We test the score on real sales. 77 counties tested so far.
Each year we score every property using only last year's records, then check who actually sold. Here is the result for the 77 counties tested so far, every result in full. A county appears once its sales records are complete enough for a fair test.
Franklin County, OH
Share of properties that sold within 12 months, by score group. Top 10% sold 2.8× as often as the county average (3.8%). Scored with only the records from before the test year.
How the test works
- Turn the clock back a year. Rebuild the county from the records published up to that day.
- Score every property. The model sees only those old records, never what happened next.
- Check real sales. Count who sold in the next 12 months, by score group.
- Publish it. Every county, including the ones where the score is weaker.
Read a row like this. In Franklin County, OH, 3.8% of properties sold in the year. Of the top-scored tenth, 10.6% sold. That is 2.8 times as often. Our weakest tested county is Multnomah County (Portland), OR at 1.3×, and it is on this page too. There, the records we have separate future sellers from the rest less sharply than elsewhere. Open my county →
Proof Explorer: see the houses behind the chart
Pick a county and a score band. We show 100 houses drawn at random from that band in the test year (never hand-picked), and which of them really sold in the next 12 months. The ones that did not sell are shown too.
Score band, highest score on the left. The bars show the share of all houses in each band that sold within 12 months.
Below are 100 random houses from this band: 8 sold and 92 did not. Most of the top band does not sell within a year. The score raises the odds; it does not promise a sale. A hundred houses is a small sample, so its count can differ a little from the share above.
sold in the next 12 months did not sell · tap a dot to see the facts
Street, owner and parcel number are hidden. Real outcomes from a test where the model saw only older records. Try to beat the AI yourself · How the score is built
Counties tested so far
How is the likely-to-sell score tested?
We rebuild every county as it looked a year ago, score every property with only the records available then, and check which properties really sold in the next 12 months. The score never sees the answer.
What does "2.4× the county average" mean?
If 3.6% of all properties in the county sold that year and 8.8% of our top 10% sold, the top 10% sold 2.4 times as often. A random list would show 1×.
Why do results differ by county?
Counties publish different records (some add court, code and probate data), and housing markets move at different speeds. We publish every county so you can judge yours.
How often is this updated?
When we retrain the model (every few months) the backtest is run again for every county and this page updates.
Check the score in your county, free for 7 days.
Motivated-seller lists from official county records, scored and explained, refreshed every Monday.
- 7 days free
- Explore: no card
- Full access: cancel before day 7, pay nothing
- Max 10 investors per county
