Proof of Reserves and Auditing Exchanges

The question these proofs try to answer

After exchange failures, customers asked a natural question: how can we know the platform actually holds what it says it holds? Proof of reserves is the name for a family of techniques and reports in which an exchange tries to show that its holdings cover what it owes customers. It is an attempt to reduce counterparty risk through transparency.

An honest solvency check needs two numbers: what the exchange has (assets) and what it owes (liabilities). A proof is only meaningful if both are covered.

How the assets side works

On a public blockchain, an exchange can demonstrate that it controls certain addresses by signing a message with the keys for those addresses, or by moving a small amount in a way that an observer can verify. Anyone can then look up the balances of those addresses and add them up.

Limits on this side:

• Timing. A balance shown at one moment can be borrowed or moved right before the snapshot and returned afterwards. A single point-in-time snapshot does not show what happened in between.
• Ownership versus control. Showing control of an address does not show that the coins are unencumbered. They might be pledged as collateral or already owed elsewhere.
• Off-chain assets. Cash, loans and other assets are not visible on a blockchain and need traditional accounting evidence.
• Overstated assets. The exchange chooses which addresses to show, and addresses it merely controls jointly with others, or has pledged elsewhere, could be presented as if they were fully its own.

How the liabilities side works

Customer balances are private, so an exchange cannot just publish them. A commonly described method uses a Merkle tree: each customer’s balance is combined with an identifier and hashed, and these hashes are combined pairwise, step by step, up to a single root hash that is published along with a total of all balances.

An individual customer can then check that their own balance was included, by following the short chain of hashes from their entry to the root. If enough customers check, it becomes harder for an exchange to quietly leave accounts out or understate what it owes.

Limits on this side:

• Only checkers are protected. If few customers verify, omissions may go unnoticed.
• Exclusions can be hidden. An operator could omit accounts or include false negative balances in ways that are difficult to spot without careful design.
• Other debts are missing. A tree of customer balances does not show loans, legal claims, or obligations to other companies.

What a reserve report does not establish

Taken together, a proof of reserves usually cannot show:

1. Solvency in full. It cannot prove that total liabilities, including those to non-customers, are smaller than total assets.
2. Absence of fraud. Related-party arrangements, hidden loans or rehypothecation can sit outside the report.
3. Good controls. Security practices, key management and internal oversight are not tested by a balance snapshot.
4. The future. A healthy report today says nothing certain about next month.

This is why many specialists describe these reports as a useful signal but not a guarantee. A disclosure that is incomplete may also create false comfort.

Audits, attestations and the differences

Different words describe different levels of assurance:

• Self-published reports come from the exchange itself and offer the least independent assurance.
• Attestations or agreed-upon procedures involve an outside firm checking specific items at a specific date, within a defined scope. The scope matters; such a report is narrower than a full audit.
• Full financial audits review the statements and controls of a company under recognized standards, though crypto-specific audit practice is still developing and standards vary by country.
• Regulatory supervision can involve required capital and segregation of client assets in some jurisdictions, but rules differ widely.

The same distinction appears with stablecoins, where reserve attestations are discussed in the context of stablecoin reserves.

How to read a proof

Some questions that help a reader:

• Are liabilities covered, or only assets?
• Is there an independent party, and what was its scope?
• Can customers verify their own balances?
• Is the report regular, or a one-off published under pressure?
• Does it say anything about loans, pledged collateral or related companies?
• Does the exchange publish its terms on whether customer assets are segregated?

If the answers are vague, treat the report as marketing rather than evidence.

The takeaway

Reserve proofs are an improvement over asking customers to trust blindly, and cryptographic methods let some parts be checked by anyone. But they address a slice of the risk. The sturdier protection against platform failure is usually to limit how much you leave on any one platform, a theme in the case studies of exchange failures. Cryptocurrency carries substantial risk, and nothing here is financial advice.