Methodology · Model v2.4 · Last reviewed Aug 18, 2026
How we rank crypto exchanges
One global ranking is not enough: availability, rails and regulation change the right answer per market. Our engine recomputes every ranking from the same verifiable dataset — this page documents the exact model.
Reviewed by M. Rodriguez, Lead Exchange Analyst · Fact-checked by D. Okafor · Aug 18, 2026 · Editorial policy · how we make money
01 · Collect
Fees, licensing, availability, liquidity and security data are pulled from primary sources: exchange fee schedules, regulator registers, proof-of-reserves attestations and order-book snapshots. Every field carries its own check date.
02 · Score
Each exchange gets 10 factor subscores (0–10). The base model weights them as shown below; context pages (country, payment, goal) shift the weights and re-filter eligibility — weights always total 100%.
03 · Gate
A ranking page is only indexed when it has enough eligible exchanges and its order genuinely differs from its parent page. Thin combinations stay out of the sitemap.
04 · Review
Every recalculation is reviewed by the named analyst and fact-checker before release, and the model changelog records what moved and why.
Base model weights
Context pages shift these weights; the shift is always disclosed on the page.
Coverage
The registry currently defines 2,226 programmatic pages, of which 2,147 pass the quality gate and enter the sitemap. Pages that fail the gate (too few eligible exchanges, or an order identical to the parent) are kept out of the index by design.
Questions about the model? Contact the editorial team → · See the rankings →