Predicting the Sale: Inside the Data That Flags Miami Sellers a Year Early

Most “motivated seller” lists are guesses dressed up as data: a pull of every property carrying a single distress flag, sold to a hundred subscribers at once. The ListoScore is a different instrument. It is a probability — a calibrated, out-of-sample estimate of whether a specific distressed property will record a qualified sale within the next twelve months.

That distinction is the whole moat. A flag tells you something is wrong at a property. A probability tells you what is likely to happen next. Only the second one lets you spend your time and your mailers where they convert.

How the score is built

The model reads the escalation of public-record distress across Miami-Dade — code violations, code liens, finance referrals, stalled permits, and lis pendens filings — together with ownership and equity context. It learns the patterns that historically preceded a real sale, then assigns each property a likelihood band: Watch, Elevated, High, or Very High.

A flag tells you something is wrong. A probability tells you what happens next.

Why “out-of-sample” is the word that matters

Any model can look brilliant on the data it was trained on. The only honest test is out-of-time: train the model strictly on what was knowable before a given date, then grade its predictions against sales that happened afterward — sales it never saw. That is how the ListoScore is measured.

A prediction only counts as correct if the property recorded a qualified, arm’s-length sale within twelve months of being scored. Predictions whose twelve-month window has not yet closed are excluded entirely. There is no credit for the undecided.

The number that justifies the discipline

Graded that way, properties the model flags “Very High” have transacted at several times the rate of a randomly chosen distressed property — a lift the public Track Record page reports live and updates as the scorecard compounds. The baseline it is measured against is not all of Miami-Dade; it is the distressed population the score actually ranks, so the lift reflects skill, not just the obvious fact that distressed homes sell.

A scorecard that grows into a moat

Every prediction is logged the day it is made, with the date its window closes, and is later reconciled against what actually happened. That predicted-versus-actual ledger is the asset a competitor cannot copy by scraping the same records — it only exists because ListoScore has been making and grading dated calls, month after month.

How investors use the band

• Triage first: work “Very High” before anything else — that is where the lift lives
• Stack with equity: a likely seller with real equity can transact clean
• Time the outreach: a fresh “Very High” flag is an early-window signal
• Verify the claim: read the live grading yourself on the Track Record page

The model does not promise that any individual owner will sell — it is a modeled estimate from public records, not a crystal ball. What it promises is concentration: it pushes the rare, genuinely motivated sellers to the top of the list so your finite time lands where the deals are.