Know What's Next
Polygon Risk models and predicts prepayment (CPR) and default (CDR) at cohort, seller, and portfolio levels — giving you a real-time edge in mortgage performance forecasting.
Mortgage risk models are broken.
Legacy systems rely on lagging data, black-box logic, and blended averages that hide what’s really happening in your portfolio.
Fragmented Risk Signals
Performance, origination, and macro indicators live in separate systems and refresh on different schedules. By the time you see the risk, the market has already moved.
Polygon Risk unifies loan-level, HMDA, MBS, and macro data in one continuously updated modeling layer.
Black-Box Models with No Explainability
Most vendors hand you a number, not a narrative.CPR and CDR outputs are opaque — impossible to audit, justify to ALCO, or defend to regulators.
Polygon Risk provides transparent, explainable ML models built from observable loan-level features, so you always know what’s driving risk.
Portfolio Averages Hide Real Exposure
Blended portfolio metrics flatten key differences by seller, channel, product, FICO, LTV, and geography. You end up managing to an average instead of managing true concentrations.
Polygon Risk models at the loan and cohort level to surface hidden pockets of prepayment and credit risk.
Predictive Without Descriptive = Blind Bets
Forecasts without a solid descriptive base mislead decisions. Descriptive dashboards without forward views keep teams stuck in hindsight.
Polygon Risk merges descriptive analytics and predictive modeling (CPR + CDR) so you see what happened, why it happened, and what’s next.
Together, these four problems define the gap between legacy risk systems and modern mortgage intelligence.
Polygon Risk closes that gap.
Polygon Risk closes that gap.
Predictive intelligence for the open-data era of mortgage risk.
Polygon Risk brings together descriptive analytics and predictive modeling in one transparent platform—so you can understand today’s exposures and anticipate tomorrow’s outcomes.
Unified Data Foundation
We integrate agency MBS loan-level data, HMDA origination records, FEMA and Census datasets, and macro-economic indicators into one consistent, normalized data layer.No silos. No vendor lock-in. Every metric you see ties back to verifiable open data.
Dual Model Engine — CPR & CDR
Polygon Risk models both voluntary and involuntary prepayment risk.
CPR Model predicts borrower behavior - refinance, turnover, seasoning effects - under changing rate and credit conditions.
CDR Model captures credit deterioration and involuntary prepayment (default, foreclosure) using loan-level performance and macro signals.
Both run on ensemble ML frameworks (CatBoost + LightGBM) optimized for precision and explainability.
CPR Model predicts borrower behavior - refinance, turnover, seasoning effects - under changing rate and credit conditions.
CDR Model captures credit deterioration and involuntary prepayment (default, foreclosure) using loan-level performance and macro signals.
Both run on ensemble ML frameworks (CatBoost + LightGBM) optimized for precision and explainability.
Cohort and Seller-Level Forecasts
Unlike one-size-fits-all models, Polygon Risk outputs predictions at the cohort level—by seller, channel, product, FICO band, LTV, occupancy, and region.You can roll up to portfolio or investor view in seconds.
Feature title
Nam pretium turpis et arcu. Vivamus elementum semper nisi. Donec mi odio, faucibus at, scelerisque quis, convallis in, nisi.
Feature title
Nam pretium turpis et arcu. Vivamus elementum semper nisi. Donec mi odio, faucibus at, scelerisque quis, convallis in, nisi.
Feature title
Nam pretium turpis et arcu. Vivamus elementum semper nisi. Donec mi odio, faucibus at, scelerisque quis, convallis in, nisi.
Contact us to learn more
sales@polygonresearch.com