DEV Community

Cover image for Asking Prices vs DLD Transactions in Python: Dubai Sellers Ask 11% More
Happy Endpoint
Happy Endpoint

Posted on Originally published at Medium

Asking Prices vs DLD Transactions in Python: Dubai Sellers Ask 11% More

How far apart are Dubai asking prices and the prices buyers actually pay? Every property portal shows the first number. The second sits in the registered sales records, and to answer the question you need both side by side, for the same kind of flat.

So I pulled the newest ready 1- and 2-bedroom apartments for sale in 8 Dubai areas from the Bayut API, 800 listings in all, and set them against registered resales of ready apartments in the same areas. The short answer: for the 218 listings I could match to sales in their own building, the median asks 11.2% more per square foot than recent buyers paid there, for the same bedroom count. Four in five of them ask more than their building's recent sales. And the area-wide comparison, which is how most people do this, can be off by half or more.

The data

Two sources, both read on 4 November 2026.

Asking prices. The Bayut API endpoint /search-property, with purpose=for-sale, property_type=apartments, completion_status=completed, rooms set to 1 or 2 and sort_order=latest. I took two pages of 25 results for each of 16 searches (8 areas, 2 bedroom counts). Each listing carries price, area, rooms, createdAt and a location list whose last entry is the building.

Paid prices. Registered Dubai sales from the Bayut transactions data, the same data the Bayut API's /transactions endpoint returns. I kept ready apartments sold as resales, 1 and 2 bedrooms, registered from August to October 2026: 2,075 sales in the 8 areas. For the building comparison further down I widened the window to May to October, so that more buildings reach 5 sales. Prices outside AED 100,000 to 500 million and prices per square foot outside 200 to 30,000 are dropped, the same cleaning the area pages on bayutapi.dev use.

The 8 areas all have sale figures on bayutapi.dev's location pages. They range from the towers of Dubai Marina, Downtown, Business Bay and Jumeirah Lake Towers (JLT) to Palm Jumeirah, Dubai Hills Estate, Jumeirah Village Circle (JVC) and Dubai Silicon Oasis.

The code

Here is the comparison for one area and one bedroom count, 1-bedroom flats in JVC. Set RAPIDAPI_KEY and it runs unchanged. The paid side is a constant: it is the median of the 477 ready 1-bed resales registered in JVC from August to October.

import json
import os
import statistics
import time

import requests

HOST = "uae-real-estate3.p.rapidapi.com"  # Bayut API on RapidAPI
BASE = os.environ.get("API_BASE", "https://" + HOST)
SQFT_PER_SQM = 10.7639

# Median paid for ready 1-bed resales in JVC, Aug to Oct 2026 (477 registered sales).
# Not fetched here: it comes from the registered sales data that the Bayut API's
# /transactions endpoint returns, cut as described under "Paid prices" above.
PAID_PSF = 1185


def headers():
    if "API_BASE" in os.environ:
        return json.loads(os.environ.get("API_HEADERS", "{}"))
    return {"x-rapidapi-host": HOST, "x-rapidapi-key": os.environ["RAPIDAPI_KEY"]}


def ready_listings(location_id, rooms, page):
    """One page (25) of the newest ready apartments for sale in a location."""
    params = {
        "purpose": "for-sale",
        "location_ids": location_id,      # from /autocomplete: JVC is 5416
        "property_type": "apartments",
        "completion_status": "completed",  # leave out off-plan resales
        "rooms": rooms,
        "sort_order": "latest",
        "page": page,
    }
    r = requests.get(BASE + "/search-property", params=params, headers=headers(), timeout=60)
    r.raise_for_status()
    return r.json()["data"]


first = ready_listings("5416", 1, 1)
time.sleep(1.5)
rows = first["properties"] + ready_listings("5416", 1, 2)["properties"]

# The same flat is often listed by several agents: same price, same size
unique = {(x["price"], round(x["area"], 1)): x for x in rows}.values()

# `area` is in square metres: convert before dividing
asking = statistics.median(x["price"] / (x["area"] * SQFT_PER_SQM) for x in unique)
print("ready 1-bed flats for sale in JVC: %d" % first["total"])
print("pulled %d, %d after removing repeats" % (len(rows), len(unique)))
print("median asking: AED %.0f per sq ft" % asking)
print("median paid:   AED %d per sq ft" % PAID_PSF)
print("asking is %.1f%% above paid" % ((asking / PAID_PSF - 1) * 100))
Enter fullscreen mode Exit fullscreen mode

My run printed 2,682 ready 1-bed flats for sale in JVC, 50 pulled and 49 left after removing one repeat, a median ask of AED 1,296 per sq ft against AED 1,185 paid: 9.4% above, the JVC row of the table below.

I compare per square foot, not whole prices, because the flats on sale and the flats that sold are not the same size. More on that in the pitfalls.

Asking vs paid, area by area

Sellers ask more than buyers paid in 13 of 14 cases

Area 1-bed asking 1-bed paid 1-bed gap 2-bed asking 2-bed paid 2-bed gap
Dubai Marina 2,286 1,649 +38.6% 1,714 1,727 -0.7%
Downtown Dubai 2,428 2,250 +7.9% 3,133 2,463 +27.2%
Palm Jumeirah 2,910 2,720 +7.0%* 3,564 2,712 +31.4%
Business Bay 2,049 1,823 +12.4% 1,821 1,670 +9.0%
JVC 1,296 1,185 +9.4% 1,465 1,071 +36.8%
Dubai Hills Estate 2,415 2,156 +12.0% 2,403 2,192 +9.6%
Dubai Silicon Oasis 1,034 884 +16.9% 1,114 1,002 +11.1%*
JLT 1,555 1,357 +14.6% 1,477 1,448 +2.0%

Median AED per sq ft. Asking: the 50 newest ready listings, repeats removed (32 to 50 left per cell). Paid: ready resales registered August to October 2026. * Fewer than 30 sales: 25 for Palm Jumeirah 1-beds, 21 for Dubai Silicon Oasis 2-beds. The chart leaves these two out.

Asking is above paid in 15 of the 16 cells. The median gap is 11.6% across all 16 and 12.2% across the 14 with at least 30 sales. The one exception is 2-bedroom flats in Dubai Marina, where the newest listings ask 0.7% less per square foot than recent buyers paid.

But look at the spread. In Dubai Marina, 1-bed listings ask 38.6% over paid and 2-bed listings ask 0.7% under. Sellers of small flats are probably not that much greedier than sellers of larger ones. A more likely explanation is building mix: which towers the newest listings happen to come from, which an area median cannot see. I did not test that for Marina on its own; the same-building match below tests it across all 8 areas.

Same building, same bedrooms

Each listing's location list ends with its building, and each registered sale carries the building too. So I matched every listing to the ready resales in its own building with the same bedroom count, registered from May to October, wherever a building had at least 5 of them. That gave 218 of the 757 listings left after removing repeats, in 101 buildings.

Four in five listings ask more than their building sold for

The median matched listing asks 11.2% more per square foot than its building sold for. 80.3% of the matched listings ask more, 27.5% ask more than 20% above, and 19.7% ask less. By bedroom count the median is 10.0% for 1-bed listings (129) and 13.7% for 2-bed (89).

The interesting part is how far the area-wide view drifts for the same listings. Measured against the area median, those 218 listings ask 14.1% more. Area by area it is much worse:

  • Downtown Dubai: 35.7% against the area median, 17.6% against their own buildings.
  • Palm Jumeirah: 25.4% against 11.1%.
  • JVC: 19.9% against 9.2%.
  • Dubai Silicon Oasis: 17.8% against 15.5%.
  • JLT goes the other way, 3.9% against 8.7%, because its new listings sit in cheaper towers than its sales.

A Downtown median mixes towers that trade at very different prices per square foot, and the newest listings are not spread across them the way sales are. If you build a valuation or a "below market" alert on area medians, this is the error you inherit.

A common rule of thumb says Dubai deals close a few percent below asking. That is a different measure: it is the discount on one flat that sold. This one compares everything on the market today with what sold, including listings that will be cut and listings that will never sell. Both can be true at once. Still, among the 218 listings I could match in these 8 areas, the typical one asks about a ninth more than its own building's recent sales. That is a sample of ready 1- and 2-bed flats in busy areas, not every listing on Bayut.

How much is listed against how much sells

The search response carries a total for the whole query, not just the page. Across the 16 searches that is 13,556 ready 1- and 2-bed apartments for sale, against 2,075 registered resales in three months.

Listed stock equals 9 to 27 months of ready resales

Divide the listings by monthly resales and you get months of stock at the recent pace. JVC has 2,682 ready 1-bed flats listed and about 159 such resales a month: 16.9 months. Dubai Silicon Oasis 1-beds are the fastest at 9.3 months, Downtown 1-beds the slowest of the 14 cells at 27.0.

Read it as a ranking, not a forecast. total counts repeat listings of the same flat, some flats are listed but not really for sale, and a sale shows up only once it is registered.

Pitfalls in the data

area is in square metres. The search results give area in square metres, not feet: the first Dubai Marina 2-bed I pulled had an area of 120.87, which is 1,301 sq ft. Divide price by area and you get AED per square metre, which looks like a price per square foot about ten times too high.

Compare per square foot. In Dubai Silicon Oasis the median 2-bed listing is 1,201 sq ft and the median 2-bed that sold was 1,296 sq ft. Whole prices say listings ask 7.7% more than buyers paid; per square foot it is 11.1%. In Downtown the bias runs the other way and is much bigger: 2-bed listings are a median 1,506 sq ft against 1,305 sold, and the whole-price gap of 45.2% shrinks to 27.2% per square foot.

Filter both sides to ready resales. In the 12 months to October, the 8 areas registered 26,924 apartment sales. Only 41.9% were resales of ready flats. 42.8% were off-plan units sold by developers, 9.4% were off-plan resales and 5.9% were finished units sold first-hand by developers. Compare a ready listing with all sales and you are partly comparing it with launch prices. On the search side, completion_status=completed keeps off-plan resale listings out.

Remove repeats. 43 of the 800 listings (5.4%) had the same area, bedroom count, price and size as another one: the same flat advertised twice, usually by two agents. In Dubai Silicon Oasis 2-beds, 18 of the 50 were repeats. Drop them before taking medians, as every figure here does, and remember that total still counts them.

Say which window you used. Against 12 months of resales instead of 3, the median gap falls from 11.6% to 8.1%. In 12 of the 16 cells, the median paid over the last three months was lower than over the full year. Ready resales in Dubai Marina (all bedroom counts) went from a median AED 1,938 per sq ft in November 2025, the first month of the window, to 1,608 in October 2026, with a peak of 2,055 in December. Asking prices are today's, so compare them with recent sales.

sort_order=latest returns fresh asks. The median listing was 1.6 days old when I pulled it, and 648 of the 757 were under a week old. The few older listings in the sample did not ask less in any consistent way, but this is the price sellers start at, before any cut.

Do not filter on isVerified. Only 73 of the 800 listings had it set. Filtering on it would throw away nine listings in ten.

What I would do next

Three follow-ups. Run the same-building match for every area and bedroom count, with a longer window for buildings that trade rarely. Re-pull the same listing IDs every week to see how many get cut, and by how much, before they sell. And take the paid side straight from the /transactions endpoint with the same filters, so the whole comparison runs from one key.

The Bayut API is on RapidAPI, and the bayut-api repo on GitHub has Python examples for its other endpoints. This LinkedIn overview covers what each one returns. For the transactions side, bayutapi.dev has a guide to Dubai DLD transaction data: what the records are and how to filter them by area and property type. If you would rather have the full registered sales history as a file, it is the Dubai Real Estate Transactions Dataset, and for anything else, get in touch.

Happy Endpoint is an independent provider, not affiliated with Bayut or the Dubai Land Department.

Top comments (0)