How we measure what others advertise.

Most comparisons quote the midpoint between best bid and best ask — a price nobody can trade at, for a size nobody specified. We instead take your exact order size and consume the venue’s live orderbook level by level, the way a real market order fills. The volume-weighted average of those fills is your effective price; the gap to the market reference is the spread cost — the part the fee table never mentions, and the part that grows with order size.

Venues are ranked by a single quantity: the net amount you end up with after every cost. Buying, that is coins received; selling, euros received. This number is independent of any reference price, immune to framing, and makes exchanges and DEXs directly comparable, because gas simply reduces the coins received — the swap quote on the comparison page runs through exactly this engine. The euro breakdown next to it exists to explain the ranking — never to produce it.

Live orderbook snapshots are continuously ingested and every figure carries its data age, visible in the table. A feed that is stale or implausible is dropped for that round — the venue simply disappears from the comparison until its data is healthy again. A missing row beats a wrong row; we would rather show fewer venues honestly than a larger table with one guess.

Some venues — Bitpanda among them — publish no orderbook. We will not pretend otherwise: their rows use the venue’s published list pricing, are marked estimated, and are never presented as live. When a venue offers a public book later, the estimate is retired the same day.

Some venue links earn us an affiliate commission; they are disclosed and they change nothing about the order of the table. Ranking is computed from market data alone — a venue that pays us nothing ranks exactly where its costs put it. This rule is enforced in code and covered by tests, because the comparison is only worth something while it cannot be bought.

For contrast: some comparison sites state in their own FAQs that affiliate partners are grouped at the top of their lists. That is a legitimate business model — it is just not a measurement. We sell the measurement.

Moving coins off an exchange is a priced decision, and the Move check measures it: for one venue, one asset and one target chain we enumerate the routes — a direct withdrawal where the venue supports the chain, or a withdrawal plus one bridge leg, quoted via LiFi (a cross-chain bridge router — it finds a path from one chain to another, we just read its price). Each route carries two numbers with distinct jobs: what arrives on the target chain, and the net units after all costs — the ranking quantity, which also charges gas paid on the source chain. Ranking on the arrival number alone would crown a route that quietly burns more gas.

The same honesty rules as everywhere else: withdrawal fees are curated exclusively from official venue sources with verification dates; a missing row means “not verified”, never “not offered”; venues without a published fee are explained, not ranked; amounts below a venue’s published minimum are not executable and therefore never ranked; bridge quotes are requested on the amount that actually arrives after the withdrawal fee. Venue processing time is not published anywhere — so we never invent it. The bridge leg is neutral: LiFi is queried keyless (no account, no special access — anyone can ask it the same question), with no integrator fee (no markup added on top of the bridge’s own price) and no affiliate relationship.

Live today: orderbooks from public exchange APIs via CCXT (an open-source connector library — it just relays each exchange’s own public data, nothing more) · EUR reference via CoinGecko · base-tier fee schedules curated from each venue’s published pricing, re-verified on change, each carrying its source and last-verified date.

Quote currency: the currency switch selects which order books are measured. On EUR we walk each venue’s EUR pair; on USD we walk each venue’s own dollar rail — a genuine fiat-USD book where the venue has one, its deep USDT book where it does not, disclosed on every row. Nothing is converted between currencies on the pricing path, and every figure states the currency it was measured in. USD and USDT books are ranked as two separate groups, because a USDT price assumes the peg — ranking the two together would count 1 USDT as exactly $1. A venue is omitted from the dollar view, with the reason stated, when it has no public dollar book or when its fee schedule is not verified to apply on that rail (pair fee classes and promotional zero-fee pairs are common; an omission is cheaper than a wrong number). The EUR↔USD reference rate has exactly one job left: re-expressing your typed budget when you flip the switch — it never touches a measured figure or the ranking.

Withdrawal costs (optional toggle): curated from each venue’s official fee schedule or public fee API, each row carrying its source and verification date — currently BTC and ETH on their native chains. Several venues adjust these fees with network conditions; our figures are the published values as of verification. Venues that publish no fixed fee (Coinbase passes the live network fee through, disclosed before you withdraw) are marked “excl. withdrawal” — we exclude rather than guess.

Convert / Instant-Buy markup: the convenient one-click buy products carry a spread the spot order book does not — our table prices spot, so we flag this markup separately where a citable figure exists (currently Coinbase and Kraken; Bitpanda has no separate spot book, its premium is what the table already prices). These are curated typical ranges with sources, never live quotes, and they never affect the ranking. A systematic, measured per-venue study is in build.

DEX swap quotes (live on the comparison page): one LiFi-aggregated route per pair — ETH, WBTC, USDC, USDT, DAI on Ethereum — quoted keyless with no integrator fee and no affiliate relationship. The headline is LiFi’s own expected and minimum output; hidden spread is measured against the same EUR reference as the exchange table; gas uses the route’s own estimate with Alchemy as fallback — and when neither source answers, the quote marks gas as unavailable and the total excludes it, flagged in the UI. No route, no number: when LiFi finds nothing we say so instead of guessing.

We list capabilities here only once they are live — this page describes the product you are using, not the one we are planning.