How to read an exchange's proof of reserves

When you leave coins on a centralised exchange, you are trusting it to hold enough assets to repay every customer. Proof of reserves is a set of techniques exchanges use to show that they do.

The two halves of the proof

A meaningful proof has to cover both sides of the balance sheet:

• Assets: the exchange shows wallet addresses it controls, often by signing a message with those addresses, so anyone can check the balances on-chain.
• Liabilities: the exchange publishes a commitment to every customer balance, usually as a Merkle tree, so the total can be compared with the assets.

A report that shows only assets is incomplete. Large wallets mean little if customer balances are even larger.

Checking your own balance

Many exchanges let you verify that your account is included in the liabilities tree. You receive a record ID and can confirm that your balance was counted in the published total. If that option exists, use it; it is the part of the proof only customers can test.

What proof of reserves cannot show

• Debts that are not customer balances, such as loans the exchange has taken out.
• Whether assets were borrowed just before the snapshot and returned afterwards.
• The quality of the assets, for example a large holding of the exchange's own token.
• What happens between snapshots, if reports are infrequent.

Attestation versus audit

Some reports are an attestation: an accountant confirms specific figures at one point in time. A full financial audit is broader and examines the whole business. Read which one you are looking at before drawing conclusions.

Key takeaways

• Look for both assets and liabilities, and for frequent reports.
• Verify your own inclusion when the exchange allows it.
• Keep only what you need on an exchange; self-custody removes this counterparty risk.