Boliga Data Scraping

Asking and sold are different sources and stay separate fields

Denmark's transparency makes something possible that most markets do not allow: comparing what was asked with what was paid.

But they remain two sources, and conflating them is the failure to avoid:

• Asking price comes from a listing, published by an agent.
• Transaction data comes from a registry process, on its own cadence.
• They are published at different times, with the transaction lagging the sale.
• Linking them is a match, not a fact — address, timing and attributes have to align.

So asking_price and sold_price are separate fields with separate source and timestamp, and asking_sold_link_confidence travels with any pairing.

asking_to_sold_pct is computed only where the link is high confidence, and is null with a reason otherwise. An unverified pairing would produce the most quotable and least reliable figure in the dataset.

And the lag is a field, not a nuisance

transaction_published_lag_days is recorded. A recent-period sold-price average is systematically incomplete because the most recent transactions have not been published yet — and that incompleteness gets mistaken for a market movement more often than almost anything else in property data.

Danish conventions, and what we do not collect

Property type conventions

Danish property categories — owner-occupied flats, houses, cooperative housing — have distinct legal and pricing characteristics. Cooperative housing in particular does not price like ownership, and pooling them produces an average across different products.

tenure_type is a dimension and we do not pool cooperative with owner-occupied in any default figure.

Area and price per square metre

Area basis recorded with every figure; price per square metre captured where displayed and computed alongside, with disagreement flagged.

Status

Transitions as timestamped events. Here a disappearance can sometimes be resolved against transaction data — but only where the link is high confidence, and otherwise disappearance_reason stays not_determined.

What we do not collect

• Buyer or seller identities, even where a registry process makes some of it available. A transaction is a commercial event; the parties are people.
• An asking-to-sold figure from a low-confidence link.
• A recent-period average presented without its publication lag.
• An address sharpened beyond publication.

The first is the one that matters most. Transparency about prices is not a licence to collect data about the people involved, and we do not.