How we verify

Every fee here is the standard tier: what an account pays on its first trade, holding no exchange token, having issued no voucher, and having traded no volume.

That choice costs us the flattering numbers. Binance is listed at 0.10% rather than the BNB-discounted 0.075%, Gate at 0.20% rather than the 0.09% you reach by holding GT, Bithumb at 0.25% rather than the 0.04% its coupon gives. Quoting the best obtainable rate is the norm in this category, and it makes every comparison meaningless, because the conditions attached differ per venue and per reader.

Where a lower rate is reachable we record it separately, with what reaching it costs you: holding a token, issuing a voucher, or renewing a coupon that expires. Both numbers appear on the page. Only the standard tier is used for ranking.

Each exchange carries the date its terms were last reconciled against that exchange's own published schedule. Not the date the page was written, and not the date it was last touched.

A verification date with nothing behind it is worse than no date at all, because it looks like evidence. So the record is public. The changelog lists what moved and when, including the occasions we had a figure wrong.

In order of preference: the exchange's own API, then its published fee schedule, then its support documentation. Third-party directories are used only when nothing above can be retrieved, and the page says so when that happens.

Gopax is the worked example. Its fee page renders in the browser and returns nothing to a fetch, so the rate sat on third-party sources until we read it from the exchange's own trading-pairs endpoint instead. All 115 of its won pairs report the same figure. An API response is better evidence than a marketing page, and anyone can repeat the check.

Tables sort by what you pay. The commission we receive is recorded in the dataset so we can judge which exchanges justify more writing, and it is never a sort input on any page.

The referral codes page ranks by the cost left after the code applies, not by the size of the discount. Sorting on the discount would push the most expensive venues to the top, because a fixed percentage off a higher base is a bigger number. A code that saves you more while costing you more is not the better code.

No invented offers. Where we cannot confirm that an exchange runs a public programme, the page says so and the signup button is disabled rather than sending you into a flow that goes nowhere.

No private cashback. Several exchanges prohibit affiliates from sharing commission with the people they refer, and void the commission for it. Every discount described here is one the exchange itself applies to your account, where you can check it.

No exchange trademark in our domain, and no paid search on exchange brand keywords. Both breach the affiliate terms we operate under.

The dataset behind every page is published as JSON with the verification dates attached, free to use with attribution. If a figure here disagrees with the exchange, the exchange is right and we want to hear about it.