A literal account of the sources, fields, filters, formulas, samples, dates, and limits behind every published figure.
The active asking-price sample uses public residential advertisements from Fincaraíz and Metrocuadrado. Market data is refreshed approximately weekly. Selected-neighborhood new-build data comes from La Galería Inmobiliaria’s licensed monthly census. Google Trends remains a separate search-interest signal, never buyers or transactions.
When available, we retain source, external identifier, DIVIPOLA municipality, stated neighborhood, operation, property type, COP price, area, bedrooms, new-build status, publication date, observation date, administration fee, and coordinates. We publish anonymous aggregates, not individual third-party listings.
Only rows observed within a 30-day active window enter aggregates. A likely cross-portal duplicate shares municipality, operation, neighborhood, normalized type, exact price, and rounded area. We exclude areas below 18 m² or above 2,000 m²; sale values outside COP 1–45 million/m²; and rents outside COP 8,000–350,000/m² per month.
Neighborhood names are standardized within each municipality; same-named areas in different cities are never merged. A listing without a stated neighborhood remains in the municipality-level group and counts in the city sample, but not as a neighborhood. General aggregates publish from 5 observations; bedroom, new-build, and resale segments from 4. Published rankings require at least 30 sale and 15 rental observations.
We publish medians, not averages. A typical 2-bedroom is the median total asking price among advertisements marked as two bedrooms, rolled up across neighborhoods with sample-size weighting. It is a comparable asking-price indicator, not an appraisal of a specific unit.
Formula: median monthly rent per m² × 12 ÷ median sale asking price per m². It compares a neighborhood or city market and does not necessarily pair the same unit. It excludes administration, property tax, insurance, maintenance, vacancy, management, financing, and tax.
Active-listing age is the median number of days since publication among still-active advertisements with a publication date and requires at least 5 observations. It does not measure time to a completed sale. The trend compares dated cuts of the weighted asking-price-per-m² index and publishes only after 28 days; every result displays its exact period.
Data is stored in COP. USD conversions use the Superintendencia Financiera’s daily TRM when available; other currencies use a timestamped market-rate service with a backup rate identified as stale. Counts are aggregate listing observations, not unique properties.
Each refresh publishes a dated cut. A material correction preserves the affected cut, documents the change, and updates the modification date. Aggregate statistics may be reused with attribution and a link while retaining date, geography, sample, unit, and limitations. Individual advertisements may not be republished or presented as deeds or closing prices.
Available cuts preserve the aggregates published on that date. The series grows with each weekly refresh.
