A gradient boosting model trained on Zillow Research and FRED historical data, not GovLandScout's own listings, predicting how each Texas county's share of price cut listings is likely to move over the next 1, 3, or 6 months. Darker red means the model expects more price cuts (rising distress); darker green means fewer (easing). Counties with no shading don't have enough Zillow history to include yet.
Click a county on the map to see its price cut history and the model's projection.
This map has nothing to do with the property listings elsewhere on this site; it's a separate research project applying a machine learning model to county level housing market data, built to explore whether historical patterns in price cuts, home values, inventory, and unemployment can predict where distress is headed next. It is not derived from GovLandScout's own scraped listings (that history is only a few weeks old, nowhere near enough to train on), it is not validated for real investment, policy, or funding decisions, and a county shaded red is not a claim that specific properties there are about to become distressed; it's a county level statistical tendency, evaluated on historical data the model wasn't trained on but still just a model, with real, measured error (see the writeup below).
Full methodology, data sources, and honest results, including where this model does and doesn't work well, are written up in model/README.md in the project's repository.