GetVerdict
Methodology

How GetVerdict gets its number.

Every figure on the site reads from settled, government-registered transactions — not portal asks, not model predictions. This page documents the data, the formula, and the parts we explicitly don't do.

SourceDubai Land DepartmentOfficial transaction registry
Sales records—settled, registered sales
Areas indexed—distinct districts
As of—latest imported sales date
01

Data sources

All numbers across GetVerdict read from cleaned DLD source tables and app-facing snapshots, led by dld_transactions and sales_evidence_snapshot. The As of stamp above shows the latest imported public sales date.

What we include

  • Sales only — trans_group_en = 'Sales'. Mortgages and gift transfers are excluded from price aggregations because they don't reflect arms-length pricing.
  • Both resale and off-plan — reg_type_en IN ('Existing Properties', 'Off-Plan Properties'). The off-plan share is shown explicitly per area so you can weigh structural mix.
  • Valid prices and sizes — actual_worth > 0 and procedure_area > 0. AED/sqm uses meter_sale_price when present, otherwise actual_worth / procedure_area; non-positive derived AED/sqm records are dropped.

Schema disclosure

The columns we touch:

event_dateday of registration · used for windows, recency, freshness
area_name_endistrict · used for aggregation, search, similar-sale matching
project_name_enproject · prioritised in similar-sale matching
building_name_enspecific building · highest-priority matching signal
rooms_enunit type (1 B/R, Studio …) · used for unit filter
reg_type_enExisting Properties or Off-Plan Properties
trans_group_enSales, Mortgages, Gifts · we filter to Sales
actual_worthtotal transaction price in AED
procedure_areaunit area in square metres
meter_sale_priceAED per square metre · primary unit-economics field

Time windows used on the site

  • Similar sales — last 365 days. Older records get exponentially down-weighted (~18-month half-life).
  • Liquidity — last 90 days.
  • Trends league — last 12 calendar months for the trace; +1 month back to compute YoY.
  • Momentum / volume change — last 3 months vs the prior 3 months.
02

How we calculate fair price

When you file a verdict, the API filters the registry, weights each similar sale, and reports a midpoint plus a band. The steps are deterministic — no model, no LLM, no portal data.

1

Select similar sales

Sales transactions in the last 365 days for the subject area, with valid price and size. We tier by specificity — same building × same unit type > same project × same unit > same project, any unit > area × same unit > area only — and the highest-priority tier with sufficient sample wins.

2

Weight by recency and similarity

Each sale gets a weight = recency × size similarity × unit match × kind. Recency uses an 18-month half-life: a sale from 9 months ago carries roughly 70% the weight of a sale from this month.

3

Compute the midpoint and band

Weighted median AED/sqm across the pool (robust to single-penthouse outliers). The band uses weighted quartile dispersion, bounded between 6% (very tight market) and 18% (sparse evidence) above and below the midpoint.

4

Score confidence, surface evidence

Larger sample, tighter dispersion, and higher-tier evidence → higher confidence. Top 5 similar sales by weight are returned with date, project, size, price, and similarity so you can audit the answer.

Minimum sample. A verdict requires at least 6 similar sales after filtering. If the registry slice is too thin, the API falls back to a static reference fixture and explicitly downgrades confidence to Low.

03

Metrics defined

Every metric on the site is defined here exactly once, with formula and where it appears.

Recorded fair · midpoint

Verdict report, Section II
weighted median AED/sqm of selected similar sales

Weights combine recency (~18-month half-life), size similarity, unit-type match, and source weight (resale vs off-plan).

Fair band (low — high)

Verdict report and Compare unified scale
midpoint × (1 − dispersion) … midpoint × (1 + dispersion)

Dispersion comes from weighted quartiles in the similar-sales pool, bounded between 6% (very tight market) and 18% (sparse evidence).

Premium / discount %

Verdict delta chip and Compare decision banner
(ask − fair_mid) ÷ fair_mid × 100

Liquidity · 90 d

Verdict and Compare unit cards
COUNT(*) WHERE event_date >= now − 90d

A measure of how often sales settle in this area or project — proxy for market depth.

Momentum · 3 mo

Shown on /trends league and row detail
avg(last 3 months avg AED/sqm) ÷ avg(prior 3 months) − 1

Captures direction this quarter — distinguishes accelerating markets from cooling ones, where YoY alone would smooth over both.

Volume change · 3 mo

Trends league and detail
sum(sales last 3 months) ÷ sum(sales prior 3 months) − 1

Liquidity direction — rising volume suggests growing market depth, falling volume suggests buyer pullback.

YoY %

Trends league
(current month avg) ÷ (same month 12 months ago) − 1

Dispersion %

Trends row detail
(12-mo high − 12-mo low) ÷ current × 100

A high number means the area's price moved a lot over the year; low means it stayed flat.

Off-plan share

Verdict, Compare, Trends
off-plan sales ÷ total sales × 100

Structural read: high share means the area is dominated by new-build sales; low share means most activity is resale.

Gross rental yield

Verdict report, Compare, rental trends
annual registered rent ÷ asking price, or ÷ fair midpoint when asking is not set

Rental evidence uses matched DLD rental contracts within the configured recency window. Buckets with implausible midpoint yields are suppressed. Rental evidence is shown as context, not as similar-sale evidence.

Project share

Verdict report (header above evidence)
similar sales in same project ÷ total similar sales × 100

If a verdict's evidence is mostly in the same project, the answer is project-specific. Lower share means we widened to area-level evidence.

Fair high

Verdict and Compare unit cards
fair_mid × (1 + band dispersion)

The high end of the fair band from the selected similar-sales pool. This replaced the old fixed 3% negotiation buffer.

Percentile placement

Verdict report under delta
% of similar sales priced at or below the asking price per sqm

“P78” means the asking price is at the 78th percentile — higher than 78% of similar settled sales, per sqm.

Verdict pill thresholds

Used on the verdict report and Compare unit cards.

Premium / discount %
Pill
Reading
< −5%
PURSUE
Materially below the registry — actionable upside
−5% to +5%
WATCH
Within fair-band noise — neither cheap nor expensive
> +5%
PASS
Materially above the registry — overpriced relative to settled sales

Momentum chip tiers

Used on /trends league.

3-mo momentum
Chip
≥ +5%
↑↑ strong up
+1% to +5%
↑ up
±1%
→ flat
−1% to −5%
↓ down
≤ −5%
↓↓ strong down
04

Confidence score

Every verdict ships with a confidence score (0–100) and a label — High, Medium, or Low. The score is a composite of three things: how much evidence we have, how clustered the prices are, and how specific the evidence is to your subject.

High70 – 100

Same-project evidence with sample ≥ 8 and tight price dispersion. The fair band is narrow; the verdict is defensible against most counter-offers.

Medium40 – 69

Area-level evidence or partial project match. Use the band, not the midpoint — and check the Top-5 similar sales before naming a number.

Low0 – 39

Sparse area or unusual unit type. We still report a band, but it's wide; treat it as an orientation, not a price target.

The score combines (1) sample size after filtering, (2) coefficient of variation of weighted prices, and (3) the highest tier of evidence reached (same building > same project > area). Each verdict surfaces its top-5 similar sales so you can sanity-check the score against the rows it's built from.

05

Risk flags

Beyond the score, every verdict carries one or more risk flags — short signals that explain unusual properties of the similar-sales pool.

GOOD_DATA_COVERAGElow

Sufficient similar-sale coverage for a transaction-informed valuation. Treat the midpoint as defensible.

LIMITED_SAMPLEwarning

Resale or off-plan side of the pool is empty. Validate the verdict against additional inputs — broker check, building visit, second similar-sale run.

MIXED_MARKET_EVIDENCEinfo

Both resale and off-plan sales are present in the pool. The band incorporates both channels — check whether the similar sales match your subject's kind.

OFF_PLAN_FLAGinfo

Subject is off-plan. Channel pricing can differ from resale on quality and delivery assumptions; weight evidence accordingly.

Flags are informational, not blocking — they explain what the algorithm noticed about the evidence so your read of the number is calibrated. Severity is reported but never overrides the price band.

06

Fees and all-in cost

The verdict includes an Estimated fees block and an All-in costfigure so the asking price isn't reported in a vacuum.

What goes in

  • DLD transfer fee — 4% of the asking price. Statutory; not negotiable.
  • Agency commission — 2.1% (buy) or 2.0% (sell) of the asking price.
  • Other — AED 5,000 fixed stub for NOC, trustee, and admin fees.
  • All-in cost = asking + total fees.

What doesn't

  • Mortgage costs (bank processing, valuation, life insurance) — not included; depends on the lender.
  • Service charges, community fees, and ongoing carrying cost — not included.
  • Custom legal review or escrow handling outside the standard DLD trustee flow.

Every fee block is marked Estimateduntil observed against the specific transaction (e.g. a signed Form-F). The rates come from DLD's published schedule and standard market commission.

07

Limitations

The things we know we don't do well, in plain text:

  • No geographic radius yet.Project and building coordinates aren't in the source schema, so proximity matching falls back to same-project then same-area evidence rather than a literal 1-km circle.
  • Interior condition, views, and floor are not in the registry feed. Two identical units at very different floors will appear identical to the algorithm.
  • Off-plan assignments and pre-handover flipscan be mixed inside the broader off-plan procedures bucket — the source schema doesn't separate them as a distinct type.
  • USD conversion is approximate. We use a hardcoded reference rate of 1 AED ≈ 0.272 USD, updated manually. USD figures shown across the site are for orientation only.
  • “UNKNOWN” building names appear in the registry feed. Where building is unknown, we display the project name to keep evidence rows readable.
  • Sparse-area fallback. If an area has fewer than 6 valid similar sales in 365 days, the verdict falls back to a static reference fixture and the confidence label drops to Low.
  • Sqft is computed. The registry stores square metres. We display sqft as sqm × 10.7639 for comfort; any rounding error originates here.
  • Refresh failures. If a nightly refresh fails, the As ofdate in the source strip will lag — that's your signal that you're looking at older data, not a silent bug.
  • Bespoke units. Custom villas, unique penthouses, and trophy assets are hard to compare; the algorithm reports a wide band and a Low score for these on purpose.

What we explicitly don't do

  • No portal scraping. No Bayut, Property Finder, or Dubizzle data feeds the numbers. Asking prices are typed by the user; everything else is registry.
  • No predictions. No model, no LLM, no “AI-powered estimate”. Every figure is a deterministic aggregation of public transaction records.
  • No mixing of mortgages or gifts into pricing. Only arms-length sales feed the price aggregations.
  • No phantom inventory. If a unit hasn't settled and registered, it doesn't exist in our numbers.
08

Frequently asked

How fresh is the data?
A nightly job at 06:00 GST refreshes the registry mirror. The As ofdate in the source strip above shows the most recent successful refresh; if it lags more than 24 hours, the upstream feed had a hiccup and we're showing the previous day's numbers.
Why is my building not covered?
We index every building that has at least one recorded sale in the last 12 months. Brand-new towers without recorded sales, owner-occupied compounds, and government inventory don't show up in the registry and therefore don't show up here.
Can I challenge a verdict?
Yes — the verdict surfaces its top-5 similar sales with date, project, size, and price. If you can name specific records the algorithm missed (or included in error), the verdict is repeatable: change the area or kind, re-file, and the band moves with the evidence.
Is my PDF accepted by banks?
No. A GetVerdict PDF is an independent opinion sourced from public transaction records — it isn't a RICS appraisal and it isn't a regulated valuation. Banks require their own panel valuer for a mortgage. The PDF is useful as a negotiation document, a desk reference, and an audit trail; it isn't a substitute for a formal appraisal.
How is GetVerdict different from Bayut or Property Finder estimates?
Portal estimates are derived from listing asks, page views, and ML on observed listings. GetVerdict reads only settled DLD sales. Where portals tell you what people are asking, the registry tells you what people actually paid — and the gap between the two is exactly what the verdict surfaces.
Do you handle off-plan?
Yes — off-plan sales are included in the evidence pool when you mark the subject as off-plan, and the kind is shown on every evidence row. Pre-handover assignments and developer launches that haven't reached DLD registration yet are out of scope until the sale is registered.
Do you offer API access?
Not yet. The verdict, compare, and trends endpoints are internal to the web app. If you have a use case that would benefit from programmatic access, reach out — early access is on the roadmap.
What happens if there's no similar-sale evidence?
Below 6 valid similar sales in 365 days, the API falls back to a static reference fixture and downgrades confidence to Low. You'll see the band widen and a LIMITED_SAMPLE flag attached — the number is still reported, just labelled clearly as thin evidence.
Attribution

Sourced from Dubai Land Department public records. GetVerdict is an independent opinion service and not affiliated with, endorsed by, or representing the Dubai Land Department. Every figure on the site is reproducible from the cited columns and filters above.