How we compile the numbers
Propstock is a research publication, not a data vendor. Every figure on this site is compiled from published sources, and every figure can tell you which ones. This page explains the process, including the parts of it that are automated.
What the Prime Index is
The Prime Index tracks prime residential values across these cities: Dubai, Hong Kong, Johannesburg, Lagos, London, Nairobi, New York, Paris, Singapore, Toronto. It is rebased to 100 at Q1 2019. The global composite is a weighted mean of member-city levels.
The composite is only published when at least half the member cities have a checked reading for the quarter, and the page always states how many are reporting. A composite built on a handful of cities is not a global figure, and we would rather show the gap.
Where the figures come from
Readings are compiled from published sources in this order of preference: primary sources first (land registries, central banks, national statistics offices, regulators), then established consultancies and brokerages, then reputable financial press reporting a named primary source.
We do not use listing-site averages, developer press releases, or portal marketing research. Asking prices are not transaction prices.
Where automation is used
The first pass of compilation is automated. A language model with web search retrieves candidate figures for each city and quarter and returns them with the sources it used. This is a research assistant, not an oracle, and it is treated as one.
Three rules constrain it:
- A figure that arrives without a citation is discarded, not published with a caveat. The pipeline drops it before it reaches the database.
- A figure outside a plausible range for its type is discarded and logged, on the basis that a number we cannot interpret is not a number.
- Nothing the automated pass produces is publicly visible. Every reading is written as a draft and requires a person to publish it.
The same applies to written coverage. Articles drafted by the research engine carry a disclosure at the top of the piece, are bylined to a desk rather than a person, and cannot be given a named human byline until a human has reviewed them. That constraint is enforced in the software, not by convention.
What our confidence grades mean
Vintage, not publication date
Every figure carries the period it describes, not the date we collected it. A Q3 yield found in November is a Q3 figure. Readings older than roughly two quarters are marked as historic on the page rather than presented as current.
Where a table cell shows a dash, we have no citable figure for it. It does not mean zero, and we do not fill it with an estimate.
What our bureaus are
Propstock operates 30 city desks. All 30 are currently research desks: they compile from published sources and do not conduct in-market reporting. We describe them as bureaus because they own coverage of a market, not because they imply a newsroom in that city.
Each city page states which kind of desk covers it. We will update this page as that changes.
Developer ratings
Where we grade a developer on delivery, escrow discipline or build quality, each grade must record the evidence it rests on. Grades without a stated basis are not rendered. Verification status is set by an editor against filed evidence, never automatically.
Commercial disclosure
Developers can pay for placement. Paid entries are labelled wherever they appear alongside editorial data, including inside data tables. Yields supplied by a developer are marked as targets until one of our desks has checked them against comparable lettings.
Commercial relationships do not affect index readings, ratings or coverage decisions.
Known limitations
We would rather state these than have a reader discover them:
- Coverage depth is uneven. Registry data in London, Dubai and Singapore is far richer than in Lagos, Nairobi or Johannesburg, so confidence grades are systematically lower in the African markets and gaps are more common.
- Prime residential definitions differ between sources. Where we combine them, the method field on the reading says so.
- Currency conversion introduces error in markets with volatile exchange rates. USD figures for Nigeria in particular should be read with that in mind.
- The index is a research product built from public data. It is not a transaction-level index and should not be treated as one.
Corrections
If a figure on this site is wrong, we want to know and we will correct it with a note on the page. Contact the relevant desk.
Nothing on Propstock is investment advice. Market data is research, published for information.