Ratings Methodology

The principle

A rating is only useful if it is consistent and checkable. Everything a score depends on is published: the criteria, what each one assesses, the number given for each, and the arithmetic that turns them into the figure at the top of the review. If you disagree with a score, you can see exactly which criterion produced it.

The criteria are set per category

There is no single house scorecard. A hardware wallet and a stablecoin are not judged on the same things, so each category has its own set of four or five criteria, chosen for what can actually be evidenced in that category. The full set — every criterion and a paragraph on what it assesses — is printed on the ranking page for that category, above the reviews it applies to.

Examples of how far they differ: hardware wallets are scored on price, secure element, open source, air-gap and verification, and asset coverage. Stablecoins are scored on reserve transparency, regulatory oversight, redemption rights, custody and segregation, and backing model clarity. Instant exchangers are scored on fee transparency, rate options, custody and KYC, refund policy, and whether the operating entity is named at all.

How the score is worked out

• Each criterion is scored 0–10 for every service in the category.
• All criteria in a category carry equal weight. There are no hidden weightings, and no criterion counts for more than another.
• The overall score is the plain average of those numbers, rounded to one decimal place. Each review prints the individual scores and the arithmetic, so you can reproduce the result yourself.
• Rankings are ordered by that overall score, highest first. Where two services finish level, they are listed alphabetically. A tie is a tie — we do not break it silently to manufacture a winner.

What we deliberately do not score

A rubric is only worth having if every criterion is evidenced the same way, so anything we could not verify to the same standard is left out rather than guessed at. That is why the exchange ranking scores published trading costs and not custody or licensing, and why the stablecoin ranking scores documented reserve and licensing facts and not peg history or market depth. Where a category leaves something important unscored, the ranking page says so and explains why.

Where the facts come from

Scores are built from published primary sources: the provider's own pricing, documentation and specification pages, licence files in its code repository, published audit reports, and regulatory registers. Every review lists the exact pages used and the date each was read, so any figure can be traced and re-checked.

Each review states its basis. Reviews marked primary sources are desk research and we do not claim otherwise; reviews marked hands-on were produced by someone who used the product. We do not open accounts, place test trades or buy the products in order to score them, and we do not score criteria our stated basis cannot evidence — which is why the rubrics measure what a provider documents rather than how an interface felt to us.

Where a provider publishes nothing on a criterion, that absence is scored as the absence it is. It is never filled in with an assumption.

Providers change prices and terms without notice. If a figure no longer matches its source, tell us and we will correct it and re-date the review.

Re-checking

Every review carries the date it was written and the date its sources were last read, and every ranking page carries a "last checked" date. Those dates are the honest record of how current a page is. We re-check when a provider makes a material change we become aware of, and we do not claim a fixed re-check cycle we cannot guarantee — trust the date on the page rather than a promise about how often it moves.

Independence

Scores are editorial and reproducible: every overall score is the plain average of the per-criterion scores shown on the page, so you can recalculate it yourself. This is a non-commercial project with no affiliate links, no advertising, and no paid placements — nobody can buy a score or a rank. See our Editorial Policy.