How Propertiz finds, matches and ranks distressed properties

Every lead starts as a public record in one of 17 counties across 4 states (Florida, Indiana, Ohio and Texas). Here is how each record is tied to a house, how owners are ranked, what the numbers on a lead mean, and how often the model is right.

Collect

Read what each county publishes, from foreclosure filings and unpaid taxes to code cases, probate and evictions, plus every parcel from the assessor or appraisal district.

Match

Tie each record to exactly one property by parcel number, address, legal description, case number or owner name. Ambiguous ones wait for review.

Underwrite

Estimate value from comparable sales or the county's appraisal, rent from HUD market rents, equity from recorded mortgages or the last sale, and DSCR at your purchase price.

Rank & explain

Score distress, recency and owner situation, filter by your buy box, and show exactly why each lead ranks where it does.

Match

Every record tied to one house

Most court filings don't print a parcel number. Each record is matched by parcel number, address, legal description, case number or a unique owner name. A probabilistic matcher accepts a match on its own only at 97% confidence with a clear lead over the next candidate. Anything less waits for a person, because mailing the wrong owner is worse than missing one.

82% of 346,056 public records in the app are tied to exactly one property today.

Score

Points that fade with time

Foreclosure filing: 30 pts, halves every 120 daysProbate: 20 pts, halves every 365 daysCode violation: 10 pts, halves every 180 days

Each signal adds published points that fade by half-life. Later stages replace earlier ones, a sale date that has passed drops out, and owners with signals in three or more categories get a stacking bonus. Every lead lists its points, reason by reason.

Sell Probability

Checked against what happened next

A logistic-regression model trained on past foreclosure filings and what happened next: did the owner sell within six months? It uses the foreclosure stage, how long the owner has held the property, what they paid against today's value, the price band, the age of the building, whether it's multifamily and how distressed the zip code is. It's retrained every week.

Top 10% ranked

15%

All owners

On the 1,645 past filings it learned from, owners in the model's top 10% sold 3.3× more often within six months. On a random fifth of those filings, set aside before training (329), pick one owner who sold and one who didn't, and the model ranks the seller higher 67% of the time (a coin flip would be 50%).

These are checks on past filings, not yet a forward test that trains on older filings and scores newer ones. Measured on Florida foreclosure filings so far: Texas doesn't disclose sale prices, and the other states are still building their history. Shown on a lead only when it's at least 1.3× the average.

Value, rent & equity

Conservative, and labeled

Value: in some counties, values come from our valuation model: up to 20 nearby comparable sales, each adjusted to the home and to today's prices, blended with the county's appraised value scaled to recent sales. Each value comes with a range and a confidence grade, and a low-confidence home shows only the range. Only some counties, because the model needs a county that publishes which sales were arm's-length, and a county switches to it only after it beats our older method on hundreds of past sales, each valued before it sold. We re-test it every week, and each county's page says how accurate it measured. Values assume typical condition: distressed homes often sell well below them. Elsewhere, the median price per square foot of at least three recent comparable sales in the same zip, times the home's size. With fewer sales, or in Texas, where sale prices aren't public, the county's appraised value is used, and the lead says which.
Rent: HUD Small Area Fair Market Rent for the zip and bedroom count, a conservative 40th-percentile figure.
Equity: recorded mortgages and their satisfactions, paid down at each year's average rate, in Jacksonville, Lakeland and Palm Bay. Elsewhere it's an estimate from the last sale, or "unknown" rather than a guess.
DSCR: rent divided by principal, interest, taxes on your price, insurance and HOA.

Estimates are for ranking, not underwriting: verify value, liens and title before an offer. Full methodology, with live match rates →

Every point, published

A lead is a property with at least one active distress signal that also fits your buy box. Each signal adds the points below, fading as it ages. Later stages replace earlier ones: a judgment replaces the filing it came from, and a sale date that has passed drops out. Leads are then split into tiers within the area and buy box you look through: A is the top 20%, B the next 40%, C the rest.

Signal	Points	How it fades
Foreclosure filed (lis pendens)	30	halves every 120 days · ignored after 540 days
Foreclosure judgment	20	full points until the sale date · ignored after 365 days
Foreclosure sale scheduled	20	full points until the sale date
Notice of default (trustee-sale states)	30	halves every 120 days · ignored after 365 days
Notice of trustee's sale	20	full points until the sale date · ignored after 120 days
Tax-deed application	30	full points until the sale date
Tax-deed sale scheduled	30	full points until the sale date
Unpaid property taxes, 2+ years	20	no decay
Unpaid property taxes, 1 year	8	no decay
Open code violation	10	halves every 180 days · max 25 points
Code-enforcement lien	15	no decay
Eviction filed by the landlord	8	halves every 180 days · ignored after 365 days · max 16 points
Probate (owner passed away)	20	halves every 365 days
Absentee owner (tax bill mailed elsewhere)	+10	owner factor: adds points, never makes a lead alone
Owner lives out of state	+8	owner factor: adds points, never makes a lead alone
Not the owner's primary home	+5	owner factor: adds points, never makes a lead alone
Long ownership	+8	owner factor: adds points, never makes a lead alone
High estimated equity	+10	owner factor: adds points, never makes a lead alone
Signals from 3+ categories	+15	stacking bonus

Value accuracy, measured

Lorain County, OH: We tested our values on 1,742 arm's-length home sales in Lorain County from April to October 2026, each estimated before the sale. Half were within 8.9% of the sale price; 55% within 10%. Our 80% range contained the sale price 81% of the time.
Stark County, OH: We tested our values on 1,988 arm's-length home sales in Stark County from April to October 2026, each estimated before the sale. Half were within 10.6% of the sale price; 47% within 10%. Our value range contained the sale price 77% of the time.

What the numbers on a lead mean

Value: in some counties, values come from our valuation model: up to 20 nearby comparable sales, each adjusted to the home and to today's prices, blended with the county's appraised value scaled to recent sales. Each value comes with a range and a confidence grade, and a low-confidence home shows only the range. Only some counties, because the model needs a county that publishes which sales were arm's-length, and a county switches to it only after it beats our older method on hundreds of past sales, each valued before it sold. We re-test it every week, and each county's page says how accurate it measured. Values assume typical condition: distressed homes often sell well below them. Elsewhere, the median price per square foot of at least three recent comparable sales in the same zip, times the home's size. With fewer sales, or in Texas, where sale prices aren't public, the county's appraised value is used, and the lead says which.
Rent: HUD Small Area Fair Market Rent for the zip and bedroom count.
Equity: value minus the estimated loan balance, from recorded mortgages where they're loaded (Jacksonville, Lakeland and Palm Bay today; free-and-clear status in Tampa), else from the last sale. When the records can't support it, equity shows as unknown.
DSCR: rent divided by principal and interest, taxes on your purchase price, insurance and HOA, with each state's typical tax and insurance rates unless you set your own. Try the DSCR calculator.

Limits

Coverage differs by county, and each county page lists what isn't included yet.
Estimates are for ranking, not underwriting. Verify value, rent, liens, the loan balance and title before an offer.
Some records, such as unpaid taxes, code violations and probate, depend on what each county publishes or answers to a records request.

See the live numbers for your county. The app's methodology page shows each source's match rate and the latest Sell Probability check.

Open the methodology →

See the sale before it's a lead

Propertiz reads county records every night in 17 counties across 4 states (Florida, Indiana, Ohio and Texas) and shows you the owners with a reason to sell. Start with 14 days of Closer, no card.

Start 14 days free →