Data Overview

Mortgage data at your fingertips

Beyond Averages: The Power of Loan-Level Data

How Our Microdata Delivers Unprecedented Market Intelligence

Microdata

In housing finance, true insight isn't found in summaries - it's in the details of every single transaction. We meticulously model microdata at the loan-level. Microdata is the DNA and the backbone of our market intelligence.

What does "loan-level microdata" mean for you?

It means we go beyond dashboards to capture the complete story: every borrower, every lender, every property, every financial nuance, directly from the source. From 124+ million records in HMDAVision to millions of FHA endorsements and billions of Agency RMBS transactions, we transform vast datasets into a singular, high-fidelity view of the market. Even with complex sources like Census PUMS data, we leverage microdata for unprecedented speed and depth in demographic insights.

Ultimate advantage

This granular precision is your ultimate advantage. Lists and tables are built with assumptions that may have nothing to do with your market and market position. We model the microdata that illustrates individual behaviors so you can understand the trends. It allows us to build powerful dashboards for instant analysis, but it's the underlying loan-level truth that empowers your AI and strategic decisions with unmatched clarity and foresight.

microdata is the dna of market intelligence

Mortgage Data Sources

Our Key Data Pillars

Mortgage Origination & Market Dynamics

We model loan-level and microdata in the interactive dashboards of Polygon Vision and Polygon Pulse.

HMDA LAR is used for insight into loan-level mortgage application and origination data for fair lending, market share, and product trends.
HUD FHA Endorsement data allows for monthly loan-level FHA origination insights into broker-wholesale lender relationships and market share.
FFIEC Demographics data supplies the neighborhood level demographic, income, and housing context for HMDA analysis.
NMLS Data modeled for insights into licensed activity - Productivity per LO, LO Turnover, and LO Tenure per lender

Community & Financial Context

Discover patterns and trends in your data effortlessly. Visualise historical data to forecast future performance.

CRA geographic data: Lending activity insights for community development and identifying underserved markets.
FDIC branch and deposits data is the source for mapping bank presence and loacal market activity.

Consumer & Economic Insights

Dive deeper into your data with interactive charts. Hover and click for more detailed insights and breakdowns.

American Community Survey (ACS) PUMS via iPUMS is modeled for detailed household and individual characteristics for understanding housing demand, relocaton opportunities, and borrower profiles.
Current Population Survey (CPS) PUMS and Annual Social and Economic Supplement (ASEC) PUMS allow for a broader economic and social context on employment, income, and household financial health.
Federal Reserve Survey of Consumer Finances (SCF) Microdata

Loan Performance, Property Risk & Value

Dive deeper into your data with interactive charts. Hover and click for more detailed insights and breakdowns.

Agency RMBS Loan-Level Disclosure Data modeled for instant and comprehensive analysis of current mortgage originations trends (not only units but also DTI, interest rate, FICO, LTV, Loan Amount, and more).
Agency Loan Performance Data: Robust insights into loan performance (delinquencies, prepayments, loan modifications, foreclosures, and more) across Fannie Mae, Freddie Mac, and Ginnie Mae.
FEMA Hazard Risk Data: Granular geographic hazard risk assessments blended for a complete picture of housing finance risk.

Learn more about our data and methodology

Frequently asked questions

About Polygon Research Data Sources

What data sources power Polygon Research?

We build on authoritative open data across housing finance: HMDA (mortgage applications and originations), agency loan-level disclosures (Fannie Mae, Freddie Mac, and Ginnie Mae), HUD FHA loan-level data, demographics microdata (ACS PUMS, CPS PUMS, ASEC), NMLS mortgage loan officer data, and FEMA National Risk Indicator (NRI) hazard risk data.

How large is your dataset, and why does scale matter?

Scale is what makes conclusions stable. We process complete datasets (not samples), including 124+ million HMDA loan-level records across 7 years, 1.87B agency loan-level transactions over 6 years, 334M+ ACS microdata records, and decades of loan performance histories—including 25 years of Fannie and Freddie loan performance transactions measured in the billions.

How many years of HMDA do you model?

We focus on the modern HMDA era starting in 2018 (when reported fields expanded), and we model multiple years through the most recently released dataset. This gives you consistent, apples-to-apples trend context across market cycles and strategy shifts—at the loan/application level.

What is HMDA LAR, and why do you rely on it?

HMDA LAR (Home Mortgage Disclosure Act Loan Application Register) is the most comprehensive public view of U.S. mortgage activity—covering applications through outcomes. It’s released annually and updated as institutions file corrections. We use it because it provides transparent, auditable market intelligence with clear data lineage.

What’s included in your HMDA modeling?

Our HMDA modeling supports segmentation across borrower, loan, pricing, lender, channel, and geography. It includes key pricing and cost signals (interest rate, rate spread, origination charges, discount points, closing costs, lender credits), plus product and underwriting context (DTI, CLTV, loan purpose/type, occupancy, units) and rich geographic filters (county, MSA, zip, tract-level designations).

What are HMDA’s limitations, and how do you handle them?

Public HMDA does not include application/closing dates or credit scores, and it doesn’t clearly distinguish some new-vs-existing home details. We address key gaps by applying transparent assumptions and ML-based estimation (for seasonality and credit score signals) so users can still analyze markets with practical, decision-ready context.

How current is your agency (Fannie/Freddie/Ginnie) loan-level data?

Agency loan-level disclosures are published monthly, and each release contains the full securitized portfolio through the end of the previous month. We continuously integrate these files to support current market-trend analysis—credit box, product mix, seller/issuer dynamics, and more.

Do you include loan performance data (prepay/delinquency), and how far back does it go?

Yes. We incorporate loan performance histories (including prepayment and delinquency trends) using agency performance datasets that extend back to 2000. This enables deep lifecycle analysis across market cycles—linking origination attributes to real servicing and performance outcomes.

What makes your FHA loan-level data valuable?

We acquire HUD’s Single-Family FHA portfolio monthly. It’s loan-level microdata with geography down to the zip code and strong visibility into broker–wholesale relationships and FHA market structure. This data is a core driver of FHA Pivot within Polygon Pulse.

What is ACS 1-Year PUMS, and how do you balance detail with privacy?

ACS 1-Year PUMS is annual Census microdata that supports deep demographic, social, economic, and housing analysis—modeled in CensusVision. Because microdata privacy matters, neighborhood-level analysis uses PUMA geographies; for tract-level views, we rely on summarized ACS (5-year) data to preserve anonymity while still enabling market insight.

What are CPS and ASEC, and what do they tell me?

CPS is a monthly survey (about 70,000 households) that tracks employment and demographic conditions; ASEC is the annual supplement used for deeper social and economic measures. We activate CPS/ASEC microdata in CPS Pivot to monitor demand-side signals (employment shifts, income context, household changes) with caution advised for small geographies.

How do you ensure data quality, lineage, and trust?

We prioritize data lineage and explainability: every insight is traceable to its source. Our pipeline includes systematic acquisition from official sources, cleansing/standardization (de-duplication, missing value handling, error correction), enrichment (geo, economic, borrower/lender context), analytical modeling, and multi-point QA cross-checks against benchmarks—then continuous refinement as new data arrives.

Are you selling raw data files?

No. We’re not a raw data reseller. Our value is in transforming open microdata into a harmonized, analysis-ready intelligence layer—delivered through SaaS apps designed for fast segmentation, explainable conclusions, and confident decision-making.

Data for Unrivalled Mortgage Market Intelligence