Methodology
Every service is scored 0 to 10 on six axes. The overall figure is a fixed weighted mean of those six, computed by the site generator from a single table of weights. No entry gets a bespoke weighting, a bonus, or a manual override.
The axes
| Axis | Weight | What the number means |
|---|---|---|
| KYC exposure | 20% | Can this service ever demand your identity? A 10 means no account, no email and no documented case of a verification request. A 0 means mandatory verification before you can transact at all. |
| Fees | 15% | The real cost of moving value through, spread included. A service that hides its fee inside the rate is scored on the effective markup, not on what it advertises. |
| Privacy | 20% | Onion service, working no-JavaScript path, no account, no logging claims, and whether the trade ends up permanently public on-chain. |
| Custody | 15% | Who holds your funds and for how long. On-chain and multisig escrow score highest, custodial-in-flight swappers sit in the middle, full custodial exchanges score lowest. |
| Track record | 20% | Years operating, breaches, seizures, and the documented pattern of freezing user funds. New services score low here by construction; that is not an accusation, it is missing evidence. |
| Coverage | 10% | Assets and chains supported. Weighted least because a narrow service that never freezes funds beats a broad one that does. |
| Agent access | not counted | Whether an autonomous agent holding nothing but a funded wallet can discover, price and complete a swap unattended. Reported separately and deliberately excluded from the total: it measures capability, not whether the service is safe to use. |
The default view
The table opens on the overall ranking, the weighted combination of all six axes. Rank by a single axis instead whenever one of them is the whole question for you.
Watch one trap in the no-KYC view in particular: on-chain DEXs score a perfect 10 there, because there is no operator to demand anything, while scoring 3 on privacy because every trade is permanently public and linkable to your address. Permissionless is not the same as private. If privacy is what you actually want, rank by that axis.
Why these weights
KYC exposure and track record carry the most weight, at 20% each, because they are the two axes that determine whether you get your money back at all. Privacy is also 20%: for most people reading a page like this, privacy is the reason they are not simply using a bank. Fees and custody sit at 15%, and coverage is deliberately the least important at 10%, because a service supporting eight assets that has never frozen a transfer is more useful than one supporting three thousand that sometimes does.
These weights are a defensible default, not an objective truth. If your priorities differ, sort by the single axis you care about rather than the composite. That is what the sort controls are for.
What we do not do
- No paid placement. No service can pay to be listed, ranked higher, or removed.
- No affiliate links. Every outbound link is plain, with no referral parameter. This costs us revenue and is the point.
- No score inflation for the operator's own service. This site is published by the team behind Omenswap, which is not currently in the table. Should it be added, it gets the same rubric as everything else and a disclosure on its own row.
- No aggregate user star ratings. They are trivially brigaded in this sector, in both directions.
- The rubric is enforced by the build, not by good intentions. The site fails to compile if any entry scores outside the scale described above: a full custodial exchange cannot be scored above 4 on custody, a service that mandates verification cannot score above 0 on KYC, an on-chain DEX cannot score above 6 on privacy, and a service younger than four years cannot claim a track record above 7. Those bounds apply to every row, including any row we would benefit from.
Where the facts come from
Fee structures, KYC policies and custody models are taken from each service's own published terms where they exist. Incident history, seizures and freeze patterns are taken from reporting and from the documented experience record in privacy communities. Each entry records the date its facts were last checked, and that date is shown on the service page.
Two honest limitations. First, marketing copy in this sector is unusually unreliable: several services on this list publish their own "best no-KYC exchange" listicles ranking themselves first, and those pages are the top search results. We treat a service's self-published rankings as advertising. Second, a "no KYC" policy is a claim about a private risk engine that nobody outside the company can inspect. A high KYC score means no documented pattern of verification demands, which is weaker than a guarantee and should be read that way.
Status labels
- Active — operating, with no pattern of unresolved complaints.
- Caution — operating, but with a documented and recurring failure mode: surprise verification, frozen funds, stuck settlements, or a serious past breach.
- Defunct — no longer exists. Scored zero and excluded from every ranking, kept on the list only so that stale guides recommending it can be checked.
How to read the table
A high total is not a recommendation. The axes pull against each other. A custodial exchange can score brilliantly on fees and record while scoring zero on KYC. An on-chain DEX scores perfectly on custody and terribly on privacy, because every trade is permanently public. Decide which axis is load-bearing for you, then rank by that one.
"No KYC" usually means "no KYC yet." Most instant swappers advertise no verification and then run a risk engine that escalates on size, source address taint, or destination. The services scored 8 or above on the KYC axis are those with no documented pattern of doing this. That is an absence of evidence, not a guarantee.
Custodial-in-flight is still custodial. Almost every instant swapper takes your coins, holds them for a few minutes, and sends different ones back. During that window they can freeze, demand documents, or be seized. Shorter exposure than an exchange account, but not the same thing as non-custodial, whatever the marketing says.
Test small, then go, is bad advice. The failure mode behind nearly every negative report in this table is identical: a small test swap clears in minutes, so the user sends a large one, and that is the one that gets held. A successful test tells you the service is online. It tells you nothing about the threshold at which its risk engine fires.
Corrections
If an entry is wrong, it should be fixed. The scores and facts live in a single services.json file that the generator reads; a correction is a change to that file with a source, not a negotiation.