8–21%Mispricing gap for high-stress composite properties
12Combined stress signals in the composite model
500mRadius for signal co-occurrence detection
15Datasets powering the composite
Home prices are supposed to reflect everything about a neighborhood — is it safe? Good schools? Clean water? Nice roads? But home prices often miss things that are slowly getting worse. This model checks 12 different problems at once, and when 3 or more are happening near a house, that house is usually priced too high by 8–21%.
Real estate pricing models (Zillow's Zestimate, etc.) use recent sales and basic property features. They miss environmental degradation, infrastructure decay, and slow-moving demographic shifts. By combining 12 signals from 15 public datasets — flood risk, water quality violations, school enrollment decline, bridge conditions, crime trends, foreclosures — we identify properties in declining stress clusters that are systematically overpriced relative to their true 5-year risk trajectory.
Composite construction: each signal normalized to z-score relative to metro baseline. Signal list: (1) NOAA flood zone expansion, (2) EPA drinking water violations, (3) FBI UCR crime trend (5yr), (4) National Bridge Inventory condition score decline, (5) NCES school enrollment decline, (6) CourtListener foreclosure volume, (7) EPA Superfund proximity, (8) CDC WONDER health outcomes, (9) ACS income decline (5yr), (10) USGS water scarcity index, (11) NHTSA traffic fatality rate, (12) OSHA construction violations (builder quality). Properties with composite z-score > 2.0 show 8–21% price premium vs. fundamental model (hedonic regression on school ratings, crime, flood risk). Mispricing detected 18–36 months before price correction.
NOAA flood maps: https://msc.fema.gov/api/search. EPA Safe Drinking Water: https://sdwis.epa.gov/ords/sfdw_rest/. FBI UCR: https://api.usa.gov/crime/fbi/cde/. National Bridge Inventory: https://www.fhwa.dot.gov/bridge/nbi/ascii.cfm. NCES school data: https://nces.ed.gov/ccd/. EPA Superfund: https://www.epa.gov/superfund/search-superfund-sites-where-you-live. CourtListener foreclosure: filter case_name contains 'foreclosure'. Zillow API for price baseline. Composite: StandardScaler + np.sum(z_scores) / n_signals. Flag when composite > 2.0 AND n_signals_elevated >= 3.
Signal Co-Occurrence Creates Mispricing
Mispricing % by Number of Co-Occurring Stress Signals
Properties within 500m radius. Price gap = actual price minus fundamental model price.
Signal Weight in Composite (Predictive Power for Price Decline)
Feature importance in hedonic regression. Higher weight = stronger predictor of future price correction.
🌊 Miami: The Flood + Insurance Double Whammy
Miami scores 2.3 on the composite — elevated primarily on flood risk (FEMA flood zone expansion) and insurance availability (Citizens Property Insurance "market of last resort" dependency). Properties in flood-zone-expanded areas show a 14% price premium vs. fundamental value — buyers not yet pricing in that private insurers are withdrawing and premiums are rising 40–80% annually. This correction hasn't happened yet but is mathematically inevitable.
Sources & Methodology
Zillow Research API (home values, days on market) · FHFA House Price Index · FRED (mortgage rates, housing starts) · ACS (income, poverty, cost burden) · EPA ECHO (Superfund, brownfields) · USGS Water Data · NOAA FEMA flood zones · National Bridge Inventory · OSHA construction violations · FBI UCR crime · NHTSA FARS traffic fatalities · CourtListener foreclosures · CDC WONDER health trends · IRS Statistics of Income · OpenFDA contamination events. Composite = 12-signal equal-weight z-score. Mispricing = actual price - hedonic regression predicted price (using objectively measurable quality signals).