Methodology ( 001 )
How the data is built.
blipton is a daily index of the listing status of every one of NSW’s 4,905,892 residential addresses. This page describes exactly how it is assembled, how accuracy is measured, and what the data does not claim to be.
1. Never poll 5 million addresses
Only about 1.1% of NSW addresses are on the market at any moment. So instead of checking millions of addresses, blipton ingests the complete set of actively advertised properties (roughly 50,000) once a day and diffs it against the full address register — about 89× less work, and a complete picture every run rather than a sample.
2. Matching listings to real addresses
Every advertised property is matched back to its principal address in G-NAF using exact, unit-aware rules. Ranged addresses are handled conservatively: a listing for “5/187-189” is unit 5 of the building at 187–189 — it is never allowed to flag the standalone house at 189. Where G-NAF has no matching unit record, the listing stays honestly unmatched rather than being forced onto the wrong address. New-development and house-and-land project advertising is classified separately and excluded from existing-dwelling status.
3. The 3-day delisting window
An address is flipped to not listed only after its listing has been absent for 3 consecutive daily runs. Listings that briefly disappear and return — portal glitches, relists, agent changes — don’t whipsaw the data. Every flip carries its date, and completed on→off episodes are retained permanently, which is what allows days-on-market statistics to be computed from observed campaigns rather than asserted.
4. Accuracy: recall and precision, not a headline number
Because 98.9% of addresses are not listed at any time, a do-nothing system scores 98.9% “accuracy”. blipton instead reports recall (of the properties genuinely on the market, how many we caught) and precision (of the properties we flagged, how many really are listed), measured on the on-market class where the signal lives. Corrections are published, not buried: when a ranged-address rule produced false positives, the rule was fixed, the affected statuses were reset, and the incident is documented.
5. Service record
Freshness is a claim that can be checked, so it is published rather than asserted. The figures below are computed live from blipton’s own run log — every daily run records whether it completed, how long it took, and what share of advertised properties it matched to an address. Publication lag is the time between a property being published and blipton observing it.
6. Sources & licences
What blipton deliberately does not capture — off-market sales, address-withheld listings, new-build projects, brand-new subdivisions — is listed openly on the home page.
Short answers to common questions are on the FAQ. Building on the data? See the API reference.