U.S. Market Profiles
A market profile describes a county's housing-market character — its long-term trajectory, supply-demand balance, demographics, and economic footprint. Most U.S. counties cluster into one of eight recurring profiles. About a third don't fit any single profile cleanly — those are surfaced separately.
A profile is descriptive context; the BoomTown Score is a 5-year price-growth forecast. They usually agree — when they diverge sharply for a given county, that county is moved to Idiosyncratic Markets rather than carry a profile label that doesn't fit.
[01] The Eight Profiles
[02] Counties That Don't Fit a Profile
[03] How These Were Derived
Profiles are derived empirically from 24 features per county — housing trajectory (Zillow ZHVI 1yr / 3yr / 5yr changes, peak-to-current drawdown, pandemic surge, post-peak decline), rent dynamics (ZORI), supply (months of supply, days on market, building permits), affordability (income-to-home ratio), labor (BLS QCEW employment + wage growth, industry HHI), and demographics (population growth, migration inflow, education, amenity score). BoomTown Score is deliberately excluded from the clustering — profiles describe character, score forecasts future price.
We use hierarchical agglomerative clustering with Ward linkage. The number of profiles (k=8) was chosen via silhouette score and dendrogram-gap analysis. Each county's assignment was validated with 100 bootstrap resamples — counties whose profile assignment was stable in ≥70% of resamples get a "clean" classification; counties stable in 50-69% are "leans toward"; counties below 50% — or whose BoomTown Score deviates more than 30 points from the profile mean — are reported as idiosyncratic.