Every figure on this site comes from three public sources joined together and re-read every morning, and every one of them can be traced back to a row. This page says exactly how: where the data comes from, how a comparable home is chosen, how the seven signals in the deal score are weighted and capped, and the floors below which a number is withheld rather than printed. Nothing here is a valuation, and nothing here is about people.
The three sources
The listing feed, daily
Every active listing in the Las Vegas valley, pulled ZIP by ZIP so no pull hits a result cap, and snapshotted: price, status, description, the day it was first seen. A relist does not reset what was seen.
The county assessor
Each listing joined to its parcel: land use, year built, living area, lot, pool, garage, taxable and assessed value, and the recorded sale history. The square footage and the year built the score uses are the county's, not the feed's.
The county recorder
Swept daily for notices of default, notices of trustee sale, lis pendens and trustee deeds. Counted by month, place and ZIP on the public pages; matched to a parcel only for signed-in users.
What counts as a comparable home
The largest signal in the score is how a home's asking price per square foot compares to similar homes for sale nearby. "Similar" and "nearby" are rules, not judgment, and the same rules pick the comps a signed-in user sees under a home.
| Rule | Value | Why |
|---|---|---|
| Same kind of home | House, townhome or condo, by the county's land-use code | The feed's own type mislabels whole streets; the assessor's code does not. |
| Nearby | Same ZIP, plus everything within 2 miles | Neighbourhood prices do not stop at a ZIP line. |
| Similar size | Within 25% of the home's square footage | Price per foot falls as size rises; a 1,200 sq ft home is not priced by a 3,000 sq ft one. |
| Similar age | Built within 20 years | A 1965 home and a 2005 home are different products at the same address. |
| Recent | Seen for sale in the last 90 days | The market moves; older asks are stale evidence. |
| Enough of them | 8 for a full read; 3 to 7 gives a damped, partial read; under 3, no discount at all | A median of two homes is an anecdote wearing a statistic's clothes. |
| Not self-posted | Owner-posted listings are scored but never used as comps | A price the seller typed is not market evidence, and letting it in would let anyone move a ZIP's number. |
Before the comparison: the adjusted basis
Two homes at the same price are not the same price if one carries $400 a month in dues or a pool. Each comp and the home itself are put on one basis first.
| Step | How |
|---|---|
| HOA dues folded in | Monthly dues times twelve, divided by a 6% cap rate, added to the price as an equivalent lump sum, capped at 10% of the price so a resort tower's fee does not swallow the comparison. |
| Pool backed out | 3% of the asking price (up to $150,000) removed before comparing, because a pool home priced level with poolless comps is underpriced, not fairly priced. Garages measured as worth nothing extra in this market and are left alone. |
| Quality corrected | Where the county's assessed value per square foot is available, the basis is corrected for it, so a finished home is not read as overpriced against dated neighbours. |
| Floored | No adjustment may take a home below half its raw price. A $60,000 condo does not become free because it has a pool. |
The HOA cap exists because, measured over 8,779 listings, the uncapped fee term correlated at -0.71 with the reported discount: the score was mostly reporting dues differences. Capped, the correlation is -0.14 and the signal stays where it is informative.
The seven signals and their weights
The deal score is a weighted sum of seven components, each scaled to the same range and capped so no single row can run away with it. The weights sum to one.
| Signal | Weight | What it reads, and its cap |
|---|---|---|
| Discount against comps | 34% | Asking price per square foot against the comparable pool's median, on an adjusted basis. Capped at 25%; anything past 50% is treated as a data error, not a deal. Reading as overpriced counts half as much against a home as being equally underpriced counts for it. |
| Listing language | 15% | Phrases in the agent's own description that say the home needs work or the seller is motivated: as-is, TLC, handyman, needs work, estate sale, probate, priced to sell, must sell and the rest. Words, not photos. |
| Price cuts | 14% | Cuts observed in Kouzr's own daily snapshots, up to three and up to 10% in total. A relist within 60 days does not reset the count. |
| Owner motivation | 13% | From the county record: how long the owner has held the home (up to 15 years counts) and roughly how much equity the recorded sale implies (up to 50%). Long tenure and deep equity mean room to move. |
| Distress on record | 11% | A notice of default or trustee sale recorded against the parcel, at full weight for the first 90 days and fading to nothing by a year. Counts only, never shown beside an address on a public page. |
| Days on market | 8% | Against the ZIP's median, capped at twice it, with relists chained into one campaign when the gap is under 60 days. |
| Amenities | 5% | A pool or garage the county record confirms, at the same asking price as homes without. A pool is also backed out of the price before the comps comparison, at 3% of price, so it is not read as overpricing. |
The verdict on a listing page is only the first signal, said in words: "12% below the area" means the asking price per square foot is 12% under the comparable median. Inside 3% either way it says "about right". The score is the whole sum; the verdict is one term of it.
What is withheld, and where
A number too thin to be a statistic is not printed. Each floor below gates the page, the sitemap and the API together, so a figure cannot exist on one surface and be missing from another.
| Figure | Floor | Below it |
|---|---|---|
| A market report, comparison or sold page for a place | 5 active homes (5 closings for sold) | Places under it have listings pages but no report |
| A ZIP or neighborhood row in a report | 5 active homes | Thin rows are left out, not dashed |
| A segment page (fixer-uppers, price cuts, under $300K and the rest) | 3 matching homes | The page 404s and leaves the sitemap at the same line |
| A foreclosure count, by month, kind or ZIP | 5 documents | Printed as a dash |
| A closing ratio (sold over final ask) | 3 priced closings | Absent until then |
| A comps-based discount on a home | 3 comparables (damped under 8) | No discount, no verdict |
| How many people saved a home | 3 people | Not shown under it |
| The reviews page | 5 ratings | The page does not exist until then |
Cadence
- Listings pulled and every home re-scored once a day, in the morning.
- Public pages refresh within an hour of the pipeline.
- Market reports snapshot the open month daily and freeze it when the month closes; the dated pages never change after that.
- The county recorder is swept daily; the assessor's recorded sales post some weeks after closing, so the newest month on a sold page is always partial.
- Rent comparables come from a separate licensed pool, and are never mixed into a sale home's comps.
What it is not
Not an appraisal, not an automated valuation, not investment advice, and not a recommendation of any place. Kouzr Inc. holds no real estate or appraisal licence and is not a member of any MLS. Every score is an estimate to be checked against your own inspection and your own agent, and every geographic page describes homes, prices, boundaries and public records, never who lives there. The terms and privacy policy say this at greater length. The vocabulary is in the glossary; the arithmetic a buyer runs themselves is in the calculators.
Questions people ask
Is the Kouzr deal score an appraisal or a home value estimate?
No. It is arithmetic on public data: how a home's asking price per square foot compares to similar homes for sale nearby, plus six other signals about the listing and the owner. It says whether a home is priced under or over its neighbours today, not what it is worth. Kouzr publishes no automated valuation and holds no appraisal licence.
Where does Kouzr's listing data come from?
From public listing feeds for the Las Vegas valley, pulled every day ZIP by ZIP so that no single pull hits a result cap. Every listing is joined to the Clark County Assessor's record for its parcel, and the Clark County Recorder is swept daily for foreclosure filings. Kouzr does not license MLS data and is not a member of any multiple listing service.
Why does Kouzr withhold some numbers?
Because a median over two homes, or a count of one foreclosure in a ZIP, is not a statistic and can identify a household. Every figure has a floor below which it is not printed; the floors are on this page and the same floor gates the page, the sitemap and the API, so a number can never appear on one surface and be absent from another.
How often is everything updated?
Listings are pulled and every home re-scored daily. Public pages refresh within an hour. Market reports take one snapshot a month that freezes when the month closes, so a dated page can be cited. The county's recorded sales post some weeks after closing, so the newest month on any sold page is always partial.
Does the score use anything about the people who live in a neighborhood?
No. Every score, filter and page describes homes, prices, boundaries and public records about the property. Protected characteristics, and proxies for them, are kept out of the model and out of every filter under a fair-housing policy.
Can I check a number myself?
Yes, and the pages are built so you can. The county record on each listing page links to the assessor's own figures; the comps behind a discount are listed in the app; the sold pages show each closing against its last ask; foreclosure counts are of documents at the county recorder. Nothing is modelled that cannot be traced to a public row.