Methodology — how every figure on this site is computed

01 Principles

We compute from registered transactions and registered tenancy contracts, never from asking prices. A listing shows what a seller wants; the registry shows what the parties agreed. Where we do use listings — the page comparing asking prices with achieved prices — the page says so itself.

We publish medians, not averages. In Dubai a handful of AED 100M deals moves an average far enough that it stops describing a typical purchase.

Every figure carries its source and the date of the data cut. Where the sample is thin we would rather publish nothing than publish something unreliable.

02 What counts as a sale

The sales population is DLD registry records with:

• trans_group_en = "Sales" — sales only; mortgages and gifts are counted separately and never mixed into prices;
• property_type_en of "Unit" or "Villa", i.e. built residential; land and whole-building deals are excluded from price per area;
• a transaction value between AED 100,000 and AED 500,000,000;
• a price per square metre between AED 3,000 and AED 200,000.

Off-plan is taken from reg_type_en = "Off-Plan Properties"; everything else is treated as ready property.

03 What counts as a rent

Rents come from registered Ejari contracts:

• property type "Flat", "Villa", "Studio" or "Complex Villas";
• residential use (property_usage_en = "Residential");
• area between 20 and 2,000 square metres;
• annual rent between AED 12,000 and AED 1,000,000, and rent per square metre between AED 50 and AED 20,000.

New contracts and renewals are kept apart where the source distinguishes them: renewals are constrained by the RERA index, so mixing them with new lettings understates the market.

04 Mortgages and gifts

Mortgages and gifts are separate populations. We report their volumes but never fold them into prices. A registered mortgage amount is a loan amount, not a property price, and loan-to-value cannot be derived from it.

05 Medians and bounds

The median is taken across the whole filtered sample: rows are ordered and the middle is taken, averaging the two central values when the count is even. There are no arithmetic averages on the data pages.

Outliers are removed by the absolute sanity bounds listed above rather than by percentile trimming. We prefer a fixed, explainable filter to a moving one: the same transaction is in or out of the calculation regardless of the day the job ran.

06 Small samples

A figure is published only when there are enough observations behind it. Thresholds depend on what is being computed:

• price-band breakdown within an area — from 5 transactions;
• median for a project or building — from 10 transactions;
• median for an area segment — from 25 transactions;
• comparison with the previous period — from 30 transactions in each period;
• rental yield — from 50 tenancy contracts;
• citywide medians — from 100–200 observations.

Below the threshold we print an insufficient-data note instead of a number. A median over three sales describes nothing.

07 Rental yield

Yield is gross: the median annual rent divided by the median purchase price of comparable property in the same area, with no costs deducted.

What it does not account for: service charges, vacancy between tenants, agency commission, transaction costs and refurbishment. Net yield is always lower, typically by 1.5–2.5 percentage points.

08 Registration lag

Records reach the DLD registry as paperwork completes, not on the day of the deal, and the most recent weeks keep filling in for several weeks afterwards. A fresh, unfinished month therefore always looks weaker than it was.

For that reason headline figures use trailing 12-month windows, and the current incomplete month is excluded from monthly charts.

09 Construction progress

The completion percentage is the value recorded at the latest DLD technical inspection, not our estimate. Inspections are irregular.

An unchanged percentage between two dates does not mean construction has stopped — more often it means no new inspection has taken place.

The declared handover date is the date held in the project register. It is neither a developer promise nor the actual completion date.

10 Areas and naming

Registry area names are cadastral and often differ from market names: "Marsa Dubai" is Dubai Marina, "Al Barsha South Fourth" is JVC, "Madinat Al Mataar" is Dubai South. We show the market name in headings and compute on the cadastral area id.

Listings are matched to the registry by project number, not by name: area names in listings and in the registry rarely coincide.

11 Update cadence

Registry data is pulled by daily jobs, the data pages are recomputed on a schedule, and the date of the latest cut is printed on every page. If a computation fails, the page keeps its previous snapshot rather than showing an empty table.

Calendar days on which the registry did not operate — UAE public holidays — mean no registrations, not a collapse in demand, and are not presented as zero activity.

12 Limitations

What this data cannot tell you:

• the price of a specific apartment — the registry publishes transactions, not valuations;
• discounts and payment terms agreed outside registration;
• deals that were never registered;
• build quality, the managing company, or real delivery dates;
• net yield after costs.

This is market statistics. It is not a valuation and not investment advice.

13 Sources and licence

The underlying data is Dubai Land Department open data. The datasets, their size and their import dates are listed on the Data sources page.

Our calculations may be cited as "Continental Club Property, based on Dubai Land Department open data". We are not affiliated with, and do not speak for, the Government of Dubai or the Dubai Land Department.