Methodology

What this data is

Immoweb listings, scraped over time. That means:

The coverage problem

The scraper did not run continuously. Monthly scrape days range from 6 to 31, and there is no data at all for .

No listings were captured at all during that period, so those months cannot be reconstructed by reprocessing.

This shapes what the dashboard reports. Medians, ratios, yields and price indices are robust, because a median is insensitive to how many observations were sampled provided the sample is not systematically skewed. Those carry the dashboard. Raw counts and volumes are a different matter: they largely measure how often the scraper ran, so they are flagged or broken out wherever they appear.

Time series are keyed on publication_creationDate, the listing's own date, never on the scrape date, so the x axis reflects when a property came to market rather than when the scraper ran.

Cleaning rules

Raw values span €1 to €32,000,000 and surfaces from 1 m² to 452,230 m². Applied throughout:

Gross rental yield

12 × median asking rent ÷ median asking price, computed per commune where both sides carry at least listings. of communes qualify.

It is a market level ratio, not an investment return. The rent and sale medians come from different properties, and the figure ignores purchase costs, registration duty, maintenance, vacancy, management and tax. Belgian registration duty alone runs to 12.5% in Wallonia and Brussels.