Methodology
How we rate services
Every factual data point is attributed to a source with a verification date. No ranking position is for sale.
1. How we assign the 0–5 rating
Each service receives a score from 0 to 5 calculated across five weighted factors. The final score is a weighted average of the five criteria, assessed using verified public sources.
| Factor | Weight |
|---|---|
| Total costs (fees, monthly, FX, ATM, issuance) | 35% |
| Security & regulatory licensing | 25% |
| Operator transparency | 15% |
| Geographic coverage | 15% |
| User experience (UX) | 10% |
The "Sovereignty" lens does not change the data: it changes the yardstick. The score is computed automatically from the factual fields already published (self-custody, KYC, audits, proof-of-reserves, on-chain screening…) with different weights. The same service can earn two opposite ratings: a regulated, custodial exchange, great for consumer protection, drops a tier for those seeking privacy and self-custody — and vice versa.
| Factor | Weight |
|---|---|
| Privacy & anonymity (KYC, data, Tor, no-logs) | 30% |
| Fund control (self-custody, no freezing) | 20% |
| Censorship resistance (permissionless, self-host) | 20% |
| Trustless / auditability (open source, audits, PoR) | 20% |
| Costs | 10% |
Same data, opposite reading
| Fact | Consumer | Sovereignty |
|---|---|---|
| Full KYC | AML-compliant, more protection | Mandatory identity, zero pseudonymity |
| Regulation (MiCA…) | Oversight and legal recourse | Subject to freezes and warrants |
| Custodial | Assisted account recovery | You don't control your keys |
| On-chain screening | Reduces illicit funds | Can block "risky" funds |
This is not a wink at illegality: it is a tool to analyze the real purposes these services pursue. We reward those who operate at the edge — no-KYC, self-custody, censorship resistance — but the boundary of legality stays the line. Exclusion criteria are identical across both lenses: scams, exit scams and services with no legal entity are delisted in both.
2. What the S/A/B/C/D tiers mean
Excellence on both costs AND security. Top-tier licensing, zero hacks, competitive fees, maximum transparency.
Excellent with minor trade-offs. Reliable across all criteria, with one or two acceptable limitations.
Good with clear limitations. Suitable for most users, but with aspects to monitor (fees, regulation, track record).
Acceptable only for specific use cases. Use only when no Tier A/B alternative meets your specific need.
Not recommended except when no alternative exists. Significant verified risks: uncovered hacks, regulatory enforcement, opaque governance.
In the Sovereignty lens the same S→D scale is recomputed on the five sovereignty factors: the tier reflects autonomy and privacy, not costs and licensing. Tier F stays reserved for closed services.
3. Exclusion criteria
- Verified fraud or exit scams documented by regulatory or judicial sources.
- Absence of an identifiable legal entity in a recognised jurisdiction.
- Terms that cannot be publicly verified (no website, no documentation, no official contact).
- Usage volume below the minimum relevance threshold (<$10K/24h for exchanges).
Historically closed or failed services are retained in the catalog with Tier F for documentary purposes only.
4. Update frequency
Systematic update of fees, cashback, geographic availability and regulatory status for all services.
Triggered by provider-announced changes, hacks, failures, regulatory enforcement or significant fee changes.
Automatically shown on any service card whose data was last verified more than 60 days ago.
5. Rating vs. net cost calculator
The rating measures a service's overall quality: security, transparency, coverage and average costs. The net cost calculator measures what that service costs or earns you based on your specific spending profile. The two values may diverge — and that's perfectly fine. A Tier A service might be suboptimal for someone who rarely spends abroad; a Tier B service might be the best choice for a very specific spending profile. Use both together for an informed decision.
6. Reputation: why it sits outside the rating
The registry measures what official documents say: fees, terms, licences. It does not measure what happened to the people who used the service. The “Reputation” block fills that gap, but it feeds neither the 0–5 rating nor the Sovereignty lens: these are different kinds of data, and mixing them would devalue the former.
There are three levels, in decreasing order of solidity — and they deliberately look different on the page:
- Incidents & regulatory actions — hacks, withdrawal freezes, insolvencies, sanctions. Dated facts, each with its source: the same evidentiary standard as an official fee schedule. We also record the outcome: a hack with full reimbursement does not weigh like one without. When we state “no known incident” we show the date we searched — an empty list is never an unsearched list.
- Public reviews — Trustpilot, App Store and Google Play scores always shown with the NUMBER of reviews, because 4.8 out of 12,000 and 4.8 out of 7 are not the same thing. The average we show is weighted by volume and ignores platforms with fewer than 10 reviews. It is not our rating: it is those platforms’ average, recomputable by anyone.
- Community signal — what people say on X, Reddit and forums. It is the weak link and we treat it as such: never a decimal number, only a coarse band (from “negative” to “positive”) with the sample size and the time window. Below 12 mentions we declare no band at all. The service’s own official accounts are excluded from the count.
Online reviews can be manipulated: in financial services there are concrete incentives to inflate or sink them. That is why the community signal never yields a score, and always comes with real quotes linked to their original source, with dates. We do not ask you to trust our summary: we give you the link to check.
Every signal carries its verification date. Past 8 months it is flagged “re-check due”: a stale reputation datum is noise, not information.