See the collateral risk hidden within otherwise similar loans
What risk sits inside the collateral?
Lenders have excellent information about borrower risk, and sound appraisals at the time of origination. But collateral risk changes after closing — some properties carry meaningfully more downside than others, independent of the borrower.
acretag ranks that collateral risk at the parcel level and layers it onto your existing processes, so you can see which properties carry more risk before it shows up in performance — without changing your risk appetite.
What PRS is — and isn't
An appraisal or AVM estimates what a property is worth at a point in time. Property-quality tools evaluate condition or valuation confidence. PRS is designed to estimate the property's relative downside risk and price durability over time. It complements those tools rather than replacing them.
From property list to a validated report
Lender property list
Share a property list from your existing portfolio or pipeline — no loan-level PII required to start.
→Data enrichment & scoring
Each parcel is enriched and run through acretag's collateral-risk model.
→Property risk score
Every property returns a risk score you can rank and segment your portfolio on.
→Portfolio validation report
We benchmark scores against your own outcomes — lift is measured once we have that data and an existing baseline.
Three ways lenders can apply the score
Portfolio and retained-risk management
- Identify collateral concentration and geographic risk exposures
- Enhance CECL and loss-forecasting segmentation
- Improve stress testing and downside visibility
Loan and pool stratification
- Show which loans are safer than they look, loan by loan
- Price a loan as "near-prime with strong collateral profile" instead of generic near-prime
- Give traders and investors a consistent risk cut across pools
Origination overlay
- Add a collateral-durability signal for thin-file and near-prime decisions
- Increase conversion on prospects pre-selected within compliance guardrails
- Keep existing risk thresholds unchanged — we improve risk identification, not relax standards
The evidence behind the score
Built directly from our research and model-validation data — no illustrative or synthetic figures.
Observed foreclosure rates rise across score tiers
4.14X separation (D10 ÷ D1) · 2.37M rows, 20,713 events
- Property-year observations, not unique borrowers or loans.
Built for lender fair-lending and model-risk governance
Rather than asserting blanket compliance, we document the specific controls your model-risk and fair-lending teams will want to see — and back every headline number with the same rigor: target outcome, sample years and model version, the same way any credit-risk metric should be reported.
- Fairness and disparate-impact testing
- Time and population stability monitoring
- Explainability and reason-factor reporting
- Full model versioning, with a revalidation date on every release
- Independent validation support
- Target, sample and model version published for every headline metric
How a run is tracked, end to end
Questions lenders ask us first
What has acretag's Property Risk Score actually demonstrated?
Research across multiple U.S. county markets achieved a 32.5% out-of-time Gini on an assessor-and-AVM accuracy benchmark, with 4.14X separation between the highest- and lowest-risk score deciles in a separate development backtest. See Expanded validation evidence for more.
How does the Property Risk Score fit alongside FICO, LTV and existing credit policy?
It's an overlay, not a replacement. acretag layers a parcel-level collateral-risk signal onto your existing processes — it doesn't change your risk thresholds or existing models.
Does the underlying research data include FICO or LTV?
No. The research and validation data behind these results is property- and valuation-based only — no borrower-level FICO or LTV data was used. This is a collateral-risk signal, not a borrower-risk model.
How is performance lift measured for my portfolio?
Lift isn't promised upfront — it's measured on your own data during a pilot back-test, once you share outcome history and an existing baseline.
How is PRS different from an appraisal, AVM or GSE property-quality score?
An appraisal or AVM estimates what a property is worth at a point in time. Property-quality tools evaluate condition or valuation confidence. PRS is designed to estimate the property's relative downside risk and price durability over time. It complements those tools rather than replacing them.
Run a portfolio back-test
Tell us a little about your portfolio and we'll follow up to scope a pilot — no commitment, no changes to your existing credit policy.
- A back-test run against your own historical outcomes
- A validation readout you can bring to your model-risk team
- No loan-level PII required to get started