Skip to content

Crypto exchange rankings

1

Fourteen years with no customer-funds breach on record, plus a verifiable reserve proof

Taker 0.26%Volume $1.4BBreach None
2

The deepest order books in almost every pair, on a reserve proof without an audit

Taker 0.1%Volume $41BBreach 2019
3

A listed company filing audited accounts — the strongest solvency evidence here

Taker 0.6%Volume $3.1BBreach None
4

A large protection fund, offset by the thinnest corporate disclosure of the majors

Taker 0.1%Volume $5.2BBreach None
5

OKX

7.0

Lowest headline taker fee of the majors, with a monthly reserve attestation

Taker 0.08%Volume $9.4BBreach None
6

MEXC

6.6

The cheapest fees on the list and the fastest new listings — with the least disclosure

Taker 0.02%Volume $4.6BBreach None
7

Covered every loss from its 2025 incident without halting withdrawals

Taker 0.1%Volume $13BBreach 2025
8

Recovered fully from its 2020 breach; widest altcoin listing of the mid-tier

Taker 0.1%Volume $1.9BBreach 2020

How the score is built

Full method
  • Solvency evidence
    25%
  • Security record
    20%
  • Regulatory and legal record
    15%
  • Trading fees
    14%
  • Liquidity
    10%
  • Transparency
    8%
  • Evidence behind this record
    8%

Common questions

Is a proof of reserves enough to trust an exchange?
No. A self-published proof of reserves is a snapshot the exchange chooses to publish, it is unaudited, and on its own it says nothing about liabilities. It is real evidence but the weakest kind — which is why an audited or publicly listed exchange scores higher on solvency than one with only a reserve snapshot.
How do you treat an exchange that has been hacked?
Separately from one that was hacked and left users short. Time since the incident is the base, and covering every loss earns back a meaningful part of the score. An exchange that made users whole is not in the same category as one that did not.
Why is reported volume only used as a band?
Spot volume is self-reported and has been inflated across the industry for years. We use it on a logarithmic scale to place an exchange in a liquidity band, never as a precise figure, and it carries only 15% of the score.