Weekly Real Estate News
Summary
ATTOM announced November 3 that its enhanced rental AVM covers nearly 100 million homes, about 40% more than previously. The firm reports 5.9% median absolute error for unlisted rentals. These are company-reported figures, not a guarantee for individual properties.
Property-data provider ATTOM said Tuesday that its upgraded rental automated valuation model can estimate market rent for nearly 100 million U.S. homes, expanding the potential reach of automated rental underwriting at a time when investors and lenders increasingly rely on property-level income estimates.
The Irvine, California, company said its November 3 release covers single-family houses, condominiums and townhomes in all 50 states and Washington, D.C. Coverage is approximately 40% greater than its previous model. ATTOM reports a 5.9% median absolute error for properties before they are listed for rent, a company-supplied performance figure rather than an independent audit.
Why the accuracy claim needs context
A median absolute error describes the middle observation in a distribution of errors. It does not mean every rental estimate falls within 5.9% of the eventual rent. The release does not provide a full distribution of errors by geography, property type or rent band, information that would matter to an investor underwriting a thinly margined acquisition.
ATTOM said its machine-learning models combine property characteristics, estimated property values, rental history and recent rental activity, weighting the evidence differently depending on the property. The training and benchmarking process uses observed rental listings after removing duplicates, placeholder prices and implausible values. Asking rents, however, can differ from executed leases, and many single-family rentals never appear in broadly accessible listing feeds.
For lenders evaluating debt-service coverage ratio loans, an estimated market rent may help with preliminary screening or portfolio surveillance. Loan eligibility and income calculations remain subject to applicable underwriting requirements and the reliability of property-specific evidence. The expanded model is not itself a change to mortgage underwriting rules.
Wider reach, unresolved questions
The company is positioning the product for rental operators, mortgage lenders, insurers, real estate portals and software developers. It says the estimates are available through its data products, including tools intended for automated integration.
Expanded coverage is useful where there is little rental history, but those are also the properties for which an automated number can be hardest to validate. The practical question for a lender or investor is whether model error widens in markets with sparse comparable rentals, rapidly changing supply or unusual property characteristics. ATTOM’s announcement does not quantify those conditions.
The release also describes a range around each estimate, with narrower bounds when supporting evidence is stronger. Users should retain the underlying data, document exceptions and compare estimates with current local leasing evidence before treating them as cash flow.
Source: ATTOM’s November 3, 2026 announcement.
