How we research and rate brokers

Reviews, comparisons, rankings, calculators and our AI assistant all read the same published research. That means every score on Brokeli can be traced back to sourced, dated figures and to the rules on this page.

01What we score

Methodology v2 is our rating profile for forex and CFD brokers. It turns a broker’s published research into four category scores — safety, fees, trading conditions and platforms — and combines them into one overall score from 0 to 10. The star rating you see on cards and reviews is derived from that overall score; it is not a separate opinion.

A score describes a broker’s published conditions. It is not a prediction of your results and not a recommendation to trade. Whether a broker suits you also depends on your country, the products you want and how you trade.

02Categories and weights

Each category has a fixed weight. The weights add up to 100% and are stored with the methodology version, so they cannot change without a new version being published.

Safety carries the largest weight on purpose: low costs are worth little if your money is not held by a properly regulated entity.

03How a metric becomes a 0–10 score

Raw figures come in different units: pips, deposit amounts, regulator tiers and yes/no answers. Before they can be combined, each one is mapped to a 0–10 score through a declared scale. Scales are stored as plain data in the methodology, never as code, and come in three types.

• Points scale: a set of anchor points joined by straight lines. A value between two anchors gets a proportional score; values beyond the first or last anchor keep that anchor’s score.
• Linear scale: an input range mapped evenly onto 0–10.
• Yes/no scale: yes scores 10, no scores 0.

Commission is converted to pips at $10 per pip per standard lot and treated as a round-turn charge, so a $3.50 commission adds 0.35 pips. Minimum deposits are compared in the account currency: USD, EUR and GBP amounts are treated as roughly equal and are not converted at live exchange rates.

Inside a category, metric scores are averaged by their weights. The overall score is the weighted average of the four category scores. Here is the calculation for an illustrative broker with sample figures:

1. 1
   Fees7.5

   A 0.7 pip typical EUR/USD spread plus $3.50 commission is 1.05 pips all-in. That sits halfway between the 0.9 (8) and 1.2 (7) anchors, so fees score 7.5.

2. 2
   Safety8.6

   Regulation tier 4 scores 8 and negative balance protection scores 10. Weighted 70/30: 0.7 × 8 + 0.3 × 10 = 8.6.

3. 3
   Trading conditions8.5

   A minimum deposit of 100 maps directly to the 100 anchor, so conditions score 8.5.

4. 4
   Platforms10

   Four platforms (MT4, MT5, cTrader and web) reach the top of the scale.

5. 5
   Overall8.46

   0.30 × 7.5 + 0.35 × 8.6 + 0.20 × 8.5 + 0.15 × 10 = 8.46 out of 10, which is 4.2 stars.

04Stars: half the overall score

The star rating is the overall score divided by two and rounded to one decimal place, with halves rounded up. An overall score of 8.4 is 4.2 stars and 5.1 is 2.6 stars. Scores are held between 0 and 10, so stars always run from 0 to 5. There are no editorial star adjustments and no bonus stars.

05Data sources and verification

Analysts record every figure as an observation: the value, its currency or unit, a link to the source and the quoted text, the date it was checked, and the legal entity, country, account type or platform it applies to. Our main sources are:

• The broker’s official website, fee schedules, account terms and legal documents
• Public registers of financial regulators, used to confirm licences, licence numbers and which legal entity serves each country
• Regulator warning lists, which also feed our scam research

1. 1
   Draft

   An analyst records the observation with its source. Drafts never appear on public pages, in our API or in AI answers.

2. 2
   Submitted

   The analyst submits the observation for review.

3. 3
   Approved or rejected

   A fact-checker compares it with the source. Rejected observations go back to the analyst; approved ones wait for publication.

4. 4
   Published

   Approved observations are frozen into the broker’s next research snapshot.

A broker’s research cannot be published until it passes our publication gates: at least one legal entity with a verified, active licence; verified availability for our priority countries; and an approved, sourced and dated observation for every required metric. Required metrics are the regulation summary and tier and the minimum deposit, plus negative balance protection, typical EUR/USD spread and commission per lot for forex and CFD brokers.

06Published research snapshots

A research snapshot is a versioned, read-only record of a broker’s approved data. Publishing is atomic: the new version goes live in a single step, and the previous version is marked as outdated but kept, so the history stays auditable.

• Reviews, comparisons, rankings, calculators, Find My Broker and the AI assistant all read the same published snapshot.
• Ratings are calculated from the snapshot under the active methodology and stored with both version numbers.
• A change to weights or scales is released as a new methodology version. Earlier versions are archived, not edited.
• Every approval, publication and rating is written to an audit log.
• Readers who set an alert on a broker are notified when a new snapshot for it goes live.

07What never affects scores

Brokeli may earn a commission when you open an account through our links. That relationship has no path into the rating. None of the following can change a score:

• Affiliate commissions, offers, or whether we have an agreement with a broker at all
• Advertising, sponsorship or labelled partner placements
• User reviews and forum posts, which are moderated and shown separately
• Scores, awards or claims supplied by the broker itself — the overall score is always recalculated from the weights
• The country you browse from, except to decide whether a broker is available to you

Rankings are built from the same published ratings. A country ranking only includes brokers recorded as available and accepting new clients in that country, and Find My Broker never recommends a broker that is unavailable where you live.

08How often we update

• Each metric has its own refresh interval: 30 days unless the metric defines another. Observations older than their interval are flagged as stale automatically every day.
• Ratings are recalculated daily from published snapshots, and rankings are refreshed every hour.
• We re-research a broker sooner when it changes fees, account terms, availability or regulation, or when a reader sends a sourced correction.
• Every review shows the date of its research snapshot, so you can see how current the figures are.

09Corrections

If a published figure is wrong, tell us. Use the contact form with the topic “Research correction” and include the broker, the country, the figure in question and a link to an official source. We check the correction through the same workflow as any other observation and, if it holds, publish a new snapshot. The previous version stays in the history.

10Methodology v2 at a glance