A house listing, checked against the city's own records
The ad against the land registry, the price against houses its own size, and a distance test on a nearby settlement.
Built 2026-10-01 · Filmed 2026-10-20
My agent's report on a house, built to show where it could be wrong.
- The ad says393 m²of building. The land registry: one lot, 432 m².
- Price per m²14%under the median of its size. Not an outlier.
- Closest houses22%dearer per m² than the far ones.
Exhibit 1Two of its checks, filmed
Two short clips, each with its own sound. Tap one to play it. Playing one stops the other.
What to notice: the ad's 393 and the registry's 432 sit side by side, and the report tags each claim as measured or secondary.
From the first half of the report film. The report is in Portuguese, so its page is too.
What to notice: the nearest band is lit while the voice says its houses cost the most, then the result box says what that does and does not mean.
From the last section of the report film. The chart below is drawn from the same page's figures.
Exhibit 2The distance test, drawn
The price first: this house is 14 percent under the median of houses its own size nearby. That is cheaper than most, but not an outlier. The report then asked whether the nearby informal settlement explains why it is cheap. It took the houses for sale in the same size range and measured each one's distance to the settlement. The distances run from 127 to 510 metres and then jump straight to 1,301, so there is an empty stretch of 791 metres, and the cut-off sits inside it. Up to 600 metres there are 12 houses. Beyond one kilometre there are 8.
The numbers behind the chart
| Group | Houses | Distance | Median R$/m² |
|---|---|---|---|
| Up to 600 m | 12 | 127 to 510 m | 5,987 |
| Over 1,000 m | 8 | from 1,301 m | 4,908 |
The nearest group is the dearest, 22 percent above the far one. This house sits 21 percent under the median of its own group, so whatever makes it cheap is not something it shares with the houses at the same distance.
Where it could be wrong
The report lists this itself, and so do I.
- The cut-off is not a round number I picked. A first cut at 500 metres wobbled: one ad is geocoded twice, at 488 and 510 metres, and the copy that survived changed the median. Cutting inside the empty stretch makes that stop mattering.
- Moving away from the settlement here also means moving from the middle of the neighbourhood toward a highway, which is cheaper for its own reasons.
- The sample is small (12 and 8 houses) and made of asking prices, not sale prices.
- So the test shows that the settlement does not explain this house. It does not show that settlements never move prices, and the report notes that the academic literature says low-income housing does lower the prices around it.
The film is shot from a copy of the report's page. Everything that names a place, an agency, a listing or a person is masked in the page source before the film is shot. The pictures are blurred and the sections about my own plans are left out. What a film cannot show: the page is interactive, and it carries more than the clips do, such as the neighbourhood safety numbers and the list of other houses for sale.
The long cutThe whole report, with my narration
A 1:58 film: the catch, the ad against the records, the attic and the settlement test. Choose English or Português under the picture.