Context behind every number.

The five-category profile model.

A profile with a completed category review displays Regulation, Risk management, Customer support, Trading costs and Reputation, each worth 20 points. Each contribution combines up to 10 points from the same normalized overall reference with up to 10 points from its specific FactFX review. The common reference baseline is not five independently supplied external subscores.

Balanced rounding to one decimal keeps the five displayed contributions equal to the total out of 100. Category criteria, rationale, scope and evidence appear in the profile. Profiles without a completed category review retain their existing offering assessment; missing categories are not populated from another broker or from an unrelated rubric. A manual administrator override retains its written reason and removes unrelated category bars. A separately recorded administrator category allocation may be displayed only when its five values sum exactly to that assigned total.

Profile-informed category allocations.

Every scored broker profile can display Regulation, Risk management, Customer support, Trading costs and Reputation without changing its existing overall score. A completed category review or a valid explicit administrator allocation is preserved. Otherwise documented profile features guide the relative allocation, with a 20-point cap per category. No random or brand-name-based offsets are used.

Company and licence disclosures, register references and the recorded match status inform Regulation. Practice accounts, documented account controls and risk-policy references inform Risk management. Existing resource reviews and linked help or learning materials inform Customer support. Published commission bases, account-model reviews, other-fee coverage and matched average-cost illustrations inform Trading costs. Company-record traceability, corporate-reporting references and explicitly documented history inform Reputation. These assess the available profile information, not independently tested outcomes; no new authorisation, protection or service claim is created.

Missing information retains a neutral factor of 1. Documented features add relative emphasis under the published policy. The initial allocation is the existing total divided by five, plus 2.5 times each factor's difference from the mean factor. A common balancing adjustment respects the 0–20 bounds and preserves the original total. Largest-remainder rounding in integer tenths keeps the five displayed values exactly equal to the displayed total. The original overall maximum and grade are retained.

The profile labels this “Profile-informed category allocation” and lists each category's contributing features. Headline minimum spreads and bonus percentages alone add no emphasis. A spread-plus-commission illustration uses a published EUR/USD average with a matched USD round-trip commission, excluding financing, conversion and execution effects. A pending total remains pending. These are editorial allocations, not five independent measurements or a safety certification.

How our assessment works.

The existing offering model uses the following rubric worth 100 points within the FactFX half. These allocations sit inside our 50% component; they are not extra weights in the final score.

• Platforms, devices and tools · 30 points. Documented trading interfaces, device coverage, charting, orders and useful tools.
• Account and product choices · 25 points. Useful account options and product coverage within the offering’s stated market.
• Pricing and account models · 25 points. Available spread, commission and account structures, including relevant terms and options.
• Support, learning and research · 20 points. Documented help channels, educational material and research resources.

The four awarded values sum to our assessment out of 100. A focused forex offering can score strongly within its market; one capable platform can meet users’ needs as well as several platform names. Pricing-model points assess the offering’s structure and choices, rather than claiming measured cheapness.

A blocked register lookup or an unavailable representative spread average does not itself reduce these offering points. Such factual checks retain their unresolved status. Published features are assessed from documents; live execution, withdrawal quality and actual platform performance have not been tested.

Each profile has individual reasons. The administrator can save a subjective override with a written rationale; that value replaces the initial assessment and the reason is retained. No fixed score range is imposed. For example, an assessment of 80 contributes 40/50 framework points. Previously recorded final totals remain unchanged on the current 100-point display.

Administrator-assigned totals.

An administrator can assign a separate total from 0 to 100 with a written reason. That total determines the displayed grade and is labelled as an administrator-assigned total. It replaces the normal composite for presentation, without changing the external reference or the underlying factual profile. Its detail panel preserves the original calculated composite for context. The two totals are not combined. Where separately documented, five administrator category allocations may explain the assigned total, each capped at 20 with the overall 100-point cap retained. They are labelled as administrator allocations and are not independently measured category results. If their sum differs from the total, those custom values are discarded; a valid total may instead use the documented profile-informed allocation.

Broker grades · the score at a glance.

The seal beside a broker's name follows its displayed composite score: A · 90 and above; B · 80–89.9; C · 70–79.9. These are displayed points on the profile's applicable scale, without an additional conversion. A changed assessment automatically changes the grade.

A score below 70 or a pending assessment has no A, B or C seal. The seal identifies a FactFX score band; it is not regulatory approval or a safety certification. Warning-only records use their own warning seals and are excluded from these grades.

Warning assessments · a separate notice-based policy.

Warning-only profiles use individually reviewed FactFX editorial scores from 10 to 25 on a 100-point scale, in half-point steps. A case's documented concern determines the applicable band; its specific notice and written review determine its score within that band. Equivalent evidence receives consistent scores. No random numbers or minimum variety target is used.

The review benchmarks are 23 for authorisation-only cases; 21 for an association-rejection notice; 16 for an explicit false-licence claim; and 11, 12 or 13 for impersonation of an authorised firm, registered firm or recognised fund respectively. These are FactFX editorial starting judgments, not scores issued by the authority. A specifically documented reuse of a genuine firm's previous address reduces a clone assessment by 1 point. Each additional named warned website reduces the assessment by 0.5, capped at 2 points. Explicitly different operating aliases or named social distribution channels can each reduce an authorisation assessment by 0.5. These reflect the notice's identity and exposure scope, not a claim that a domain count proves an extra offence.

For association-rejection notices, literal name comparisons can refine the 21-point benchmark: a shared distinctive word with another descriptor, minus 0.5; the full core company name in the warned domain, minus 1; a single-letter lookalike of the core name, minus 1.5. This is FactFX's written judgment about the names recorded in the notice. It does not upgrade a suspicious-website notice to an official clone or fraud finding. An abbreviation without clear literal overlap receives no name adjustment.

Each record needs an official notice, a matching policy and an individual case review. A new or changed notice without that review, an unavailable licence lookup, an unsupported allegation or an unmatched suspicious-website concern remains unscored. Unknown trading terms, missing genuine-company websites, different address countries and ordinary spelling discrepancies add no deduction. These warning assessments are separate from the catalog’s 50:50 offering model and do not invent trading subscores. A lower score expresses the review's warning concern; it is not a probability of fraud, a measured loss estimate or the regulator's numeric rating. “Suspected” and domain-specific identity concerns retain their stated certainty, and an authorisation warning remains limited to its jurisdiction.

Keep the scope in view.

The reference follows its recorded brand, group or product-review scope. It does not establish current authorisation for a particular company or eligibility for an account. Profile records retain their company, website edition and account scope; current-register checks and provider-published statements keep their actual status.

The catalog combines dated reference snapshots with clearly identified archived records. Each profile shows its reference date and historical status. These are recorded assessments, not a live feed; an archived reference is not described as current.

Spreads, commissions, account features and deposit-bonus terms remain separately sourced profile information. Documented offering features inform our assessment, while bonus percentages and headline spreads receive no separate automatic points. Unknown facts remain unknown. Comparing costs requires a representative average and commission for the same account, size, currency and period; an advertised minimum is not that average.

Our assessment and any administrator override are judgements, not independently tested facts. Commercial relationships do not justify hidden increases.

FactFX does not claim hands-on execution, withdrawal, support, usability, accessibility, speed or reliability tests. A composite rating does not resolve an outstanding factual check, certify regulatory status, guarantee safety or predict returns.