Non-QM is one of the hardest mortgage markets to size because it does not show up cleanly in standard public reporting. Lenders may know the market is active, but they often lack a reliable way to see where Non-QM demand is growing, which competitors are gaining share, and which borrower or property segments are driving the opportunity.
Polygon Research solves this by applying a transparent loan-level methodology to public mortgage data, turning raw regulatory records into market intelligence lenders can actually use.
Analyzing the Non-QM market requires more than national statistics. According to our analysis, the Non-QM lending surpassed $239 billion in 2025, representing 10% of total mortgage originations. Unlike FHA or conventional loans, there is no standard flag or universal definition for Non-QM in public data sources.
Different vendors and investors have developed their own definitions based on varying methodologies. Some focus on securitization pathways. Others use product feature proxies. Many rely on incomplete datasets.
We provide a loan-level classification methodology grounded in the complete ATR/QM regulatory framework from 12 CFR §1026.43. The foundation is HMDA loan-level data, enhanced with Polygon Research’s Non-QM classification methodology, geography, lender identity, pricing fields, product features, and borrower/property characteristics.
In this blog, we offer a guide that shows you how to analyze Non-QM using loan-level data to answer critical business questions such as where are Non-QM loans concentrated geographically; which product features trigger volume in specific markets; how does your market share compare to competitors by MSA or county; what are the growth trajectories in your footprint.
Loan-level HMDA data provides answers with census-tract precision.
Non-QM has moved from a specialty lending category into a meaningful part of mainstream mortgage origination. But the market is not evenly distributed, and it is not driven by one product type.
A national Non-QM estimate can tell you the market is large. It cannot tell you whether your opportunity is concentrated in investor properties, self-employed borrowers, jumbo loans, elevated pricing, private securitization pathways, or a specific local geography.
That is the difference between market commentary and market intelligence.
Inside Polygon Vision, Non-QM can be analyzed at the same level where lenders actually make decisions: MSA, county, census tract, lender, borrower profile, loan purpose, occupancy, product feature, pricing outcome, and purchaser type.
This matters because each team asks a different question.
Executive leadership wants to know the size of the opportunity and whether the institution is gaining or losing share.
Product teams want to know which Non-QM features are actually driving volume.
Sales and marketing teams want to know where demand is concentrated.
Compliance teams want to know whether access patterns vary by geography or census tract income classification.
Secondary market teams want to understand pool composition, geographic concentration, and execution pathways.
Those are not questions aggregate statistics can answer. They require loan-level market intelligence.
Non-QM is highly local. A lender cannot understand the market by looking only at national production totals.
Loan-level data shows where Non-QM originations are actually happening by MSA, county, and census tract. For example, the Detroit MSA originated 3,195 Non-QM loans in 2024, representing 43% year-over-year growth. That single market tells a more useful story than a national average because it points to where demand is forming, which lenders are active, and whether the growth is concentrated in specific local corridors.
This is where Polygon Vision becomes especially useful. Users can move from national trends to local market structure and then down to census-tract precision. That makes it possible to distinguish broad market growth from localized opportunity.
For lenders, this changes the conversation. Marketing budgets can follow actual demand patterns. Branch and referral strategies can be evaluated against local volume. Broker recruitment can focus on geographies where Non-QM activity is already proven. Product managers can compare markets rather than assuming that one national Non-QM strategy will fit every footprint.
The key point is simple: Non-QM opportunity is not evenly spread across the country. The markets that matter are the markets where the loans are actually being originated.
Not every Non-QM loan is Non-QM for the same reason. That is why product mix analysis is essential.
In 2025, Polygon Research identified 697,605 Non-QM loans, but the classification drivers were not evenly distributed. The largest category was Other Purchaser + IMB, representing 34.6% of Non-QM loans, followed by Private Securitizer at 28.5%. Another 13.8% of loans were classified as Non-QM for two or more reasons, showing that many loans do not fit neatly into a single product or regulatory trigger.
Product features also played a meaningful role. Balloon payment features accounted for 7.8%, interest-only payment structures accounted for 6.9%, and rate spread accounted for 6.6% of Non-QM classifications. Smaller categories, including extended loan terms and non-amortizing features, represented a much smaller share nationally but may still matter for specific lender, investor, or product strategies.
The national mix provides context, but the strategic value comes from seeing how these drivers change by geography, lender, borrower profile, occupancy, and property type.
Some markets show stronger investor-property and DSCR-style activity. Others lean more heavily toward self-employed borrower demand, bank statement programs, jumbo production, pricing-driven classifications, or specialized product structures. A lender analyzing a Sun Belt investor market may see a very different Non-QM profile than a lender focused on a high-cost coastal market or a Midwestern metro.
This is where product strategy becomes evidence-based. Instead of asking, “Should we offer Non-QM?” the better questions are:
Polygon Vision allows users to answer those questions with loan-level evidence rather than relying on national commentary or anecdotal broker feedback.
Non-QM analysis becomes more powerful when it moves from market size to competitive position.
Loan-level data identifies the top Non-QM lenders in any geography. Users can see who is originating loans, how volume changes over time, which markets they are active in, and what types of Non-QM loans appear to be driving their activity.
That matters because Non-QM is often discussed as a product category, but lenders experience it as a competitive market. A lender may be underperforming in one MSA, gaining share in another, and missing a specific borrower segment entirely. Without local benchmarks, those patterns are difficult to see.
Polygon Vision helps users compare their footprint against the market. You can evaluate market share by geography, borrower profile, occupancy type, loan purpose, product feature, pricing outcome, and lender peer group. That makes it possible to identify whether a gap is strategic, operational, product-related, or geographic.
This also has fair lending value. Because the data can be mapped to census tracts, lenders can examine whether Non-QM activity is reaching different income classifications and communities consistently. That does not replace a full fair lending review, but it gives compliance and strategy teams a shared view of where the market is active and where access may be uneven.
Non-QM is not just a current market opportunity. It is also a market shaped by regulatory change, rate cycles, investor appetite, and product innovation.
HMDA loan-level data allows Polygon Research to analyze Non-QM patterns from 2018 through 2025, including the period around the expiration of the Temporary GSE Patch and the transition to price-based QM standards. That historical view matters because a single year can be misleading. Rate cycles can temporarily suppress or inflate volume. Investor appetite can shift. Product categories can expand or contract. Local markets can mature at different speeds.
By looking across multiple years, users can distinguish temporary movement from structural growth. They can see whether a market is newly emerging or consistently active. They can track whether a lender’s position is stable, improving, or deteriorating. They can also see whether product mix is changing over time.
For Non-QM, this historical context is especially important because the category is not static. The market evolves as credit availability, investor demand, property investment activity, and borrower documentation needs change.
Polygon Research classifies every HMDA record using a transparent methodology based on the ATR/QM regulatory framework, pricing thresholds, product features, purchaser type, and business-purpose indicators.
Each loan is evaluated against year-specific criteria, including:
The methodology is designed to be inspectable. Users can understand not only whether a loan is classified as Non-QM, but why.
Business-purpose and investor-property lending, including DSCR-style activity, is treated separately because these loans may fall outside the consumer-purpose ATR/QM framework while still representing an important part of the broader Non-QM and private-credit market.
This distinction matters. A lender evaluating consumer-purpose Non-QM opportunity may need a different strategy than a lender targeting investor-property demand. Polygon Vision helps separate those signals instead of collapsing them into one vague category.
Polygon Research identified 697,605 Non-QM loans in 2025, representing $239.3 billion in origination volume.
These national statistics mask substantial regional variation. High-cost coastal markets show different loan amounts than Midwest metros. Investor-heavy Sun Belt regions demonstrate distinct income profiles. Markets with high self-employment concentration show different product preferences.
Understanding your specific market requires local loan-level intelligence rather than national averages.
A strong Non-QM strategy does not belong to one department. It connects strategy, product, sales, compliance, and capital markets.
Executive teams use Non-QM data to size the total addressable market, identify where the lender is under- or over-performing, and decide whether Non-QM deserves more investment.
Product teams use the data to understand which features are driving volume, whether local demand supports a specific program, and how product mix differs across markets.
Sales and marketing teams use the data to prioritize geographies, referral channels, and borrower segments based on actual market activity.
Compliance teams use census-tract-level analysis to compare the lender’s activity with market benchmarks and evaluate whether access patterns require further review.
Secondary market and capital markets teams use loan-level intelligence to understand pool composition, geographic concentration, investor demand, and execution pathways.
The common thread is precision. Non-QM strategy becomes stronger when each team works from the same loan-level view of the market.
Loan-level Non-QM analysis turns a hard-to-measure market into a measurable one.
Market share becomes more precise. Instead of relying on industry estimates, users can calculate competitive position by MSA, county, census tract, lender, loan purpose, occupancy, and borrower profile.
Competitive intelligence becomes more actionable. Users can identify which lenders are active in their target markets, how their volume is changing, and what types of loans appear to be driving their growth.
Product strategy becomes more grounded. Users can see whether DSCR-style activity, bank statement programs, jumbo loans, pricing-driven classifications, or other product features are actually present in the markets they serve.
Fair lending analysis becomes more specific. Users can examine market activity across census tract income classifications and compare their footprint with local opportunity.
Secondary market analysis becomes more informed. Users can evaluate geographic concentration, pool characteristics, and execution patterns before making assumptions about investor demand.
The value is not just knowing that Non-QM exists. The value is knowing where it exists, who is winning it, what is driving it, and whether your institution is positioned to compete.
National Non-QM statistics provide context. Local market intelligence drives decisions.
A national estimate can tell you that Non-QM reached 10% of originations in 2025. But it cannot tell you whether your institution is gaining share in Detroit, missing investor-property growth in a Sun Belt market, or under-serving a borrower segment in your own footprint.
Similarly, knowing that purchaser type drove 63.1% of national Non-QM classifications in 2025 is useful context. But the more strategic question is whether that same pattern holds in your markets — and whether competitors are building volume around different product features.
That is the reason Polygon Research built Non-QM market intelligence inside Polygon Vision. Lenders need to move beyond national market commentary and understand the market at the level where strategy is executed.
You do not compete in the national average. You compete in specific markets, with specific lenders, serving specific borrower and property segments.
In our Non-QM Market Intelligence webinar, we show how Polygon Research identifies and classifies Non-QM loans using public mortgage data, and what that reveals about market structure, pricing, competition, borrower profiles, and local opportunity.
The session includes mortgage consultant Dana Georgiou, who advises small and mid-sized lenders on Non-QM strategy and operations. Together, we walk through a practical framework for analyzing Non-QM in a specific market, including loan-level characteristics, pricing patterns, competitive positioning, and the demographic factors that can shape demand.
For lenders, investors, and housing stakeholders trying to understand where Non-QM fits into the current market, the webinar provides a useful starting point — and Polygon Vision provides the data environment to go deeper.
Move from national Non-QM commentary to local market intelligence. Polygon Vision helps you analyze lender share, geography, borrower and property characteristics, product classification triggers, pricing patterns, and historical trends using loan-level mortgage data.
Learn how to analyze the Non-QM market using loan-level HMDA data. Discover geographic trends, competitive positioning, product mix analysis, and growth opportunities across all U.S. markets.
Non-QM is one of the hardest mortgage markets to size because it does not show up cleanly in standard public reporting. Lenders may know the market is active, but they often lack a reliable way to see where Non-QM demand is growing, which competitors are gaining share, and which borrower or property segments are driving the opportunity.
Polygon Research solves this by applying a transparent loan-level methodology to public mortgage data, turning raw regulatory records into market intelligence lenders can actually use.
Analyzing the Non-QM market requires more than national statistics. According to our analysis, the Non-QM lending surpassed $239 billion in 2025, representing 10% of total mortgage originations. Unlike FHA or conventional loans, there is no standard flag or universal definition for Non-QM in public data sources.
Different vendors and investors have developed their own definitions based on varying methodologies. Some focus on securitization pathways. Others use product feature proxies. Many rely on incomplete datasets.
We provide a loan-level classification methodology grounded in the complete ATR/QM regulatory framework from 12 CFR §1026.43. The foundation is HMDA loan-level data, enhanced with Polygon Research’s Non-QM classification methodology, geography, lender identity, pricing fields, product features, and borrower/property characteristics.
In this blog, we offer a guide that shows you how to analyze Non-QM using loan-level data to answer critical business questions such as where are Non-QM loans concentrated geographically; which product features trigger volume in specific markets; how does your market share compare to competitors by MSA or county; what are the growth trajectories in your footprint.
Loan-level HMDA data provides answers with census-tract precision.
Non-QM has moved from a specialty lending category into a meaningful part of mainstream mortgage origination. But the market is not evenly distributed, and it is not driven by one product type.
A national Non-QM estimate can tell you the market is large. It cannot tell you whether your opportunity is concentrated in investor properties, self-employed borrowers, jumbo loans, elevated pricing, private securitization pathways, or a specific local geography.
That is the difference between market commentary and market intelligence.
Inside Polygon Vision, Non-QM can be analyzed at the same level where lenders actually make decisions: MSA, county, census tract, lender, borrower profile, loan purpose, occupancy, product feature, pricing outcome, and purchaser type.
This matters because each team asks a different question.
Executive leadership wants to know the size of the opportunity and whether the institution is gaining or losing share.
Product teams want to know which Non-QM features are actually driving volume.
Sales and marketing teams want to know where demand is concentrated.
Compliance teams want to know whether access patterns vary by geography or census tract income classification.
Secondary market teams want to understand pool composition, geographic concentration, and execution pathways.
Those are not questions aggregate statistics can answer. They require loan-level market intelligence.
Non-QM is highly local. A lender cannot understand the market by looking only at national production totals.
Loan-level data shows where Non-QM originations are actually happening by MSA, county, and census tract. For example, the Detroit MSA originated 3,195 Non-QM loans in 2024, representing 43% year-over-year growth. That single market tells a more useful story than a national average because it points to where demand is forming, which lenders are active, and whether the growth is concentrated in specific local corridors.
This is where Polygon Vision becomes especially useful. Users can move from national trends to local market structure and then down to census-tract precision. That makes it possible to distinguish broad market growth from localized opportunity.
For lenders, this changes the conversation. Marketing budgets can follow actual demand patterns. Branch and referral strategies can be evaluated against local volume. Broker recruitment can focus on geographies where Non-QM activity is already proven. Product managers can compare markets rather than assuming that one national Non-QM strategy will fit every footprint.
The key point is simple: Non-QM opportunity is not evenly spread across the country. The markets that matter are the markets where the loans are actually being originated.
Not every Non-QM loan is Non-QM for the same reason. That is why product mix analysis is essential.
In 2025, Polygon Research identified 697,605 Non-QM loans, but the classification drivers were not evenly distributed. The largest category was Other Purchaser + IMB, representing 34.6% of Non-QM loans, followed by Private Securitizer at 28.5%. Another 13.8% of loans were classified as Non-QM for two or more reasons, showing that many loans do not fit neatly into a single product or regulatory trigger.
Product features also played a meaningful role. Balloon payment features accounted for 7.8%, interest-only payment structures accounted for 6.9%, and rate spread accounted for 6.6% of Non-QM classifications. Smaller categories, including extended loan terms and non-amortizing features, represented a much smaller share nationally but may still matter for specific lender, investor, or product strategies.
The national mix provides context, but the strategic value comes from seeing how these drivers change by geography, lender, borrower profile, occupancy, and property type.
Some markets show stronger investor-property and DSCR-style activity. Others lean more heavily toward self-employed borrower demand, bank statement programs, jumbo production, pricing-driven classifications, or specialized product structures. A lender analyzing a Sun Belt investor market may see a very different Non-QM profile than a lender focused on a high-cost coastal market or a Midwestern metro.
This is where product strategy becomes evidence-based. Instead of asking, “Should we offer Non-QM?” the better questions are:
Polygon Vision allows users to answer those questions with loan-level evidence rather than relying on national commentary or anecdotal broker feedback.
Non-QM analysis becomes more powerful when it moves from market size to competitive position.
Loan-level data identifies the top Non-QM lenders in any geography. Users can see who is originating loans, how volume changes over time, which markets they are active in, and what types of Non-QM loans appear to be driving their activity.
That matters because Non-QM is often discussed as a product category, but lenders experience it as a competitive market. A lender may be underperforming in one MSA, gaining share in another, and missing a specific borrower segment entirely. Without local benchmarks, those patterns are difficult to see.
Polygon Vision helps users compare their footprint against the market. You can evaluate market share by geography, borrower profile, occupancy type, loan purpose, product feature, pricing outcome, and lender peer group. That makes it possible to identify whether a gap is strategic, operational, product-related, or geographic.
This also has fair lending value. Because the data can be mapped to census tracts, lenders can examine whether Non-QM activity is reaching different income classifications and communities consistently. That does not replace a full fair lending review, but it gives compliance and strategy teams a shared view of where the market is active and where access may be uneven.
Non-QM is not just a current market opportunity. It is also a market shaped by regulatory change, rate cycles, investor appetite, and product innovation.
HMDA loan-level data allows Polygon Research to analyze Non-QM patterns from 2018 through 2025, including the period around the expiration of the Temporary GSE Patch and the transition to price-based QM standards. That historical view matters because a single year can be misleading. Rate cycles can temporarily suppress or inflate volume. Investor appetite can shift. Product categories can expand or contract. Local markets can mature at different speeds.
By looking across multiple years, users can distinguish temporary movement from structural growth. They can see whether a market is newly emerging or consistently active. They can track whether a lender’s position is stable, improving, or deteriorating. They can also see whether product mix is changing over time.
For Non-QM, this historical context is especially important because the category is not static. The market evolves as credit availability, investor demand, property investment activity, and borrower documentation needs change.
Polygon Research classifies every HMDA record using a transparent methodology based on the ATR/QM regulatory framework, pricing thresholds, product features, purchaser type, and business-purpose indicators.
Each loan is evaluated against year-specific criteria, including:
The methodology is designed to be inspectable. Users can understand not only whether a loan is classified as Non-QM, but why.
Business-purpose and investor-property lending, including DSCR-style activity, is treated separately because these loans may fall outside the consumer-purpose ATR/QM framework while still representing an important part of the broader Non-QM and private-credit market.
This distinction matters. A lender evaluating consumer-purpose Non-QM opportunity may need a different strategy than a lender targeting investor-property demand. Polygon Vision helps separate those signals instead of collapsing them into one vague category.
Polygon Research identified 697,605 Non-QM loans in 2025, representing $239.3 billion in origination volume.
These national statistics mask substantial regional variation. High-cost coastal markets show different loan amounts than Midwest metros. Investor-heavy Sun Belt regions demonstrate distinct income profiles. Markets with high self-employment concentration show different product preferences.
Understanding your specific market requires local loan-level intelligence rather than national averages.
A strong Non-QM strategy does not belong to one department. It connects strategy, product, sales, compliance, and capital markets.
Executive teams use Non-QM data to size the total addressable market, identify where the lender is under- or over-performing, and decide whether Non-QM deserves more investment.
Product teams use the data to understand which features are driving volume, whether local demand supports a specific program, and how product mix differs across markets.
Sales and marketing teams use the data to prioritize geographies, referral channels, and borrower segments based on actual market activity.
Compliance teams use census-tract-level analysis to compare the lender’s activity with market benchmarks and evaluate whether access patterns require further review.
Secondary market and capital markets teams use loan-level intelligence to understand pool composition, geographic concentration, investor demand, and execution pathways.
The common thread is precision. Non-QM strategy becomes stronger when each team works from the same loan-level view of the market.
Loan-level Non-QM analysis turns a hard-to-measure market into a measurable one.
Market share becomes more precise. Instead of relying on industry estimates, users can calculate competitive position by MSA, county, census tract, lender, loan purpose, occupancy, and borrower profile.
Competitive intelligence becomes more actionable. Users can identify which lenders are active in their target markets, how their volume is changing, and what types of loans appear to be driving their growth.
Product strategy becomes more grounded. Users can see whether DSCR-style activity, bank statement programs, jumbo loans, pricing-driven classifications, or other product features are actually present in the markets they serve.
Fair lending analysis becomes more specific. Users can examine market activity across census tract income classifications and compare their footprint with local opportunity.
Secondary market analysis becomes more informed. Users can evaluate geographic concentration, pool characteristics, and execution patterns before making assumptions about investor demand.
The value is not just knowing that Non-QM exists. The value is knowing where it exists, who is winning it, what is driving it, and whether your institution is positioned to compete.
National Non-QM statistics provide context. Local market intelligence drives decisions.
A national estimate can tell you that Non-QM reached 10% of originations in 2025. But it cannot tell you whether your institution is gaining share in Detroit, missing investor-property growth in a Sun Belt market, or under-serving a borrower segment in your own footprint.
Similarly, knowing that purchaser type drove 63.1% of national Non-QM classifications in 2025 is useful context. But the more strategic question is whether that same pattern holds in your markets — and whether competitors are building volume around different product features.
That is the reason Polygon Research built Non-QM market intelligence inside Polygon Vision. Lenders need to move beyond national market commentary and understand the market at the level where strategy is executed.
You do not compete in the national average. You compete in specific markets, with specific lenders, serving specific borrower and property segments.
In our Non-QM Market Intelligence webinar, we show how Polygon Research identifies and classifies Non-QM loans using public mortgage data, and what that reveals about market structure, pricing, competition, borrower profiles, and local opportunity.
The session includes mortgage consultant Dana Georgiou, who advises small and mid-sized lenders on Non-QM strategy and operations. Together, we walk through a practical framework for analyzing Non-QM in a specific market, including loan-level characteristics, pricing patterns, competitive positioning, and the demographic factors that can shape demand.
For lenders, investors, and housing stakeholders trying to understand where Non-QM fits into the current market, the webinar provides a useful starting point — and Polygon Vision provides the data environment to go deeper.
Move from national Non-QM commentary to local market intelligence. Polygon Vision helps you analyze lender share, geography, borrower and property characteristics, product classification triggers, pricing patterns, and historical trends using loan-level mortgage data.
Because there is no standard flag or universal definition for Non-QM in public data sources, Polygon Research developed a loan-level classification methodology grounded in the complete ATR/QM regulatory framework from 12 CFR §1026.43. Every HMDA record from 2018 through 2024 is evaluated against year-specific criteria including price-based thresholds (APR vs. APOR by loan amount tier), points and fees caps, product feature restrictions (balloon, interest-only, negative amortization), and purchaser type. The classification logic is transparent and reproducible every loan's Non-QM designation can be inspected to understand the specific triggering factors. Business-purpose loans including DSCR products are flagged separately, since they fall outside ATR/QM scope but represent a significant market segment.
In 2025, Polygon Research identified 697,605 Non-QM loans, representing $239.3 billion in origination volume. That equals 10.22% of total mortgage originations by loan count and 9.95% by dollar volume. The average Non-QM loan amount was $343,032, with an average interest rate of 7.463% and average applicant income of $279,417. Growth is being driven by several forces rather than one single product category. Investor-property lending, DSCR-style loans, self-employed borrower demand, bank statement programs, private securitization channels, jumbo lending, and pricing-related QM thresholds can all contribute to Non-QM activity. The mix varies significantly by market. To understand the real opportunity, lenders need loan-level analysis by geography, lender, borrower profile, property type, product feature, and pricing outcome.
National market share figures provide context but cannot drive resource allocation, product strategy, or competitive positioning because lenders compete in specific geographies, not nationally. The Detroit MSA, for example, originated 3,195 Non-QM loans in 2024 representing 43% year-over-year growth a pattern invisible in aggregate data. Loan-level HMDA data maps each origination to a specific census tract, revealing exactly where volume concentrates, which product features drive demand locally, and where competitors are gaining or losing ground. Understanding whether DSCR products, bank statement programs, or pricing-driven classifications dominate your market requires local loan-level intelligence that national averages cannot deliver.
Loan-level Non-QM intelligence supports every function that touches market strategy or compliance. Executive leadership can measure total addressable market and competitive share at the MSA, county, or census tract level. Product development can identify which underwriting features and pricing approaches actually generate volume in specific geographies. Sales and marketing can allocate resources based on demonstrated local demand rather than national trends. Compliance teams can overlay Non-QM originations against census tract income classifications to identify gaps in LMI communities before regulatory review. Secondary market teams can benchmark pool composition and geographic concentration against market standards before securitization. All of these applications require the loan-level granularity that aggregate statistics cannot provide.