Why our numbers differ from portal averages
Updated
INGAR price-per-m² figures are typically 10–30% lower than the averages published by real-estate portals. The difference is not an error — it is methodological. INGAR publishes the median of a standard asset comparable across cities (2-bedroom apartment, 50–80 m²), while portals average their entire inventory, where the premium segment is overrepresented. When average is compared against average on the same basis, INGAR samples converge with external references within ±1–2% (Bogotá vs. Ciencuadras: −1.1%; Madrid vs. Idealista: −1.9%).
Representativeness audit (2026-07-18): own average vs. external reference average
| City | Own avg. | Reference | Gap | Verdict |
|---|---|---|---|---|
| Bogotá | 2,252 | Ciencuadras 2.277 | −1,1% | healthy sample |
| Madrid | 6,744 | Idealista 6.876 | −1,9% | healthy sample |
| Miami | 5,877 | Estadísticas city 5.425 | +8% | healthy (methodological gap: asset definition) |
| Santiago | 2,895 | Portalinmobiliario RM 3.305 | −12% | healthy sample; FX bug fixed (2,785→2,643, changelog on the city page) |
| São Paulo | 2,035 | FipeZAP 2.358 | −13,7% | bias corrected via stratified re-sampling (1,595→1,600, changelog on the city page) |
Units: USD/m². "Own average" = average of all captured listings (no standard-asset filter), comparable with the external reference basis. The remaining 6 hub cities showed no gaps requiring verification.
FAQ
Why use the median of a standard asset instead of the market average?
Because the average of "everything listed" is not comparable across cities: it mixes studios with penthouses and inherits each portal's bias (premium listings are posted and promoted more heavily). The median of the ICHS standard asset (2-bedroom, 50–80 m²) measures the same good in all 11 cities — the only honest way to answer "how many m² does USD 200,000 buy here versus there?".
How do you know the sample is not biased?
It was audited (2026-07-18) with a decisive test: comparing our own average against the reference portal's average for each market, on the same basis. Result: Bogotá −1.1% vs. Ciencuadras, Madrid −1.9% vs. Idealista, Miami +8% vs. city statistics — healthy samples. The audit also detected and corrected two issues visibly: a CLP→USD conversion bug in Santiago (2,785→2,643, −5.1%) and a pagination bias in São Paulo (re-sampled stratified by zone; the number barely moved: 1,595→1,600). Both corrections are recorded in each city page's changelog. Since then, every pipeline run executes a representativeness self-test (drift vs. external reference ±15% + per-neighborhood coverage) and a snapshot that fails it is not published.
Why do your numbers match the portals in some cities?
When a city's inventory is dominated by the standard asset, median and average converge: this happens in Buenos Aires and Lima, where our number lands very close to market references. The gap appears where the listed inventory overrepresents the premium segment (Madrid: 61% of listings sit in 8 center/premium districts) or where the standard asset is the market's compact segment (Miami: the typical 2-bedroom exceeds 80 m²). Each city page declares basis, sample and date so the difference is always traceable.
Sources
- [1] Metodología ICHS (activo estándar SA-2B-50-80) · 2026-07-18 A
- [2] Ciencuadras — precios Bogotá · 2026-07-18 B
- [3] Idealista — precios Madrid · 2026-07-18 B
- [4] Índice FipeZAP — São Paulo · 2026-07-18 A
- [5] Portalinmobiliario — RM Santiago · 2026-07-18 B
Ready to invest in Uruguay?
Book a call with an INGAR advisor and we'll map your exact numbers, no surprises.
Book a call