The constraint in non-qualified mortgage credit modeling has always been the data gap. Agency credit frameworks were calibrated on conforming, full-documentation borrowers; NonQM lending draws from a population whose income verification, asset profiles, and documentation types sit entirely outside that training set. RiskSpan, an Arlington, Va.-based provider of data, modeling and analytics solutions for loan and structured finance investors, announced general availability of Credit Model 7.1 on July 17, 2026, a model built specifically for the NonQM segment inside its platform.
What the model addresses
The mechanism behind the problem is straightforward. A conforming mortgage borrower submits W-2s, pay stubs, and tax returns that map cleanly to agency risk parameters. A NonQM borrower might rely on bank statements, asset depletion calculations, or debt-service coverage ratios on investment properties. Standard models, trained on agency data, produce risk estimates that are structurally misaligned with that population.
Purpose-built means the model was designed from the ground up for those alternative documentation types, rather than adapted from a conforming framework. That distinction matters to loan and structured finance investors pricing NonQM pools. Mispriced credit assumptions flow directly into security valuation and risk-adjusted return calculations. Getting the underlying model right is not a preference; it is the precondition for accurate analysis.
Where this sits in the RiskSpan stack
Credit Model 7.1 is delivered inside the RiskSpan Platform, which the company positions as a data, modeling and analytics solution for its loan and structured finance investor base. The integration matters because it places the NonQM credit model in the same environment where investors manage related analytics work. Pulling outputs from a separate system and reconciling them with platform data introduces friction; keeping the model native to the platform eliminates that step.
The 7.1 designation signals a versioned product lineage. This release targets the NonQM segment specifically, which the company frames as purpose-built for that asset class rather than a general-purpose update.
RiskSpan did not disclose pricing, client counts, or performance benchmarks for the model in the announcement.