Research Methodology

What we measure

We score the quality and completeness of publicly available documentation. We do not score profitability, popularity, marketing appeal, or the subjective feel of an interface, because none of those can be verified by a reader working from public material.

This is a deliberate limitation. A high documentation score means a provider commits its terms to writing clearly. It is not a prediction of outcomes and not a recommendation.

The six weighted categories

• Disclosure clarity (20%): entity identification, permissions, risk language and how easily each is found.
• Cost transparency (20%): whether the full cost stack can be reconstructed from published documents.
• Execution documentation (20%): routing model, slippage, gap handling and order-holding location.
• Custody language (15%): segregation, custodians, insolvency treatment and compensation eligibility.
• Education quality (15%): proportion of transferable market education versus product onboarding.
• Support process (10%): documented channels, response commitments and external escalation routes.

Scoring scale

• 5 — Complete, dated, specific and easy to locate; worked examples provided.
• 4 — Complete and specific, but requires assembly across documents.
• 3 — Present and broadly adequate; general language in stress scenarios.
• 2 — Partial; material questions unanswered in public documents.
• 1 — Absent, contradictory, or contradicted by a public register.

Verification steps

• Locate the legal entity name and registration number in the provider's own documents.
• Search the relevant regulator's public register by number, not by name.
• Compare permitted activities against advertised services.
• Reconstruct the cost stack for a standard scenario at two holding periods.
• Read the order-execution policy for routing, slippage and gap language.
• Record which questions could not be answered from public material.

Limits we state openly

We cannot audit internal systems, verify execution statistics we did not observe, or confirm private complaint outcomes. Where a claim rests on evidence we could not obtain, we mark it unverified rather than scoring it favourably or unfavourably.

Documentation also changes. A score describes what was published on the date of review, which is why every page — including the Global Reserve review — displays that date.

Reproducibility

Weights are published before application and are not adjusted per platform. Any reader following the same steps against the same public documents should reach broadly the same category scores, and where they do not, the disagreement should be traceable to a specific document rather than to taste.