Go to FxProHow the score is calculated
Every broker on this site carries one number between 2.0 and 5.0. This page says what goes into it, what it weighs, where each figure was read, and — the part that usually goes unsaid — what happens when a figure is not there at all.
The formula
Each axis is first scaled to 0–100 against fixed anchors. The axes that have data are then averaged, each by its weight, and the result is mapped onto the 2.0–5.0 range. The floor is 2.0 rather than zero because a broker that reached this base is already one that serves clients in United Arab Emirates; the scale separates them from each other, it does not rate them against a firm that does not exist.
The anchors are absolute, not the best and worst value in the country. A population-relative scale moves every broker's score whenever a single new record is added, and one outlier flattens the rest of the field into a cluster. Here, a $7.50 round turn scores the same today as it will next quarter.
The six axes
| Axis | Weight | What is measured | Where it comes from |
|---|---|---|---|
| Trader reviews reviews | 25% | A public rating read on a 0–5 scale. 3.0 scores 0 and 5.0 scores 100, in a straight line between the two. Fewer than 30 reviews and the axis counts as missing rather than low — a 5.0 out of four reviews is not evidence. | Bank fields reviews. The platform the rating was read from is stored with the value. |
| Round-turn cost cost | 20% | The cost of one round turn of one standard lot of EUR/USD, in US dollars: raw spread × 10 plus the round-turn commission, or the standard spread × 10 when only that is known. $5 or less scores 100, $25 or more scores 0. | Bank fields spread_eurusd_raw_pips, commission_round_turn_usd, spread_eurusd_standard_pips. Below one dollar a round turn is treated as half-known and dropped: an ECN broker quoting 0.0 pips takes its money in the commission, and the missing half once made it the cheapest broker in a country at $0.00. |
| Licence and entity safety | 20% | Points, capped at 100: 60 for a licence from this country's own regulator, 30 for being onboarded by a local entity rather than an offshore one, 10 for no regulator warning on record. | Bank fields local_regulator_authorised, entity_type, regulator_warning. Knowing only that no warning was found is not knowing anything, so the axis needs the licence or the entity type before it scores at all. |
| Withdrawal speed payouts | 15% | Hours from a withdrawal request to the money leaving, parsed from the broker's own stated terms. 0 hours scores 100, 120 hours — five business days — scores 0. | Bank fields withdrawal_hours, withdrawal_time_hours. Stated as prose in the source ("up to five business days") and parsed, not guessed. |
| Track record track_record | 10% | Years since the firm was founded, counted to 2026 and capped at 20. Everything at or above the cap scores 100. | Bank fields year_founded. |
| Platforms and programs platforms | 10% | Points, capped at 100: MT4 25, MT5 25, cTrader 20, mobile app 15, demo account 15. | Bank fields platforms, mobile_app, demo_account. |
Weights sum to 100%. They are stored in the scoring configuration, not in the page template: this table is generated from the same file the scorer reads, so it cannot describe a formula that is no longer in use.
When a broker publishes nothing on an axis
A missing figure is not a bad figure. An axis with nothing behind it drops out of the calculation and the remaining weights become the denominator — a broker measured on four axes is scored on those four, not punished with zeros for the other two. Scoring a gap as zero would be a claim we cannot support, and it is the most common way a review site invents a ranking.
There is a floor on how little may be known. When the axes that do have data carry less than 55% of the total weight, no score is shown at all: the page states that the base is too thin and names what is missing. In United Arab Emirates, 74 of 75 brokers clear that threshold.
The card lists the axes one by one and marks the ones with nothing behind them, so a 4.1 built on six axes and a 4.1 built on four are not presented as the same statement.
Shown on a card is not the same as ranked in a table
Enough weight to compute a score is not enough to compare it. A broker with no review data is scored on its remaining strengths and lands above firms that were measured on everything, which is an artefact of the gap rather than a finding about the broker.
So a score enters a ranking only when both Trader reviews and Round-turn cost are present. Those two carry 45% of the formula between them. A score without them is still true and still shown on the broker's own card, with a note saying why it is not in the tables. In United Arab Emirates that leaves 54 of 75 brokers eligible for a ranking.
A figure for this country, or for the whole group
A broker often publishes one set of terms worldwide and another for the entity that actually onboards residents of United Arab Emirates. Minimum deposit and maximum retail leverage are ranked only when the value was measured for this country; a worldwide figure for those two is shown on the card with a group marker and is kept out of every comparison. Without that rule a headline leverage published for an offshore entity wins a local "high leverage" page it has no business being on.
For the other fields a group-wide figure is accepted and marked as such. A value recorded with an unclear scope — the source does not say who it applies to — is never ranked anywhere; in United Arab Emirates there are 3 of those.
Regulator warnings
A warning on record does not change the arithmetic. It would be a second, invisible
adjustment on top of the formula, and the published score would stop matching the published
weights. What it does instead is remove the firm from the editorial shortlists
(beginners, most_established), where a reader is being pointed at a specific broker rather than shown the field.
When a comparison page is published at all
A comparison page exists in a country only when at least 8 brokers there carry a rankable value for the metric that drives it. Below that the page would be three rows under a headline that implies a survey, which is the shape of a page that misleads without stating anything false.
What is actually measured in United Arab Emirates
Coverage is uneven and published as it is. 75 brokers were measured for United Arab Emirates; this is how many of them carry each axis.
| Axis | Brokers with this axis | Share |
|---|---|---|
| Trader reviews | 68 of 75 | 91% |
| Round-turn cost | 60 of 75 | 80% |
| Licence and entity | 75 of 75 | 100% |
| Withdrawal speed | 65 of 75 | 87% |
| Track record | 72 of 75 | 96% |
| Platforms and programs | 75 of 75 | 100% |
Read it as a warning as much as a statistic: the axes at the bottom of this table are the ones where a score leans on the renormalisation described above.
Where the figures come from
Each figure is stored with the sentence it was read from, so a number can be traced back to the wording that produced it instead of to a spreadsheet cell. Of the 801 values the formula reads in United Arab Emirates, 791 carry that quoted sentence and 471 carry a direct link to the page it came from. 67 of the 75 broker records list their sources.
Licence status is read against the register of Three regulators by jurisdiction: (1) federal Securities & Commodities Authority (SCA) on the mainland - formally renamed Capital Market Authority (CMA), uaecma.gov.ae, effective 1 Jan 2026 (verify branding); (2) Dubai Financial Services Authority (DFSA), dfsa.ae, within the DIFC free zone; (3) Financial Services Regulatory Authority (FSRA) within Abu Dhabi Global Market (ADGM). rather than against the broker's own description of itself. Records for United Arab Emirates were last re-captured between 2026-07-09 and 2026-08-27; the recalculation date is printed on every page that carries a score.
What the score does not measure
The formula is the six axes above and nothing else. Execution quality, slippage, overnight swap rates, the behaviour of customer support and the experience of a withdrawal dispute are not in it — not because they do not matter, but because we have no measurement of them that would survive being quoted. They are not folded in as an opinion either: there is no editorial adjustment applied to a score after the arithmetic.
There is also no commercial input of any kind. Placement is not sold, no row in any table is paid for, and the affiliate arrangements described in the affiliate disclosure are applied after the order is computed, never before it. How that is kept true is set out in the editorial policy.