What every number on this site measures, how it is computed, and where it should not be trusted.
Listing and sale records come from public real-estate listing data, refreshed on a daily cycle. School data comes from the U.S. Department of Education’s public school file; district and neighborhood boundaries from U.S. Census TIGER; flood zones from FEMA’s National Flood Hazard Layer; wildfire hazard from the USFS Wildfire Hazard Potential raster. Automated valuation and long-term rent estimates are licensed from a third-party property-data provider.
Each market page shows the observation date of the most recent sale in our data and the sample the page is built from.
Median sold price. Median closing price of sales that closed in the trailing 30 days. Where a state does not publicly disclose closing prices, the last list price is used and the figure is marked with an asterisk.
Sale-to-list. Median of closing price divided by list price, over sales with both values present.
Sold over asking. Share of those sales that closed above list price.
Days on market. Median days on market as reported on the listing at close.
Price per square foot. Median of closing price divided by living area.
Active listings. Count of listings currently marked for sale, including homes under contract (pending or contingent), which are excluded separately where relevant.
Months of supply. Active listings divided by the trailing 30-day sales pace. Lower means tighter supply.
New listings (30d). Active listings first observed in our data in the last 30 days. A genuine relist restarts this clock.
With price cuts. Share of active listings that have recorded at least one price reduction since we began tracking them.
Absorption rate. Trailing 30-day sales divided by current active listings.
Buyer’s / seller’s gauge. A 0–100 composite of months of supply, sale-to-list, days on market, and price-cut share. It is a directional summary, not a forecast.
A statistic is reported only when its window contains at least 5 qualifying sales; below that the metric is shown as unavailable rather than estimated from a handful of transactions. Market pages with fewer than 5 recent sales and fewer than 25 active listings are published for reference but marked noindex, because the sample is too small to be reliable.
Comps follow standard appraisal practice: same property type, then progressively wider recency tiers (90, 180, 365 days), a living-area bracket of roughly ±20–35% around the subject, and a bedroom filter, relaxing only as far as needed to reach a defensible set. The subject’s own prior sale is excluded. Each comp shows why it qualified — distance, recency, size difference, and bedroom difference.
The headline value estimate is the median price per square foot of the qualified comps applied to the subject’s living area, rounded to the nearest $1,000, shown alongside the third-party automated valuation so a disagreement between the two is visible rather than hidden.
Non-disclosure states. Idaho and several other states do not publicly disclose closing prices. In those markets, sale figures fall back to the last list price and are labeled as such.
Year-over-year comparisons. Our own sale history is roughly a year deep, so year-over-year figures appear only where a full prior-year window exists; otherwise they read as still accumulating.
Public-record error. Listing feeds carry duplicates, delayed status changes, and occasional mis-geocoded records. We de-duplicate, normalize city names, and drop records that fall outside a market’s state, but errors survive.
Not an appraisal. Every estimate here is a starting point for a professional’s judgment, not a valuation product, and should not be relied on for lending, tax, or legal purposes.