Methodology: How we source & update fees
FeeEdge compares trading fees across 33 major crypto exchanges. Accuracy is the whole point, so here's exactly where our numbers come from and how often they're refreshed.
What we track
For each exchange we track the entry-tier (regular / VIP-0) maker and taker fee for both perpetual futures and spot. The calculator then adjusts these for your 30-day volume tier, your maker/taker mix, native-token discounts, and, on perps, an estimate of funding costs. Trading-fee amounts are commissions only: monthly volume times your blended maker and taker rates. The calculator does not multiply those commissions by an asset-liquidity factor. Every figure on the site is an estimate to help you compare, not financial advice; always confirm the exact rate on the exchange before trading. Venues that publish a 0% maker and 0% taker book sit in their own band above the numbered ranking.
Where the rates come from
Each exchange/market uses the most reliable source available, in this order:
- Live exchange API. Where an exchange exposes a clean public fee endpoint, we pull it automatically. Kraken futures are fetched this way. Kraken spot volume:0 stays the curated 0.25% maker / 0.40% taker book; the official spot volume table is not applied.
- Automated scraping. For venues without a usable API, we scrape their published fee page through Bright Data's infrastructure, with a verified per-exchange parser and strict sanity checks (a value that can't be confidently parsed is rejected rather than shown). BingX futures is sourced this way today, and we add venues as reliable parsers are verified.
- Hand-verified rates. For the rest, we use published rates that we verify by hand and re-check on a recurring schedule. Each exchange carries the date we last confirmed its book.
How often it updates
An automated job refreshes every exchange daily. Live-API and scraped venues update from their real source on each run. Hand-verified published books carry the date we last confirmed them.
The "Updated" date on a ranking card follows that venue's verification stamp for published books. A live or scrape fetch only replaces it when the fetch is newer than that venue's stamp. An older fetcher write does not make a published book look stale. Exchange and compare pages use the same rule. Each calculator row shows that date and time so you can see how current a given rate is.
Who reviews the data
Pages carrying the FeeEdge Research byline are maintained by the independent operator described on the About page. Automated checks can detect and reject suspicious values, but a person confirms fee changes before they appear in the public fee-change log. Exchange and comparison pages link to official fee schedules when an official public page is available.
What we don't capture perfectly
Funding on perps is a signed cashflow for an explicit long or short hypothesis, not an absolute cost. Collection is BTC perpetual only, so ETH-only stays fees-only, and BTC plus Other or any mixed book is a labeled BTC reference, not a full-instrument match. A missing live quote is not treated as zero and is not replaced by the 0.01%/8h hypothesis when ranking venues. That 0.01%/8h figure is a labeled scenario hypothesis only. Source intervals stay on the quote (hourly on Hyperliquid, 4h on edgeX); 8h is the comparative unit. A momentary credit is not a guaranteed yearly saving. Withdrawal fees are shown per asset and network, fetched daily from the venues that publish them on a public endpoint and hand-verified for the rest. Spreads and slippage are the costs that genuinely can't be pinned to a single number, since they move with the order book, so we surface them as context rather than a fixed figure. MEXC's published UI perps base is 0% maker / 0.02% taker as of 31 Aug 2026, but that pair is not a universal book: rates can vary by region and event, and API futures use a separate 0.06%/0.08% schedule. When in doubt, the exchange's own fee page is the source of truth.
What the monthly total includes
The calculator total is trading-fee commissions. On perpetual futures it also includes signed BTC funding when every destination venue has a live BTC quote and the profile can use that quote. Withdrawals stay out of the total: they need frequency, asset, and network. Native-token discounts apply only when you turn on a quoted pay-in-token rate. A missing funding quote is shown as a labeled 0.01%/8h hypothesis and is not used to rank venues.
How funding is labeled
Collection is BTC perpetual only. BTC-only profiles are a BTC match. BTC plus Other, or any mixed book, is a BTC reference: an approximation, not those instruments' own rates. ETH-only or no instrument stays fees only. Long and short are exposure hypotheses, not a mixed-book net. Historical 30-day averages are a scenario built from stored settlements where we have them. Time-weighted snapshots are labeled when a venue has no paid-settlement history.
Validity, not confidence scores
We do not invent confidence percentages. Public states are objective: current, manually verified, estimate, stale, unavailable, and BTC reference. Origin, date, and conditions sit on the Updated control and in Why this ranking, not as extra chrome on every card.
When we do not name a cheapest winner
Rank order still lists venues from lowest to highest estimated total. If the gap between the top two is smaller than the estimate margin (display cents, plus the labeled BTC-reference funding leg when that approximation is in the total, plus the funding-leg gap on a 30-day historical scenario), the ranking says within margin or tied. It does not pin a cheapest winner on a difference smaller than those labeled components.
Known limits
Funding collection is BTC perpetual only. Settlement history is stored for Binance, Bybit, OKX, and Hyperliquid only. The fee-audit website CSV is the authorized Binance.com (intl/EU) Trade History export validated 10 September 2026. Fees on that file stay in BNFCR; FeeEdge does not convert them to a USDT rate. Binance.US is unvalidated. Many VIP tiers stay curated in code. Those limits are also listed next to results and on the admin data-health view.
Use and cite the data
Journalists, researchers, and newsletter writers may cite the State of Crypto Exchange Fees report with a link and the displayed update date. Sites can also embed the free comparison widget. For a dated machine-readable record of confirmed changes, use the fee-change RSS feed.
Found something off?
If a rate looks wrong, tell us at support@feeedge.com and we'll re-verify it. Accuracy reports are genuinely appreciated.