Proof of Reserves Explained

Kraken, Coinbase and Ledger all describe the basic method: an independent accountant takes an anonymised balance sheet snapshot, compiles it into a Merkle tree, and signs the Merkle root. The root can be shared publicily; the individual balance leaves can be compared with the root to verify inclusion.

CryptoExID highlights that this only attests to the calculation the exchange defined, using data it supplied.

This has limits. Firstly, published reserves are only a snapshot. An exchange could build up assets for the audit, then let them run down over the next month. Banxa speaks of a "snapshot that can be dressed up".

Secondly, published liabilities are the snapshot, plus nothing more than the exchange defined as customer balances. Banxa speaks of "assets shown without liabilities".

Thirdly, a proof of reserves is only a defined update and attestation, not a full audit, even if conducted by an independent accountant. Coinbase and CryptoExID both distinguish an "attestation" from a "full audit".

So the main concern should be the exact scope: not a pretty picture of solvency from the top, but a specific brief attestation to a particular point on a balance sheet.

For this reason, a proof of reserves is most useful when the exact scope noted is easily understood - and can be trusted. While a three-way comparison between assets, a liability snapshot, and issued shares can be regarded as a solid process, Before relying on one, check that:

• Liabilities are published.
• Auditors and signers are independent.
• Customers can verify each account is included in the liability tree.

A proof of reserves is then a process which quickens the understanding and encourages good behaviour, rather than confirming underlying balances.