Last updated: June 8, 2026
Each broker in the fxgnosis rating receives a score from 0 to 5 across seven independent factors. The final position is determined by a weighted average of these factors.
Five factors rest on objective data (Regulation, Trading Conditions, Payments, Reviews, Localization): licences, fees, the set of payment methods and platforms, public ratings, verifiable localization signals. Two factors, Reputation and Support, include an editorial score against a clear rubric (see below). The methodology is the same for all brokers regardless of any affiliate relationship; any update is dated below and applied to the entire rating simultaneously.
The score is an absolute rating on a strict scale, not a broker's position relative to others in the list. A five means a broker with no weak spots on any factor: a set of Tier-1 licences, service to local clients under a top-tier licence, best-in-class trading conditions and a spotless reputation. That combination is almost out of reach for brokers actually available in this market.
So even a reliable broker for the local market naturally stays in the 3–4 range, and comparing that score with the usual 4.8–4.9 on sites that grade on a lenient scale is misleading: there almost every broker lands in the top zone, whereas here a five matches a global benchmark rather than an average level. A low score reflects real limitations on verifiable factors, not the editors' attitude.
Composed of two parts.
Group-level (0–3 points): the broker's full licence portfolio.
Per-geo (0–2 points): the licence under which the broker actually serves local clients. Tier-1 → 2, Tier-2 → 1, Tier-3 → 0.5, no licence → 0.
Final score: sum of both parts, capped at 5. A broker with no licences legitimately gets 0 (a meaningful result, not missing data), so this factor always participates in the calculation.
Composed of two parts.
Age bonus (0–1) is calculated automatically: 20+ years → 1.0, 10 years → 0.5, 5 years → 0.25, under 3 years → close to 0.
Editorial score (0–4) is assigned based on monitoring of independent platforms (ForexPeaceArmy, BrokerScam, industry forums), regulatory actions, and fines:
Without an editorial score, the factor is dropped (excluded with weight redistribution).
Weighted sum of six sub-factors; a sub-factor with no data is dropped and its weight is redistributed to the others:
The score depends on which tiers of payment methods are present and which direction they support, not on the number of listed methods. Within a single tier, deposit and withdrawal are combined as "at least one method," so five cards of the same tier give no advantage over one.
Each present tier contributes:
contribution = tier weight × (0.4 for deposit + 0.6 for withdrawal)
Withdrawal carries more weight: being able to take money out matters more to a trader. Tier weights: Tier 2 = 3.5 (base), Tier 3 = 1.6 (premium), Tier 1 = 0.9 (weak reserve). The sum of contributions is capped at 5.
Payment method tiers for Mexico:
There is no separate penalty for missing local methods: Tier-3 is rewarded with its own weight, and its absence simply skips the premium.
Logarithmically weighted average across two sources, TradingView and Trustpilot:
(tvRating × log₁₀(tvReviews + 10) + tpRating × log₁₀(tpReviews + 10)) / (log₁₀(tvReviews + 10) + log₁₀(tpReviews + 10))
Log weighting prevents a broker with a few dozen enthusiastic reviews from outranking one with thousands of organic ones. If a broker is listed on only one source, only that source is used. If listed on neither, the factor is dropped and its weight is redistributed across the others.
Editorial score based on test live-chat enquiries with concrete MX questions (SPEI/OXXO payments, verification, Hacienda taxation):
Without a support score, the factor is dropped.
This factor measures only the localization that is not already captured by other factors: local payment methods are part of Payments, support language is part of Support, and the local regulator is part of the per-geo half of Regulation. What remains is a residual of four verifiable signals (weighted sum; a sub-signal with no data is dropped):
Each signal: present → 5, absent → 0, unknown → not counted. If no signal is set at all, the factor is dropped and its 8% weight is redistributed.
Missing data on a factor must not silently turn into either a penalty or a reward. If there is no data on a factor or a sub-signal, it does not get 0 and does not get 5: it is dropped from the calculation, and its weight is proportionally redistributed across the factors that do have data. The final score is therefore always based only on verifiable information. For example, if a broker's spread is not confirmed, the trade-cost sub-factor does not pull the score down to a token value; it simply does not participate.
Licences and licence numbers are verified directly against public regulator registers (FCA Register, CySEC public list, FSC Mauritius licensees, etc.). Trading conditions come from brokers' official websites and live client cabinets at the time of verification. Review ratings are taken from public TradingView and Trustpilot listings, unmodified. Localization signals (site and education language, account currency, local contact) are checked by editors against the website and the client cabinet. Editorial scores for reputation and support are assigned by the fxgnosis team based on monitoring of independent platforms and test enquiries.
A full recalculation runs automatically on the last day of each month: all seven factor values are recomputed and positions in the top are refreshed. Off-cycle recalculations are triggered by urgent events (licence suspension by a regulator, mass confirmed complaints); such cases are noted in the broker's card.