demand is in.
Whether lenders of similar size achieved a higher share in the same tracts. Peers set the bar the examiner will use.
Scan 02 · Marketing
Minority Detail crosses race with ethnicity, so Hispanic White and Non-Hispanic Black are separate rows rather than a blended average. The gap between application mix and population mix is tracked across four HMDA years, and the last column converts it into the number of additional applications that would close it.
One metro, all lenders, 2025 applications from HMDA against Census population. Run for a single institution, the same table shows where your own marketing is landing and where it is not.
What the scan says
Non-Hispanic Black residents are 8 percent of this market and 5 percent of its applications. The gap has held between 2.2 and 2.7 points for four years. Closing it takes 598 more applications.
The last column is a marketing target. It already has a geography attached, because the redlining scan shows which tracts those applications would come from.
Whether the gap is yours or the whole market's. A gap every lender shares is a demand question. A gap only you show is a reach question.
Scan 03 · Underwriting
The disparity check puts the Non-Hispanic White denial rate beside the minority denial rate for every loan purpose, action type or any of sixty other dimensions, and reports the difference as the Minority Delta. Three more views sit one click away: denial reasons ranked against last year, odds of denial, and disparity odds against a control group.
Illustrative figures for one lender in one metro, 2025, lender identity removed. Every view runs the same way on any peer or on the market, with the same filters.
What the scan says
Purchase denials are level across groups. The delta lives in refinance and home improvement, and the top two denial reasons are credit history and DTI. That combination points at specific levers.
When DTI and insufficient cash lead the purchase denials for a group that peers are approving.
When credit history drives a delta in a purpose where the market delta is small.
When the gap sits in a purpose you barely offer. Here, refinance and home improvement. The lenders closing those loans nearby show what the product looks like.
Scan 04 · Pricing
Higher-priced loans (HPL) are the fair lending risk that closes rather than denies. HMDAVision reports HPL originations, the HPL rate and the odds of HPL by loan purpose, for your institution and for every lender in the market, then carries the result into the pricing disparity index.
One metro, all lenders, 2025 HMDA. Odds of HPL is the ratio of the HPL rate to the non-HPL rate. Select your institution and the same table becomes your line against this one.
What the scan says
Cash-out refinance carries a higher-priced rate more than five times the purchase rate in this market. A lender whose cash-out book concentrates in majority-minority tracts has a pricing question before anyone asks it.
Compare your HPL rate by purpose and tract type against the market line. Where yours sits above it, the pricing review starts there, with the loan file already filtered.
Rate spread, discount points and fees for the same borrower group across peers, to separate a product-mix effect from a pricing one.
Scan 05 · Steering
Loan type, purpose, term and channel distribution by borrower group, benchmarked against peers in the same market. Where similarly situated borrowers land in different products, the scan shows the split and which lenders do not show it.
A group concentrated in one product or one channel is a training and product-menu question for the sales floor, and the peer figure says whether the pattern is yours or the market's.
Context before conclusions
The geography exploration view puts housing stock age, occupancy and vacancy beside every tract. A thin application count in a tract that is mostly renter-occupied asks a different question than the same count in an owner-occupied tract with new construction.
410,104
Housing units across the metro's 249 tracts
63.5%
Owner-occupied, against 36.5 percent renter-occupied
12.3%
Vacant, rising to 18.6 percent in new-construction tracts
One metro, 2025. Census housing characteristics joined to HMDA originations at the tract level.
The summary
The Fair Lending Summary reduces the five scans to indexed ratios by tract type: majority-minority, majority-Black and majority-Hispanic. Each ratio is shown for your institution and for the peer set you defined, so a number reads against what comparable lenders achieved in the same market rather than against a national assumption.
Share ratios compare the percent of tracts of a type to the percent of applications or originations made there; 1.0 is proportional. Disparity indices are the protected-group rate over the control-group rate. Illustrative, lender identity removed.
What the summary says
Three redlining ratios sit well above peers. The underwriting indices are within range of them. That is the board slide, and it took one filter to produce.
Bring the index and the peer figure into the same room. A ratio above peers is a plan with a geography, a product and a target already attached from the scans above.
From scan to plan
Because all five scans run on the same loan-level record with the same filters, a finding in one carries straight into the next. The redlining gap names the tracts, the marketing gap sizes the applications, the underwriting delta names the lever, and the summary puts the result beside peers.
01
Run all five against the same market, period and loan population, for your institution and for the peers you choose. Every filter is recorded.
02
Segment the finding by purpose, channel, income group and tract type until the driver is specific enough to name.
03
Set the application target, open the credit policy review, add the product, or brief the board with the index beside the peer figure.
Comparable by design
Fair lending claims are sensitive to definitions. Every view in HMDAVision shows the period, geography, peer group and loan filters that produced it, so the work can be followed by counsel, by an examiner or by whoever runs it next year.
Same scope
Your institution, peers and market on one geography, one period and one loan population.
Filters recorded
The selection that produced each figure is written into the footer of the view.
Visible vintage
HMDA year and Census tract vintage stay attached to the result.
Loan-level applications and originations: action taken, purpose, product, channel, rate spread, higher-priced flag, denial reasons, applicant race, ethnicity, sex and age, and census tract.
Tract population by race and ethnicity, minority share, income group, housing units, occupancy, vacancy and housing stock age.
Lender size, type and channel, chosen by you and recorded with the result.
HMDA year or preliminary release, Census tract vintage, and the date the peer group was defined.
Common questions
Is this a substitute for a compliance or legal review?
No. It is the same analysis an examiner runs on the same public record, done first and done by you. Findings still go to compliance and counsel. What changes is that they arrive with the peer figure and the driver already attached.
Can I run the scans on a competitor?
Yes. HMDA is public, so every scan on this page runs on any reporting lender. That is how the peer columns are built, and how you see who is originating in the tracts you are not.
Can public data explain why a gap exists?
It can show where the gap is, how large it is against peers, and which purposes, channels and denial reasons carry it. Internal credit policy, pricing decisions and staffing are not in the public record. The scan tells you where to look inside.
Which institutions should run them?
Any HMDA reporter: banks, credit unions and independent mortgage banks. Brokers and real estate agents carry fair lending obligations too, and can run the market-level scans on the lenders they send business to.
Your next move
Start with one question at mortgagedata.ai, or open HMDAVision on your own institution with a guided seven-day trial.
Free tier is three answers a day, fifteen a month, no card.