Methodology

The question

Most property research in Dubai compares communities: one district against another, one year against the next. That answers where prices moved, but not why one unit outperformed its neighbour in the same building.

Our question is narrower and harder: when two units are comparable on paper, which characteristics explain the difference in their registered prices, and under which conditions? The answer matters for anyone buying more than one unit, because a portfolio is a set of individual units, not a set of communities.

The data

Registered sale transactions from the Dubai Land Department (DLD), covering about two decades. These are registered prices, not asking prices.
Unit-level records, where licensed, so that the same unit can be followed through several sales and compared with its neighbours.
Rental contracts and building records, to separate price effects from rental effects and to describe each building.
Listing data, to measure the gap between asking prices and registered prices, a signal of liquidity and bargaining power.

We are currently securing access to the full unit-level data set. Until then, the work is on the framework, the test design and public data.

60+ factors in ten categories

The model groups more than 60 candidate factors into ten categories. Each factor is a hypothesis to test, not an assumption.

Category	What it covers
Macro-location	Community and submarket, distance to metro stations, business districts and the coast, schools nearby, how finished the neighbourhood is, freehold status
Building and project	Height, age, number of units, amenities, service charges, construction quality, elevators, how the plot is oriented
Unit features	Floor, view, orientation and sunlight, noise, corner position, layout efficiency, size and bedrooms, outdoor space, finish
Developer and brand	Delivery and handover track record, branded and unbranded residences, payment plan design
Market cycle	Timing relative to peaks and troughs, the off-plan premium over time, time since handover
Transaction type	Off-plan or secondary, cash or financed, single or bulk purchase, resale restrictions
Legal and ownership	Freehold zones, title status, escrow and project registration, residency thresholds
Financing	Payment plan structure, mortgage availability, investor sentiment
Liquidity	Time on market, the gap between asking and registered prices, turnover per building, share of investors
Rental	Short-term and long-term rental rules, vacancy, furnishing, size effects on rental income

The framework is public. The exact factor definitions, their weights and the results are not (see "What we publish" below).

How we test

Same building first. Every comparison starts inside a building or project. Statistically, this is a building fixed effect: it removes everything the units share, such as community, developer, age and amenities, and leaves the differences between units.
Conditions, not averages. A factor rarely works the same way everywhere. We test interactions, for example floor level together with elevator capacity, view, facade orientation and noise, to find out when an effect appears and when it disappears.
Non-linear effects. Many effects grow, flatten or reverse. Floor level, for example, is modelled with curves rather than a straight line.
Time controls. The market cycle and the off-plan premium both change over time. Quarter and year effects keep timing from being mistaken for a unit effect.
Segments. Price drivers differ between the mass market and the luxury market. A study of 140,039 Dubai transactions found that price determinants vary across price levels and are non-linear (Worku, 2017). We therefore also estimate effects by segment, using quantile regression.
Bulk purchases separately. Research on Dutch land registry data finds that real estate portfolios most often trade at a price discount (Ishaak, Liu, Hardeman and Remoy, 2024). Bulk deals are not flagged as such in public transaction data, so we identify likely bulk transactions ourselves and keep them from distorting unit-level results.
Out-of-sample checks. A result is only used if it holds on data the model has not seen.

From factors to portfolios

These are the portfolio construction approaches we plan to test. They describe a method, not a recommendation, and they will only become a service once licensing is in place.

Choose the unit, not only the building. Within one tower, exposure to view, floor, orientation and noise differs from unit to unit. Selection happens at unit level.
Spread factor exposure. A portfolio can be concentrated in one theme without looking like it, for example many units with the same handover date or the same type of view. We aim to spread exposure across communities, segments, handover dates and factors.
Plan the exit at entry. Turnover per building, the gap between asking and registered prices, and the number of identical units handed over at the same time all affect how easily a unit can be sold or let later.
Price the off-plan discount properly. A launch discount has to be weighed against the capital tied up and the time to completion. We explain this with a worked example.
Put negotiated terms in writing. Discounts for buying several units are only reliable when they are written into the sale and purchase agreement.
Track every unit. After purchase, each unit is compared with registered transactions of comparable units, so that the portfolio is measured against evidence rather than against list prices.

What we publish and what we do not

We publish the framework, the methods and the open research questions, on this website and on LinkedIn.
We do not publish factor weights, scores or rankings, and we do not publish findings about individual developers, projects or buildings. The research results are the company's own work and are used for clients.
We do not publish return forecasts or performance figures.
Fees and commissions. We intend to be paid through a combination of client fees and commissions, and to disclose every fee and commission to the client in writing before a purchase. For off-plan purchases, the intention is to agree a baseline commission with developers and to pass any commission above that baseline on to the client as a discount on the purchase price. On the secondary market, commissions are often shared between several parties, such as the listing and the buying broker. Those shares will be disclosed as well. These arrangements are intended and remain subject to licensing and final terms.

Frequently asked questions

Is there a high-floor premium in Dubai?

No peer-reviewed study has tested it for Dubai yet. International studies find that the premium depends on conditions: in Taipei, a premium per floor in buildings with elevators but a discount per floor in walk-up buildings; in South Korea, a discount for top-floor units linked to heating and cooling costs. Testing this inside Dubai's towers is one of our first research questions. Read more.

What data does UAE Portfolio use?

Registered sale transactions from the Dubai Land Department over about two decades, unit-level records where licensed, rental contracts, building records and listing data. We use registered prices, not asking prices.

Does UAE Portfolio rate developers or projects?

Not publicly. Developer track record is one of the ten factor categories in the model, but we do not publish scores, rankings or findings about named developers, projects or buildings.

How will UAE Portfolio be paid?

Through a combination of client fees and commissions, every one of them disclosed to the client in writing. Developer commission above an agreed baseline is planned to be passed on to the client as a discount on the purchase price.

Is UAE Portfolio licensed?

Not yet. UAE Portfolio is a validation-stage project. The waitlist tells us whether to build the service, and for whom. Any future service will be provided only under written terms, once the required licences are in place.

How can I get access?

Join the waitlist on the home page. If you are looking at off-plan projects on sale today, the off-plan map from our licensed partner broker is linked there too.

Sources

Chen, Y.-H., Peng, C.-W. and Yang, C.-H. (2022). Vertical Housing Price Gradient and Ground Floor Premium in Taipei City. International Real Estate Review 25(2), 199 to 216.
Hwang, S. and Ma, J. (2022). Top-floor discounts in residential buildings: Evidence from South Korea. Real Estate Economics.
Ishaak, F., Liu, P., Hardeman, E. and Remoy, H. (2024). Portfolio sales: price premium or price discount? ERES working paper eres2024-206.
Mohammad, S. I., Graham, D. J. and Melo, P. C. (2015). The effect of the Dubai Metro on the value of residential and commercial properties. Journal of Transport and Land Use.
Worku, G. B. (2017). House price drivers in Dubai: nonlinearity and heterogeneity. International Journal of Housing Markets and Analysis 10(3), 384 to 409.