FxBrokersListing Broker Review and Scoring Methodology

FxBrokersListing evaluates forex and CFD brokers using a structured scoring system based on measurable broker data. The purpose of the methodology is to make broker ratings consistent, transparent and reproducible.

Every broker is evaluated across ten main categories. Individual metrics are scored directly on a scale from 0 to 5. Metric scores are combined into subcategory scores, subcategory scores are combined into category scores, and category scores are weighted to calculate the final FxBrokersListing Rating out of 5.

Where a metric can be measured numerically, FxBrokersListing uses a predefined quantitative rule rather than an editorial judgment. Continuous market and operational metrics use the distribution-calibrated scoring process in Section 2.2 unless a documented external consumer-protection or contractual anchor applies. Binary and categorical metrics use closed, testable criteria.

Each broker review identifies the reviewed client jurisdiction, serving legal entity, and account type used for scoring. A score applies to that stated review configuration and is not automatically assumed to apply to every entity or account operated under the same broker brand.

1. Overall FxBrokersListing Scoring System

All calculations use unrounded values. The final published rating is rounded to two decimal places.

Example: a Trust & Safety score of 4.20 contributes:

1.1 Standard Feature Rule

Some metrics measure whether a specific feature is available rather than measuring a continuous numerical value.

The 2.5 conditional state applies only when the feature is available within the scope being scored but is subject to a documented eligibility condition. The scoring scope is the reviewed entity/account unless the individual metric explicitly measures options available across the reviewed entity or jurisdiction, such as account-option metrics. A feature available only outside the applicable scoring scope is not treated as conditionally available for that review.

1.2 Feature-Count Influence Rule

A metric whose primary input is only the number of available features, channels, account variants or platform families is treated as a breadth metric, not as a direct measure of trading quality. Outside the Tradable Instruments and Market Research categories, where breadth is part of the subject being measured, pure count/breadth metrics may contribute no more than 30% of a category score in total. The remaining category weight must be based on costs, protections, processing outcomes, task completion, verified user outcomes or other directly observable performance measures.

A broker cannot receive additional points merely because the same underlying capability is offered under multiple brands, platforms or marketing names. Duplicate functionality is counted once unless the methodology explicitly measures cross-platform consistency.

1.3 Broker-Excluded Category Average

Each broker review displays a comparison benchmark beside each published category score. This benchmark is the simple arithmetic mean of the same category score across the eligible FxBrokersListing comparison universe, excluding the broker currently being viewed.

The benchmark is displayed only when at least 100 other brokers have a published score for that category under the same methodology version. The broker being viewed is never included in its own category benchmark.

To be included in the benchmark, a broker must:

1. have a published score for the category under the same methodology version;
2. have sufficient verified data for that category under Section 2.1;
3. use the same category definition and scoring rules; and
4. contribute only one score to the benchmark comparison set.

Category scores marked Insufficient Data are excluded. The mean is calculated from unrounded category scores and rounded to two decimal places only for display.

The broker-excluded category average is a context benchmark only. It does not change the broker's category score, category weight, overall FxBrokersListing Rating or data-coverage calculation.

2. Evidence and Verification Standards

FxBrokersListing uses the most authoritative source reasonably available for each data point.

1. Official regulator registers and government records
2. Broker legal and contractual documents
3. Official broker fee schedules and trading specifications
4. Official broker product, account and platform pages
5. Direct FxBrokersListing testing where applicable
6. Official platform or payment-provider documentation
7. Eligible third-party financial-data or broker-research sources that meet the source-eligibility test below

When sources conflict, the source higher in the hierarchy normally takes precedence unless there is evidence that it is outdated or does not apply to the reviewed legal entity.

2.0 Third-Party Source Eligibility Test

A third-party source may supply a scored numerical input only when all of the following are verifiable:

1. the publisher and page or dataset are identifiable;
2. the value is dated or the observation period is stated;
3. the value is identified as measured, average, median or otherwise statistically defined, rather than a broker advertisement copied without verification;
4. the methodology or measurement basis is described sufficiently to identify what was measured;
5. the data apply to the relevant broker, account type and instrument; and
6. the source was published or updated within the maximum age specified by the metric.

If any required condition fails, the third-party value is not used as a scored observation.

A metric-specific evidence rule may impose a stricter source age, observation period, account-selection rule or statistical requirement than this general eligibility test. Passing Section 2.0 makes a third-party source eligible; it does not override stricter requirements stated for an individual metric.

2.1 Confirmed Absence, Unverified Data and Not Applicable

Confirmed absence

If reliable evidence confirms that a feature is unavailable, it receives the score specified for absence, normally 0. Confirmed absence is different from a feature that could not be verified.

Unverified information

An unverified metric is not automatically treated as zero. It is marked Unverified and excluded from scoring only when the minimum data-coverage rules below are still satisfied.

For factual availability, protection, cost and trading-condition inputs, every observation must resolve to one of three states: Verified Present, Confirmed Absent or Unverified. Failure to locate sufficient evidence is not by itself proof of absence.

Disclosure and transparency metrics are different because the observable being scored is the disclosure itself. For those metrics, a required disclosure may score as not disclosed when the prescribed source review confirms that the disclosure is not present; that result must not be used to infer that the underlying feature, protection or trading condition itself is absent.

A subcategory receives a published score only when at least 70% of its original metric weight has verified, scoreable data. When the 70% threshold is met, unverified metrics are excluded and the verified applicable metric weights are proportionally normalized. If less than 70% is verified, the subcategory is marked Insufficient Data.

A category receives a published score only when subcategories representing at least 70% of that category's original weight have published scores. If the threshold is met, the scored subcategory weights are proportionally normalized. Otherwise, the category is marked Insufficient Data.

An overall FxBrokersListing Rating is published only when:

1. Trust & Safety has a published category score;
2. Fees & Commissions has a published category score; and
3. published category scores represent at least 90% of the original overall category weight.

If overall score coverage is between 90% and 99.99%, unscored categories are excluded, and the remaining category weights are proportionally normalized. The review must display the resulting Rating Coverage % beside the overall rating. If coverage is below 90%, no overall numerical rating is published.

Rating Coverage and Evidence Coverage

Rating Coverage measures the percentage of the original overall category weight represented by categories that have a published score and are included in the final FxBrokersListing Rating. This is the coverage measure used to determine whether an overall numerical rating may be published under the 90% rule above.

Rating Coverage does not represent the percentage of all underlying metric evidence that was verified.

Evidence Coverage separately measures how much of the applicable underlying scoring evidence has verified, scoreable observations before any normalization of missing metric, subcategory or category weights.

For this calculation, each metric's original overall-equivalent weight is determined from its original category, subcategory and metric weights before any missing-data normalization. Metrics correctly classified as Not Applicable are excluded from both the numerator and denominator. Unverified applicable metrics remain in the denominator but not the numerator.

Rating Coverage and Evidence Coverage are reporting measures only. Evidence Coverage does not alter metric scores, subcategory scores, category scores, category weights or the final FxBrokersListing Rating.

Not applicable

A metric may be treated as not applicable only when the feature genuinely does not apply to the reviewed broker configuration and the scoring rule permits exclusion. A feature that is merely difficult to verify is not Not Applicable.

Not-applicable and permitted unverified items are therefore removed only under the coverage rules above, and the remaining scoreable weights are proportionally normalized.

2.2 Threshold Calibration and Benchmark Governance

Continuous market, cost, time, amount, rate and volume metrics are scored against a frozen FxBrokersListing calibration universe rather than against editor-selected bands. The calibration universe consists of one current observation per published broker review configuration with verified data for the metric. A distribution-calibrated metric requires at least 60 verified broker observations.

For each methodology version, FxBrokersListing freezes and records the calibration date, sample size, median and the 10th through 90th percentile cut points for every distribution-calibrated metric. The frozen calibration record must also retain an ordered value-frequency table, empirical rank table or equivalent record sufficient to reproduce the exact percentile and tie treatment for any observed value. Those records form the Calibration Appendix for that methodology version. A distribution-calibrated metric cannot be used in a published rating unless its calibration record is present.

Percentile direction is declared in the metric: lower is better for costs, spreads, fees, minimums and processing/response times; higher is better for market coverage, instrument counts, support-language coverage and qualifying research volume. Ties use the mid-rank formula above.

Distribution-calibrated scores express relative standing within the frozen FxBrokersListing calibration universe for that methodology version. They are not absolute quality thresholds unless the individual metric explicitly uses an external or consumer-utility anchor.

A fixed absolute scoring threshold may replace distribution scoring only when the threshold is derived from an external rule or an explicitly defined consumer-utility breakpoint—for example, a statutory protection amount, a regulatory requirement, a contractual yes/no protection or a zero-fee condition. The methodology record must identify the external anchor and rationale. Unsupported editorial breakpoints are not permitted.

Calibration is reviewed at least once every 12 months. New cut points take effect only under a new methodology version or effective date, and all brokers displayed together are recalculated under the same frozen calibration snapshot. Commercial relationships, broker size and brand recognition cannot alter the calibration universe, score, threshold or weight.

2.2A Weight Calibration and Sensitivity Control

Category and subcategory weights express the dimensions FxBrokersListing intends the rating to represent, and the methodology is tested against the broker dataset before a methodology version is published. The calibration record must contain two separate sensitivity tests:

1. Leave-one-metric-out test: each scored metric is removed individually from its immediate parent scoring group. The remaining metric weights within that parent group are proportionally normalized, and all affected subcategory, category and final broker ratings are recalculated. The test is repeated separately for every scored metric.
2. Subcategory-weight perturbation test: each scored subcategory weight is separately increased by 20% and decreased by 20% relative to its stated methodology weight. For each test, the other subcategory weights within the same category are proportionally normalized so that the category continues to total 100%, and the affected final broker ratings are recalculated.

These sensitivity tests are diagnostic governance tests unless this methodology states an explicit publication limit for the affected metric. The specific publication limit for pure feature-count or breadth metrics remains the rule stated below.

A pure feature-count or breadth metric outside Tradable Instruments and Market Research may not be published at a weight that allows that single metric to change the final 5-point rating by more than 0.10 points between its minimum and maximum score. If the sensitivity test exceeds 0.10, its weight must be reduced before the methodology version is released.

The calibration record reports the median and 95th-percentile absolute change in final broker ratings under the perturbation tests. Weight changes are versioned and applied to the complete comparison set; they are never adjusted for an individual broker.

2.2B Calibration Universe Eligibility and Freeze Integrity

The composition of the calibration universe must be reproducible in addition to the percentile calculations derived from it. This section documents and audits the application of the existing calibration-universe rules in Section 2.2; it does not create a separate scoring basis or alter any existing metric definition, threshold, weight or score.

For every distribution-calibrated metric, the Calibration Appendix must retain a Calibration Universe Manifest identifying every broker review configuration considered for inclusion in the frozen calibration sample and its inclusion or exclusion status under the existing methodology rules.

For each configuration, the manifest must record, at minimum:

• broker review configuration identifier;
• broker brand;
• serving legal entity;
• reviewed client jurisdiction;
• applicable account type where relevant;
• observed value where available;
• observation or verification date;
• inclusion or exclusion status;
• the existing methodology rule supporting any exclusion; and
• calibration snapshot identifier.

Inclusion in the manifest does not itself make an observation scoreable. Eligibility continues to be determined solely by Sections 2.0, 2.1, 2.2 and the applicable metric-specific evidence and selection rules. The manifest records how those existing rules were applied.

Where multiple published pages, account variants or brand names ultimately represent the same review configuration for the metric being calibrated, the calibration record must identify the single observation selected under the applicable methodology rule so that the one-observation-per-review-configuration requirement can be independently reproduced.

Once a calibration snapshot has been frozen for publication, brokers or observations may not be selectively added to or removed from that snapshot in order to alter a percentile, cut point or score. A factual correction to the frozen calibration dataset must be documented. If the correction requires replacement of the published calibration snapshot, the replacement must follow the methodology-version or effective-date controls already stated in Sections 2.2 and 16 and must be applied consistently to the affected comparison set.

Commercial relationships, broker prominence, review traffic, advertising status, affiliate status and expected scoring effect must not determine inclusion in or exclusion from the calibration universe.

The Calibration Appendix should also report descriptive composition statistics for the frozen universe where practical, including the number of contributing review configurations and material scope characteristics such as reviewed jurisdictions. These statistics are transparency measures only and do not alter any broker score.

2.2C Measurement Comparability and Observation Metadata

Numerical observations used under this methodology must be documented sufficiently to establish that they correspond to the observable already defined by the applicable metric. This section adds documentation and comparability controls only; it does not replace or modify any metric-specific evidence rule.

For every continuous numerical metric, the research record must identify, where applicable:

• the statistic being used, such as mean, median, upper bound, fixed value, observed completion time or published rate;
• the observation period;
• the observation date or date range;
• the relevant account type;
• the relevant instrument or product;
• the reviewed legal entity and jurisdiction;
• the original measurement unit;
• any currency conversion or unit conversion;
• any market-hours or time-zone basis stated by the source;
• whether the value is broker-published, independently measured or directly tested by FxBrokersListing; and
• any other metric-specific condition required by the existing methodology to reproduce the observation.

Researchers must not silently transform one statistical concept into another. A minimum must not be recorded as an average, a range midpoint must not be recorded as a measured mean, and a single observation must not be represented as a historical average unless an existing metric rule explicitly permits that treatment.

Where a source value does not correspond to the measurement basis required by the existing metric, the researcher must document that mismatch and apply the evidence and missing-data treatment already prescribed by Sections 2.0, 2.1 and the relevant metric rule. This section does not create an alternative estimate or score.

Where two or more otherwise eligible sources of the same authority level provide candidate observations for the same input, the research record must document the basis for source selection using applicability to the reviewed configuration, conformity with the metric definition, verification date and any metric-specific evidence rule. A researcher may not choose between candidate observations according to which one produces the more favorable broker score.

Measurement metadata form part of the reproducibility record and must be retained with the scored observation.

2.2D Weight Rationale and Construct Register

The numerical weights in this methodology remain exactly as stated in their respective sections. In addition to the sensitivity testing required by Section 2.2A, FxBrokersListing must maintain a Weight Rationale and Construct Register for each methodology version so that the intended meaning of those existing weights can be audited.

The register must identify, for each category and subcategory:

• the dimension of broker performance, protection, functionality, accessibility, transparency or client experience it is intended to represent;
• the rationale for its stated weight;
• the principal underlying observations through which that dimension is measured;
• material overlap with other categories or subcategories, if any;
• the existing controls used to avoid unintended double counting; and
• the reason the dimension is included in the overall rating rather than presented only as descriptive information.

For individual metrics, the register may group closely related metrics where they serve the same clearly defined measurement purpose. The register should document whether each metric or metric group primarily measures cost, protection, accessibility, operational performance, breadth, transparency, functionality, direct task completion, contractual or regulatory status, or another explicitly defined construct.

The register is an audit document only. It does not independently change a category, subcategory or metric weight. Any numerical weight change remains a methodology-version change under Section 16 and must be applied consistently to the comparison set.

2.2E Calibration Stability and Drift Review

Before a new calibration snapshot replaces an existing frozen calibration snapshot, FxBrokersListing must perform a calibration stability review. This review documents the effect of the replacement calibration; it does not authorize any broker-specific adjustment or any deviation from the scoring rules in Section 2.2.

For every distribution-calibrated metric, the review should compare the existing and proposed calibration datasets and record, where available:

• sample size;
• median;
• 10th through 90th percentile cut points;
• material changes in the distribution of observations;
• the number and percentage of common observations moving between scoring bands;
• the median absolute change in metric score caused solely by recalibration; and
• the 95th-percentile absolute change in metric score caused solely by recalibration.

Where recalibration materially changes category or final broker ratings, the calibration record should identify whether the change resulted principally from changes in underlying broker observations, changes in the composition of eligible observations, factual data corrections, a methodology-version change, or another documented cause.

Calibration drift analysis is a governance and transparency control. A frozen calibration remains in force until replaced in accordance with Sections 2.2 and 16.

2.3 Closed Taxonomy and Classification Rules

Every scored categorical classification must use an enumerated definition in this methodology. Researchers may not award points for an unlisted “other,” “equivalent,” “established,” “high quality,” “meaningful” or similar category by editorial judgment. A new category can affect scores only after it is defined through a methodology-version change and applied to the full comparison set.

Descriptive labels such as leverage labels and instrument-range labels may use plain-language adjectives only when the underlying numeric boundaries are explicit. Those labels do not create additional score points.

For a scored closed-taxonomy checklist or count metric, an enumerated factual item is counted as absent only when its absence is confirmed. If the status of any enumerated factual item is Unverified and the metric does not state a specific item-level missing-data rule, the entire metric is Unverified under Section 2.1 rather than silently treating the unresolved item as zero. Disclosure checklists remain scoreable from verified disclosure/non-disclosure because the disclosure itself is the measured observable. This rule does not change any listed item, denominator, maximum score or weight.

2.3A Selection and Tie-Breaking Rules

Whenever this methodology specifies a selection rule among multiple accounts, promotions, legal entities, products or other candidate observations, all explicitly stated selection criteria in the relevant section must be applied in their stated order.

If two or more candidates remain tied after all explicitly stated criteria have been applied, a researcher may not choose between them using editorial judgment or by selecting whichever candidate produces the most favorable broker score.

Standard-account selection: where no account is identified as the default or Standard account and two or more qualifying commission-free retail accounts have the same lowest published opening minimum, use the account with the lowest verified minimum trade size. If they remain tied, use the account with fewer documented retail eligibility restrictions. If they remain indistinguishable under these conditions, use the account appearing first alphabetically by its official account name.

Raw/ECN/commission-account selection: apply the existing rule of lowest published account-opening minimum first and lower published EUR/USD round-turn all-in cost second. If accounts remain tied, use the account with the lowest verified minimum trade size. If they remain tied, use the account appearing first alphabetically by its official account name.

Promotion selection: when two or more qualifying promotions produce exactly the same Standardized Promotion Value under the $500 test, use the promotion with the higher Restriction Burden score. If still tied, use the promotion with the higher Terms Transparency score. If still tied, use the promotion whose official promotion name appears first alphabetically.

Other unresolved scoring selections: where a scoring rule produces a genuine tie not covered above, and choosing between the tied observations could change the numerical score, the research record must document the tie and apply a deterministic rule defined for that methodology version. An editor or researcher may not resolve the tie according to brand preference, commercial relationship or the score that would result.

2.4 Calibration Basis Register

2.5 Data Normalization Rules

• Monetary values stated in another currency are converted to USD using the verification-date spot exchange rate recorded in the research file.
• Processing times are converted to hours. Where a source gives a range, the published upper bound is used unless the metric explicitly states otherwise.
• Forex spreads are recorded in pips. Fractional-pip quotations are converted to standard pips.
• XAU/USD spreads are recorded in price points, where 0.01 price points equals one XAU/USD pip under this methodology.
• Percentage fees are measured against the transaction or notional amount defined by the relevant scoring rule.
• When a broker publishes different conditions by legal entity or jurisdiction, only the conditions applicable to the reviewed configuration are scored.

3. Trust & Safety — 20%

3.1 Applicable Regulation and Regulatory Safeguards — 45%

Regulator Classification

The regulator applicable to the legal entity that actually serves the trader is used.

FxBrokersListing classifies regulators using the following five verifiable regulatory safeguards:

1. Ongoing regulatory capital requirement
2. Mandatory segregation of retail client funds
3. Retail conduct-of-business rules
4. Public supervisory/enforcement powers
5. Statutory investor compensation or insolvency protection framework

If one or more safeguard statuses are Unverified and the unresolved status could change the applicable regulator group, the regulator-classification metric is Unverified under Section 2.1 rather than assigning a weaker group because evidence is incomplete.

Authorization and License Verification

One point is awarded for each verified condition:

Maximum score: 5.

3.2 Client-Fund Protections — 20%

Investor Compensation

The verified monetary coverage limit is still disclosed in the broker review, but it does not create arbitrary score bands. The score measures the legal status of the protection arrangement.

If applicability of a compensation arrangement cannot be determined from eligible evidence, the metric is Unverified under Section 2.1; inability to verify a scheme is not scored as confirmed absence.

Client-Fund Segregation

Negative Balance Protection

3.3 Legal Entity and Ownership Transparency — 15%

Legal-Entity Transparency

One point is awarded for each verified disclosure:

Ownership and Control Transparency

One point is awarded for each verified item:

Maximum: 5 points.

3.4 Operating and Regulatory Record — 10%

The regulatory-record score starts at 5.00. Verified regulatory events create deductions.

Each distinct regulator enforcement action is counted once. If one enforcement action contains multiple sanctions arising from the same underlying case, only the largest applicable base deduction from the table below is applied to that action. Separate enforcement actions with different official case or decision records are scored separately and their adjusted deductions are added together. The same underlying conduct is not double-counted merely because it is described in more than one official publication.

Penalty size by itself does not determine severity because the same nominal fine can represent different significance across firms and jurisdictions. When one enforcement case contains multiple event types, Section 3.4's single-case rule applies.

Recency Multiplier

The minimum possible score is 0.

3.5 Account and Security Protections — 10%

The first three metrics use the standard 5 / 2.5 / 0 feature rule.

Additional Security Safeguards

Qualifying controls include:

• Biometric login
• Withdrawal whitelist
• New-device verification
• Login alerts
• IP/session management
• Separate withdrawal confirmation
• Anti-phishing code
• Transaction PIN

Each of the eight enumerated controls counts once. The maximum score is 5. No unlisted control may be substituted without a methodology-version change.

4. Fees & Commissions — 15%

4.1 Standard-Account Trading Costs — 35%

Benchmark Account Selection and Spread Evidence

The standard-account benchmark uses the broker's named default or Standard retail account when one exists. If no account is identified as the default or Standard account, FxBrokersListing uses the commission-free retail account with the lowest published account-opening minimum that is available to the reviewed client jurisdiction.

If reliable evidence confirms that the reviewed broker configuration has neither a named default/Standard retail account nor any commission-free retail account that qualifies under the fallback rule above, the Standard-Account Trading Costs subcategory scores 0 as a confirmed absence of the account structure being measured. If a qualifying account exists but a required spread input cannot be verified, Section 2.1 applies instead.

For a fixed-spread account, the currently stated fixed spread is used. For a floating-spread account, a representative spread must be supported by one of the following sources, in order of preference:

1. an official broker average or representative spread for the reviewed account;
2. official broker historical spread statistics from which an average is directly reported;
3. an independently measured average or median from a third-party source that passes Section 2.0, published or updated within the previous 12 months and statistically defined by that source.

An advertised minimum, "from" spread or best-case spread is not treated as an average. If only a minimum can be verified for a floating-spread account, the representative-spread metric is marked Unverified and handled under Section 2.1 rather than converting the minimum into an estimated average.

EUR/USD Representative Spread

GBP/USD Representative Spread

XAU/USD Representative Spread

XAU/USD is normalized to price points.

1 XAU/USD pip = 0.01 price points.

S&P 500 CFD Representative Spread

If one benchmark instrument is genuinely not offered, that instrument is excluded and the remaining instrument weights are proportionally normalized.

4.2 Raw/ECN/Commission Account Costs — 25%

This benchmark uses the commission-based retail account with the lowest published account-opening minimum that is available to the reviewed client jurisdiction. If multiple qualifying accounts have the same opening minimum, the account with the lower published EUR/USD round-turn all-in cost is used. Commissions quoted in a currency other than USD are converted to USD using the verification-date spot exchange rate recorded in the research file.

If reliable evidence confirms that the reviewed broker configuration offers no qualifying commission-based/raw/ECN retail account, this subcategory scores 0 as confirmed absence. If such an account exists but a required cost input cannot be verified, the affected metric is Unverified under Section 2.1.

The representative raw spread used in the all-in cost calculation follows the same evidence hierarchy as Section 4.1: official average or representative spread first, official historical spread statistics second, and an independently measured average or median from a third-party source that passes Section 2.0 third. An advertised minimum or “from” spread is not substituted for a representative raw spread.

The resulting EUR/USD all-in cost uses the EUR/USD distribution-calibrated scoring rule above.

Round-Turn Commission per Standard Lot

4.3 Non-Trading Fees — 20%

Inactivity Fee

If no inactivity fee exists, this metric scores 5.0. If a fee exists, both inputs are scored under Section 2.2 and combined using the weights above. If one input cannot be verified, Section 2.1 applies.

Currency Conversion Fee

Other Recurring Fees

4.4 Swap and Financing Costs — 10%

Measurement Protocol

The benchmark uses EUR/USD and XAU/USD on the same account type used for the Standard-Account Trading Costs section. For each available benchmark instrument, FxBrokersListing calculates the annualized cost of the more expensive debit side (long or short). If only one side carries a debit, that debit side is used. If both sides carry a debit, the larger annualized debit is used.

When swap is quoted as money, points or pips, the published daily charges for one standard lot are converted to the account currency using the broker's contract specifications. One full trading week is calculated, including any published triple-swap day. The weekly debit is then annualized over 52 weeks and divided by the position notional value.

If the broker directly publishes an annualized financing percentage, that percentage is used without re-annualizing it. The final representative value is the median of the available EUR/USD and XAU/USD annualized debit-side costs. If only one of the two instruments is offered, the available instrument is used. If neither benchmark can be verified, the metric is Unverified under Section 2.1.

4.5 Fee Transparency — 10%

Each verified disclosure receives 0.5 points:

Maximum: 5 points.

5. Trading Platforms & Tools — 12%

5.1 Platform Availability and Variety — 10%

Distinct Platform Families — 60%

The score saturates at three platform families, so additional platform names do not keep increasing the rating without evidence of better trading functionality.

Desktop, web and mobile versions of the same platform family count once.

Named Third-Party Platform Choice — 40%

The metric measures availability of an externally portable third-party platform ecosystem; it does not rank one named platform as inherently better than another.

Only the named platform families above receive points in this metric. No open-ended “other established platform” points are awarded.

For TradingView, Section 5.1 measures availability of TradingView as a supported trading platform family. Section 5.5 separately measures direct broker-account connectivity for order execution through TradingView; chart access or branding alone does not satisfy the integration metric.

Maximum: 5 points.

5.2 Advanced Trading and Automation — 15%

Each metric uses the standard 5 / 2.5 / 0 feature rule. Hedging, scalping and news-trading permissions are scored under Section 14.1 rather than being double-counted in this category.

5.3 Core Trading Workflow and Functionality — 40%

Each task is scored 5 when completed successfully in direct testing, 2.5 when completion is possible only with a documented platform/account restriction, and 0 when the task cannot be completed. A feature advertised on a webpage does not receive task-completion credit unless the test succeeds or direct testing is impossible for a documented reason and equivalent official technical documentation verifies the function. When that documentation fallback is used, the same 5 / 2.5 / 0 conditions apply: 5 for verified unrestricted completion, 2.5 for verified restricted completion, and 0 only when non-function is confirmed.

Maximum: 5 points.

5.4 Mobile and Web Trading — 25%

Android and iOS use the standard feature rule.

Web Trading

Cross-Platform Consistency

Compare the following ten functions across all supported mobile/web/desktop versions:

1. Market orders
2. Pending orders
3. Stop-loss
4. Take-profit
5. Charts
6. Indicators
7. Alerts
8. Account management
9. Trade history
10. Watchlists

The score is continuous from 0 to 5 and uses the exact consistency percentage; no editorial bands are applied.

5.5 Integrations and Connectivity — 10%

Trading API Definition: For this metric, a Trading API qualifies only when it allows the reviewed client configuration to programmatically submit, modify, cancel or execute trading orders. A read-only API providing market data, account information, reporting or price feeds without programmatic order execution does not qualify as a Trading API. If trading functionality is available only to specified accounts, client types or eligibility conditions, the standard conditional-feature rule applies.

Each metric uses the standard feature rule.

6. Deposits & Withdrawals — 8%

In Sections 6.1–6.4, when a representative value is calculated across payment-method families and more than one eligible method or provider exists within the same family, the family-level value is the median of the verified method-level values in that family. The cross-family calculation then uses those family-level values. This prevents a family with many branded providers from receiving disproportionate influence. A representative median is scoreable only when the required value is verified for every eligible family included in that metric; a known eligible family with an unresolved required value makes the affected metric Unverified under Section 2.1 rather than being silently omitted from the median.

6.1 Withdrawal Processing and Accessibility — 35%

Withdrawal Processing Time

Where multiple payment-method families are available, the median of the published upper processing-time bounds is used.

Minimum Withdrawal

Where method-specific minimums differ, the representative minimum withdrawal is the median of the minimums for payment-method families available to the reviewed jurisdiction. A method with no stated minimum is treated as $0 only when the broker explicitly confirms that no minimum applies.

Withdrawal Accessibility

The score starts at 5. Deduct:

Minimum score: 0.

The $10,000-per-day condition is the predefined consumer-utility breakpoint for this accessibility rule: a lower daily cap can require a client withdrawing a five-figure balance to split the request across multiple days. It is an accessibility breakpoint, not a safety or broker-quality judgment.

6.2 Range of Payment Methods — 10%

Payment Method Families

Normalized families include:

• Bank transfer
• Cards
• E-wallets
• Regional/local bank methods
• Supported digital-asset channels
• Mobile/digital wallets

Payment-Method Family Classification Rules

The six payment-method families are mutually exclusive for scoring purposes. A specific payment method or provider may contribute to only one family, even when the provider could reasonably be described using more than one general payment term.

• Bank transfer: conventional domestic or international bank-account transfers or wire transfers that are not classified as a separately identified regional/local payment system.
• Cards: funding or withdrawal through payment-card networks using a debit, credit or prepaid card transaction.
• E-wallets: standalone online stored-value or electronic-money wallet services through which the client maintains or uses a wallet account. A service classified as a Mobile/Digital Wallet below is not also counted as an E-wallet.
• Regional/local bank methods: separately identified country- or region-specific bank-transfer, instant-bank-payment or local payment-rail systems through which the transaction is initiated from or settled through the client's bank account. A method classified here is not additionally counted as Bank Transfer.
• Supported digital-asset channels: funding or withdrawal through cryptocurrency, stablecoin or another supported blockchain-based digital-asset transfer.
• Mobile/digital wallets: device-based or tokenized wallet services in which payment is initiated through a mobile or digital wallet credential rather than through a standalone electronic-money wallet account.

When a provider supports more than one payment function, classification is based on the specific transaction method actually made available by the broker to the reviewed client configuration. The same transaction method cannot increase more than one payment-method-family count.

Only the six enumerated families count. Multiple providers within one family do not increase the score.

Only the six enumerated payment-method families above count. Multiple providers within the same family do not increase this score.

Regional/Local Payment Coverage

6.3 Deposit Processing and Accessibility — 25%

Deposit Processing Time

Where multiple payment-method families are available, the representative deposit processing time is the median of the published upper processing-time bounds for the families available to the reviewed jurisdiction.

Where payment-method minimums differ, the representative transaction-level minimum deposit is the median of the verified method-family minimums available to the reviewed jurisdiction. A method with no stated minimum is treated as $0 only when eligible evidence explicitly confirms that no transaction-level minimum applies.

Deposit Accessibility

One point is awarded for each verified condition:

6.4 Funding and Withdrawal Fees — 20%

A standard transaction value of $500 is used.

Deposit fees and withdrawal fees are scored separately using their original 40% and 60% weights. For each side, calculate the Effective Fee % for every eligible payment-method family using the $500 transaction, then use the median of the family-level Effective Fee % values as that side's representative observed value.

6.5 Funding Disclosure Completeness and Consistency — 10%

Funding Information Completeness

0.5 point for each disclosed item:

1. Deposit methods
2. Withdrawal methods
3. Deposit processing times
4. Withdrawal processing times
5. Deposit minimums
6. Withdrawal minimums
7. Deposit fees
8. Withdrawal fees
9. Supported currencies
10. Payment/KYC restrictions

Maximum: 5 points.

Official Source Consistency

Start from 5 points.

A material contradiction exists only when two current official sources for the same reviewed entity/account state mutually incompatible facts about a scored input and choosing one source rather than the other would change that metric's score, availability status or eligibility. Wording differences that do not change a scored input are not contradictions.

Minimum score: 0.

7. Tradable Instruments — 10%

7.1 Forex Market Coverage — 30%

7.2 Range of Asset Classes — 30%

Normalized classes:

• Forex
• Stocks/share CFDs
• Indices
• Commodities
• Cryptocurrencies
• ETFs
• Bonds
• Futures
• Options

Only the nine enumerated asset classes count. Each class counts once regardless of the number of instruments inside that class; instrument depth is measured separately in Section 7.3.

7.3 Depth and Number of Instruments — 25%

FxBrokersListing Instrument-Range Labels

These labels describe instrument quantity and do not independently change the broker's score. The words used in these labels are fixed display labels tied only to instrument count; they are not qualitative judgments about execution, suitability or broker quality.

7.4 Specialist Market Access — 15%

Only the four enumerated specialist-market classes above affect this metric. Each uses the standard feature rule.

8. Leverage Classification

Maximum leverage is presented as a descriptive classification. Higher leverage does not automatically increase the broker's overall rating. The terms Low, Moderate, High and Ultra High are fixed dataset labels only; they are not safety, quality or suitability judgments.

9. Account Opening — 5%

9.1 Range of Account Types — 10%

Two named accounts are counted separately only when at least one of the following verified differences applies: spread-only versus commission pricing; a different execution model; a maximum-leverage difference of at least 25%; an account-opening minimum at least 2× the other account; a minimum trade-size difference; access to a distinct instrument set; or a documented eligibility class such as retail versus professional/VIP. A different marketing name or a different platform alone does not create a separate account type.

Marketing-name variants and platform-only variants remain excluded. Because this is a pure breadth metric, its category influence is limited by Section 1.2.

The score saturates at four account structures. Additional account marketing variants do not increase the score.

9.2 Entry Accessibility and Minimum Deposit — 35%

This subcategory measures the minimum account-opening deposit. Minimum forex trade size is scored under Section 14.3 so that the same metric is not counted in both Account Opening and Trading Conditions & Risk Management.

Minimum Account Opening Deposit

9.3 Specialist Account Options — 15%

Each metric uses the standard feature rule.

9.4 Account Opening and Verification — 40%

Online Account Opening

KYC Processing Time

Onboarding Transparency

0.5 points is awarded for each disclosed item:

1. Required identity documents
2. Country restrictions
3. KYC steps
4. Expected verification time
5. Minimum opening deposit
6. Account types
7. Client legal entity
8. Available account currencies
9. Applicable fee information
10. Leverage or margin terms

Maximum: 5 points.

10. Customer Support — 7%

Customer Support is scored only from directly verifiable broker information and FxBrokersListing testing. Public comments, review sentiment, star ratings and review-platform company-response statistics are not used in this category.

10.1 Response Speed and Accessibility — 55%

Initial Human Response

Response time is measured from the timestamp of FxBrokersListing's support request to the first non-automated human reply. Where testing is performed more than once during the same review cycle, the median measured response time is used. Automated acknowledgements, queue messages and chatbot-only replies do not count as a human response.

Non-Response Observation Window: The observation window for a support request is the first 24 scheduled human-support hours following submission of the request. Only hours during which the broker states that the relevant human-support service is operating count toward the 24-hour observation window. If no non-automated human reply is received during those 24 scheduled support hours, the test is recorded as a confirmed non-response and the metric scores 0. Automated acknowledgements, queue notices and chatbot-only messages do not stop the observation window. If a technical failure prevents a valid request from being submitted or received, the result is Unverified rather than a confirmed non-response.

Live Support Accessibility

10.2 Contact Channels — 15%

Each channel family is counted once. Maximum score: 5.

10.3 Availability and Language Coverage — 30%

Support Hours

Convert the documented recurring human-support schedule to hours per seven-day week. For 24/7 use 168 hours; for 24/5 use 120 hours. Where different channels have different schedules, use the schedule of the broadest live-human channel available to ordinary clients in the reviewed jurisdiction.

Support Languages

Count only languages for which human support is explicitly documented or directly verified. Website translation alone does not count as support-language availability.

11. Market Research — 6%

Raw content volume cannot by itself dominate this category. Market and education volume subcategories total 55% and use distribution-calibrated scoring; 35% is reserved for freshness, defined curriculum coverage and accessibility, and 10% for practical tools.

11.1 Market Research and Analysis — 30%

Market Analysis/Outlooks

Count broker-produced market research items published during the previous 30 days. A qualifying item must contain market-specific analysis rather than only a headline, price table, calendar entry or promotion. For written content, the item must contain at least 250 words of analytical text after excluding navigation, disclaimers and repeated boilerplate. For audio/video content, it must contain at least 3 minutes of market-specific analysis. Automatically syndicated news and duplicated copies of the same item are excluded.

In-House Analyst/Insight Content

Count qualifying broker-produced analysis attributable through a byline, author page or official publication credit to an identifiable broker analyst, named research team or named internal research function during the previous 30 days. Generic unsigned news reposts and third-party syndicated content do not count.

Market Webinars and Events

Count market-focused events during the previous 90 days.

11.2 Educational Resources — 25%

Structured Lessons or Course Modules

Count unique lessons or modules that have a distinct learning objective or lesson page. Navigation pages, category pages, duplicate translations and the same lesson published in multiple formats are counted once.

Educational Webinars in Previous 90 Days

Educational Videos/Tutorials

Written Educational Guides

eBooks/Downloadable Guides

Glossary/Beginner Reference Entries

11.3 Practical Research Tools — 10%

Each metric uses the standard feature rule.

11.4 Freshness, Curriculum Coverage and Accessibility — 35%

Update Frequency

Measure the median number of calendar days between the five most recent qualifying research publications under the Market Analysis/Outlooks definition above. If only two to four qualifying publications exist, use all available intervals between those publications. If fewer than two qualifying publications exist, recurring update frequency is confirmed absent and the metric scores 0.

Curriculum Topic Coverage

Curriculum Topic Coverage is the metric labeled Depth and learning structure in the Section 11.4 subcategory table above.

One point is awarded for each topic area with at least one dedicated broker-produced lesson or guide of at least 250 words or 3 minutes:

1. Forex/CFD market mechanics and terminology
2. Order types, platform operation and trade execution
3. Technical analysis
4. Fundamental/macroeconomic analysis
5. Risk management, leverage and position sizing

Maximum: 5 points.

Accessibility

These are closed access states. Deposit size does not create an additional arbitrary score band.

If access cannot be determined from eligible evidence, the Accessibility metric is Unverified under Section 2.1 rather than scored 0.

12. Trader Experience Classifications

Trader experience classifications are separate from the overall FxBrokersListing numerical rating. They describe the type of trader for whom the broker's verified account structure, platforms, costs, tools and accessibility are most aligned.

These classifications do not add or subtract points from the broker's overall score.

For each profile, a criterion is counted as met only when it is verified as met. A profile is classified as met as soon as its verified met count reaches the stated threshold. A profile is classified as not met only when the verified met count plus all still-Unverified criteria is below the threshold. If neither conclusion is possible, that profile classification is withheld until sufficient data are verified. Unverified criteria are therefore never automatically counted as failed.

12.1 Beginner-Oriented

A broker is classified as Beginner-Oriented when at least 5 of the following 7 criteria are met:

In the criteria below, Platform usability score means the Section 5.3 Core Trading Workflow and Functionality subcategory score. Market Research and Education score means the Section 11 Market Research category score, which includes Educational Resources. These definitions identify the existing scores referenced by the original criteria; they do not change the 5-of-7 threshold or any criterion value.

1. Minimum opening deposit ≤ $200
2. Account opening is fully online
3. Platform usability score ≥ 3.5/5
4. Market Research and Education score ≥ 3.0/5
5. Standard-account trading-cost score ≥ 3.0/5
6. Copy trading or social trading is available
7. Mobile and web trading score ≥ 3.5/5

Beginner Criteria Met = Number of the 7 conditions satisfied

A broker qualifies as Beginner-Oriented when:

Beginner Criteria Met ≥ 5

12.2 Intermediate-Oriented

A broker is classified as Intermediate-Oriented when the verified-criteria rule above establishes that it does not meet the minimum threshold for either the Beginner-Oriented or Professional-Oriented classification.

Therefore:

Beginner Criteria Met < 5

and

Professional Criteria Met < 4

For the Intermediate-Oriented rule, the two inequalities above are applied only after each profile has been definitively classified as not met under the Unverified-criteria rule in Section 12. A raw verified-met count below the threshold is not sufficient when unresolved criteria could still cause that profile to qualify.

This classification represents brokers whose verified features and trading conditions do not strongly satisfy either of the two more specialized experience profiles.

Intermediate-Oriented is a classification rule only. It does not represent a lower or higher broker-quality score than Beginner-Oriented or Professional-Oriented.

12.3 Professional-Oriented

A broker is classified as Professional-Oriented when at least 4 of the following 6 criteria are met:

1. Raw/ECN/commission-based account is available
2. Trading API or FIX connectivity is available
3. Algorithmic/EA trading is available
4. At least 1,000 tradable instruments are available
5. Trading Platforms & Tools category score ≥ 4.0/5
6. EUR/USD all-in raw-account cost ≤ 1.0 pip

Professional Criteria Met = Number of the 6 conditions satisfied

A broker qualifies as Professional-Oriented when:

Professional Criteria Met ≥ 4

12.4 Brokers Meeting Both Profiles

The Beginner-Oriented and Professional-Oriented classifications measure different characteristics and are not assumed to be opposites.

A broker that satisfies both:

• at least 5 of the 7 Beginner-Oriented criteria; and
• at least 4 of the 6 Professional-Oriented criteria

may display both Beginner-Oriented and Professional-Oriented classifications.

Such a broker is not classified as Intermediate-Oriented.

12.5 Classification Logic

The classification rules are therefore:

• Beginner-Oriented: Beginner Criteria Met ≥ 5
• Intermediate-Oriented: Beginner Criteria Met < 5 and Professional Criteria Met < 4
• Professional-Oriented: Professional Criteria Met ≥ 4
• Beginner + Professional: Both respective thresholds are met

In the summary logic above, the Intermediate-Oriented “<” conditions mean that the corresponding profile is definitively not met under the Section 12 Unverified-criteria rule; they do not instruct researchers to count unknown criteria as failed.

These classifications are determined entirely from the predefined criteria above. No editor or researcher may manually assign an experience classification based on subjective judgment.

Trader experience classifications do not change the broker's overall FxBrokersListing numerical rating.

13. Bonus & Promotions — 2%

Bonus & Promotions measures the value, accessibility, restrictions and transparency of active retail promotions available to the reviewed client jurisdiction. Because promotional rules can be restricted or prohibited by law in some jurisdictions, the category is scored only where such promotions are legally permitted for the reviewed configuration.

If retail trading promotions are legally prohibited for the reviewed client relationship, this category is marked Not Applicable and Section 2.1 applies. If promotions are legally permitted but the broker has no active qualifying promotion, the category score is 0.

13.1 Promotion Availability and Eligibility — 20%

13.2 Standardized Promotion Value — 25%

FxBrokersListing measures the maximum verified promotional value that an eligible retail client can obtain from a standardized $500 qualifying deposit. Percentage bonuses are applied to $500 subject to the published cap. Fixed bonuses are counted at their verified amount. A no-deposit promotion may be counted when it is available to the reviewed client and does not require a deposit before the promotional credit is granted. Future rebates, loyalty points and contest prizes are excluded from this metric unless their value is guaranteed at the time the qualifying condition is met.

Where more than one active promotion qualifies, the promotion producing the highest standardized value is selected and the same promotion is used for Sections 13.3 and 13.4. This prevents the methodology from combining the value of one promotion with the easier terms of another.

13.3 Restriction Burden — 35%

One point is awarded for each verified condition below. The score therefore runs directly from 0 to 5 without editorial bands.

A condition receives a point only when the applicable promotion terms verify it. If the terms expressly impose the restriction, that item receives 0.

13.4 Terms Transparency — 20%

Each verified disclosure receives 0.5 points:

1. Eligibility by client jurisdiction
2. Eligible account types
3. Qualifying deposit or activity requirement
4. Bonus percentage or fixed amount
5. Maximum promotional amount or cap
6. Trading volume or turnover requirement, if any
7. Withdrawal treatment for deposited funds
8. Withdrawal treatment for profits
9. Expiry or duration rule
10. Full promotion terms are publicly accessible before participation

Maximum: 5 points.

14. Trading Conditions & Risk Management — 15%

Trading Conditions & Risk Management measures the rules that govern how positions may be opened, maintained and closed, together with the broker's disclosure of leverage, margin, execution and liquidation conditions. Higher maximum leverage does not automatically produce a higher score.

14.1 Strategy Permissions — 20%

Each metric uses the standard 5 / 2.5 / 0 feature rule. A strategy is treated as available only when it is permitted for the reviewed entity/account under the applicable trading terms.

14.2 Leverage and Margin Transparency — 25%

Each disclosure uses the standard feature rule: 5 when the applicable condition is directly and completely disclosed for the reviewed configuration, 2.5 when the disclosure is available only for specified instruments/accounts or is incomplete, and 0 when the prescribed source review confirms that the required disclosure is not provided. Inability to complete the prescribed source review is Unverified under Section 2.1.

The leverage labels in Section 8 remain descriptive only. The numerical amount of maximum leverage is not scored as inherently better or worse.

14.3 Position Sizing and Exposure Controls — 20%

The other three metrics use the standard feature rule.

14.4 Execution Conditions and Policy Disclosure — 20%

One point is awarded for each verified disclosure:

1. Execution model or execution arrangement is identified
2. Market execution and/or requote treatment is explained
3. Positive and negative slippage treatment is disclosed
4. Stop/limit order treatment during gaps or fast markets is disclosed
5. An applicable order-execution policy or equivalent execution document is accessible

Maximum: 5 points.

14.5 Margin and Liquidation Risk Controls — 15%

Each metric uses the standard feature rule. This subcategory measures the existence and clarity of risk-control rules rather than rewarding a particular leverage level.

15. Review Updates and Data Dates

Broker conditions may change after a review is published. Regulation, spreads, commissions, leverage, payment methods, instruments, account types and platform availability may all change.

FxBrokersListing therefore records the relevant verification date for material broker information and updates ratings when changes are verified.

16. Methodology Versioning

Material changes to the methodology receive a new methodology version or effective date.

Examples of material changes include:

• Category-weight changes
• Subcategory-weight changes
• New scoring thresholds
• Changes to missing-data treatment
• Changes to calculation formulas

Existing broker ratings displayed in the same comparison set should use the same methodology version. When a material methodology change is introduced, FxBrokersListing recalculates affected ratings before presenting old-version and new-version scores as directly comparable.

17. Quality Control and Reproducibility

Each scored broker record should retain enough information for the rating to be reproduced.

The research record should include:

• Observed value
• Source
• Verification date
• Relevant legal entity
• Metric score
• Applicable weight
• Any exclusion or not-applicable treatment
• Rating Coverage percentage and Evidence Coverage percentage for each affected subcategory/category
• Account type and client jurisdiction used for the observation
• Any currency conversion rate or normalization step used
• Calibration snapshot ID and sample size for every distribution-calibrated metric
• Performance percentile used to convert each distribution-calibrated value into a score
• Direct-test evidence for each task-completion metric where testing is feasible
• Category-benchmark snapshot date and methodology version used for each displayed broker-excluded category average
• Number of eligible comparison brokers included in each category benchmark after excluding the broker being viewed
• Unrounded broker-excluded category mean used to produce the displayed benchmark
• Calibration basis type for every metric: distribution, external anchor, direct formula or closed categorical rule
• Sensitivity-test result for every pure feature-count/breadth metric subject to Section 2.2A

The same verified data entered into the same methodology version should produce the same FxBrokersListing rating.

17.1 Independent Research Replication and Inter-Researcher Reliability

FxBrokersListing quality control must include periodic independent replication of scored broker research records. This control tests whether the existing methodology can be applied consistently by different researchers and does not create a separate score component.

A sample of broker reviews should be independently reviewed by a second researcher using the same methodology version and the same applicable evidence rules. Where practical, the second researcher should determine the underlying observations and classifications without being shown the first researcher's final metric scores.

The replication review should compare, where applicable:

• numerical observed values;
• categorical classifications;
• Verified Present, Confirmed Absent and Unverified determinations;
• legal-entity and jurisdiction selection;
• account-selection decisions;
• source eligibility;
• Not Applicable determinations;
• metric scores;
• subcategory calculations;
• category calculations; and
• final rating calculations.

The quality-control record should report agreement rates for categorical observations, material differences between independently collected numerical observations, classification disagreements, calculation discrepancies, source-selection discrepancies and the documented resolution of each material disagreement.

A disagreement must be resolved through the evidence hierarchy, metric definition, selection rules, calculation rules and other provisions of the applicable methodology version. It must not be resolved according to the broker score preferred by an editor or researcher.

Where replication identifies an incorrect observation or calculation, the correction must comply with the source-retention, versioning and calculation-control requirements of this methodology. Inter-researcher reliability results are quality-control evidence only and do not add or subtract points from any broker score.

17.2 Source Snapshot and Evidence Preservation

A reproducible research record must preserve sufficient evidence to establish not only what source was used but what the relevant source stated when the observation was verified.

For every material scored observation, the research record should retain, where technically and legally practical:

• source URL or official document identifier;
• publisher or issuing authority;
• page, section, table or document location where applicable;
• access or retrieval date;
• verification date;
• relevant source publication or update date where available;
• a saved copy, screenshot, archived version or other durable record of the relevant source content;
• the specific text, table entry, platform result or observed value supporting the scored input; and
• the researcher responsible for verification.

Where a source is dynamic, account-gated or capable of changing without a public revision history, preservation of the observation should include sufficient contemporaneous evidence to demonstrate what was observed on the verification date.

For direct testing, the retained evidence should identify the tested platform or interface, reviewed configuration, test date and result of the scored task. Where supported by the research system, stored evidence may also include a file checksum or equivalent integrity identifier so that later changes to an archived evidence record can be detected.

Source preservation does not alter the source hierarchy in Section 2. It exists to make application of that hierarchy auditable after publication.

17.3 Missing-Data and Coverage Bias Audit

FxBrokersListing must periodically review the distribution of Unverified, Insufficient Data and Not Applicable outcomes across the broker comparison dataset. This audit evaluates the operation of the existing missing-data rules and does not change the thresholds, normalization formulas or scoring treatment in Section 2.1.

The review should identify whether missing evidence is concentrated in particular categories, metrics, jurisdictions, legal-entity structures, broker types, source types or other material segments of the comparison universe.

The audit should report, at minimum:

• the percentage of observations classified as Unverified by metric;
• the percentage of subcategories classified as Insufficient Data;
• the percentage of categories classified as Insufficient Data;
• the distribution of Rating Coverage;
• the distribution of Evidence Coverage; and
• metrics for which missing evidence materially limits comparison coverage.

The audit must distinguish, where the research record permits, between confirmed absence, failure of disclosure, lack of eligible evidence, inability to complete direct testing, Not Applicable treatment and other documented causes of non-scoreable data.

Missing-data analysis is a methodology-governance diagnostic only. It does not permit researchers to infer missing observations, change an individual broker's score or alter the normalization rules in Section 2.1. If the audit indicates a systematic measurement problem, that issue must be considered during the next methodology review, and any resulting scoring-rule change must follow Section 16.

17.4 Calculation and Implementation Validation

The implementation used to calculate FxBrokersListing ratings must be tested separately from the underlying research process so that software, spreadsheet, database or formula implementation errors do not alter the methodology.

Before publication of a methodology version, and after any material change to the scoring implementation, automated or equivalent validation checks should confirm that:

1. all overall category weights total 100%;
2. all applicable subcategory weights total 100% within their parent category;
3. all applicable metric weights total the amount specified by their parent scoring group;
4. metric scores remain within the permitted 0–5 range;
5. missing-data normalization follows Section 2.1 exactly;
6. Not Applicable observations are excluded only where permitted by the existing methodology;
7. Rating Coverage is calculated from original applicable category weights;
8. Evidence Coverage is calculated from original overall-equivalent metric weights;
9. final ratings use unrounded intermediate values;
10. final published ratings are rounded only at the required publication stage;
11. distribution-calibrated metrics cannot be calculated without the required valid calibration snapshot;
12. percentile direction is correctly applied as higher-is-better or lower-is-better;
13. tie treatment follows the mid-rank formulas in Section 2.2;
14. confirmed absence is not substituted for Unverified information;
15. category benchmarks exclude the broker being viewed;
16. category benchmarks use only eligible same-version observations;
17. broker-excluded category averages use the required minimum comparison sample;
18. methodology-version rules are applied consistently across the comparison set; and
19. no manual broker-specific adjustment is introduced after formula calculation.

The calculation system should maintain reference test cases with known inputs and known expected outputs. Re-running those cases under an unchanged methodology version must reproduce the same metric, subcategory, category, coverage and final-rating results.

A change to software, spreadsheet formulas, database logic or other implementation infrastructure must not silently alter the methodology calculation. Implementation validation is a quality-control mechanism only and does not create an additional scoring rule.

17.5 Metric Dependency and Double-Counting Audit

FxBrokersListing must periodically review the methodology for unintended duplication of substantially identical underlying observations. This audit documents dependencies in the existing methodology and does not authorize ad hoc removal or reweighting of any metric.

The audit should identify cases where:

• the same factual feature contributes to more than one scored metric;
• one metric is mathematically derived from another scored metric;
• two metrics are driven predominantly by the same underlying observation;
• one operational characteristic receives repeated credit under different terminology; or
• a disclosure metric and a substantive-performance metric rely on the same source fact.

Statistical correlation between metrics does not by itself establish double counting. The review must determine whether apparently related metrics represent distinct constructs under the existing methodology or repeat substantially the same underlying characteristic.

Material dependencies and the methodology rationale for retaining them must be documented in the quality-control record. Any scoring-rule change arising from the audit must follow Section 16 and be applied consistently to the relevant comparison set.

17.6 Published Corrections and Research-Record Change Log

FxBrokersListing must distinguish between a correction to underlying broker data and a change to the methodology itself.

A data correction occurs when the methodology remains unchanged but a previously recorded factual input is demonstrated to be incorrect, outdated or applicable to the wrong legal entity, jurisdiction, account, instrument or observation period.

A methodology change occurs when a scoring definition, formula, threshold, classification, weight, evidence rule, missing-data rule or other scoring treatment changes and remains governed by Section 16.

For every material correction to a published broker rating, the research record should retain:

• the value previously recorded;
• the corrected value;
• the source supporting the correction;
• the correction date;
• the reason for the correction;
• the affected metric, subcategory, category and final rating where applicable; and
• confirmation of the methodology version applied to the recalculation.

Corrections must not be used as a mechanism for selectively introducing a new methodology rule. Where a methodology change is required, Section 16 applies.

18. Editorial Independence

Commercial relationships do not change broker scores.

Affiliate relationships, advertising agreements or other commercial arrangements do not alter:

• Category weights
• Metric weights
• Numerical thresholds
• Evidence standards
• Scoring formulas

19. Final FxBrokersListing Rating Formula

20. Calculation Controls

• No broker-specific manual score adjustment is permitted after the formula is calculated.
• Any correction to an observed input value must be supported by a retained source and verification date.
• Scores are calculated from unrounded intermediate values and rounded only at the final publication stage.
• A methodology-version change must not be applied selectively to individual brokers within the same comparison set.
• No distribution-calibrated metric may be scored without the frozen calibration snapshot required by Section 2.2.
• No researcher may substitute an unlisted category, subjective equivalence or ad hoc score for a closed-taxonomy rule.
• Pure feature-count breadth must respect the category influence cap in Section 1.2.
• Every displayed category average must exclude the broker being viewed, use only eligible same-version category scores, include at least 100 other brokers, and be calculated from unrounded category scores.
• Every distribution-calibration snapshot must have a reproducible Calibration Universe Manifest under Section 2.2B.
• Every scored continuous numerical observation must retain the measurement metadata required by Section 2.2C.
• No researcher may silently transform a source value from one statistical measurement basis into another through undocumented estimation.
• The Weight Rationale and Construct Register must be retained for each methodology version under Section 2.2D.
• Calibration stability and drift analysis must be completed before a replacement calibration snapshot becomes effective under Section 2.2E.
• Periodic independent research replication must be performed under Section 17.1.
• Material scored evidence must be preserved in a form sufficient for later audit under Section 17.2.
• Missing-data patterns must be reviewed under Section 17.3 without changing individual broker scores or the Section 2.1 missing-data rules.
• The calculation implementation must pass the validation controls in Section 17.4.
• Potential metric duplication must be reviewed under Section 17.5 without ad hoc broker-specific reweighting.
• Data corrections and methodology changes must be separately identified and recorded under Section 17.6.

21. How to Interpret the Rating

The overall score is a weighted summary of the broker data measured by this methodology. FxBrokersListing does not use the final rating as a substitute for the underlying category results.

Two brokers with similar overall ratings may have different characteristics. For example, one broker may receive more points from regulation and trading costs while another may receive more points from platform functionality and market coverage.

For this reason, broker reviews should display the overall rating together with the underlying category scores wherever practical.

Distribution-calibrated component scores should be interpreted as relative standing within the frozen calibration universe used by the applicable methodology version; fixed legal/contractual anchors and direct-formula metrics retain the absolute meanings stated in their individual rules.