Exchanges
What proof of reserves actually proves
Proof of reserves became a standard exchange feature for an obvious reason: after a large enough failure, “trust us” stopped working. The response was cryptographic — publish evidence that customer funds exist rather than asking people to assume it.
That is a real improvement over nothing. It is also a much narrower claim than the marketing around it suggests, and the gap between the two is where a reader should be paying attention.
The mechanism, briefly
Two halves are needed to say anything about solvency.
Assets. The exchange identifies the addresses it controls and publishes the balances. Strong versions go further and sign a message with the corresponding keys, which demonstrates control rather than merely asserting it — an important difference, since anyone can point at an address they do not own.
Liabilities. The exchange publishes what it owes customers. The Merkle tree does the clever part here: every customer balance becomes a leaf, the tree is hashed up to a single root, and each customer can verify their own balance was included in the total without seeing anyone else’s. It turns an auditing problem into something individual users can check for themselves.
Solvency is the comparison of the two. Assets alone are not a claim about solvency at all.
What it does not tell you
Most published “proof of reserves” is only the asset half. An exchange showing large holdings has demonstrated that it holds a lot of crypto. It has not said what it owes against them. A company can hold a great deal and still owe more.
It is a snapshot, and snapshots are schedulable. An attestation describes one moment. Assets that arrive before it and leave after it were present for the measurement. This is not a hypothetical weakness of the design — it is inherent to any point-in-time check, which is why frequency and unpredictability matter more than the size of any single figure.
On-chain assets are not the whole balance sheet. Loans, legal claims, obligations to other entities in the same corporate group — none of it appears on a chain. An exchange can be fully backed in the narrow sense and still be in difficulty.
Self-published attestations are self-graded. The value of an attestation comes substantially from someone independent putting their name to it. Where that signature is absent, what remains is a company publishing a favourable description of itself in a format that looks technical.
How to read one
Four questions, in order of how much they change the answer:
- Does it include liabilities, or only assets? If only assets, it is not evidence of solvency and should not be read as any.
- Can you verify your own balance is in the tree? If there is no way to check your own inclusion, the liability side is an assertion.
- Who attested to it, and did they sign it? A named third party staking their reputation is a materially different claim from an internal report.
- How often, and on what schedule? Frequent and irregular is harder to arrange around than quarterly and announced.
The honest summary
Proof of reserves is evidence about assets. It is not evidence of safety, and it is not an audit. Treated as the first of several questions it is genuinely useful. Treated as a reassurance that funds are secure, it is doing work it was never built to do.
The same reasoning transfers directly to the other place this claim is made. A fiat-backed stablecoin rests on a published reserve, and the peg holds because a redemption right is credible — so an attestation that stops at assets leaves the same gap there as it does here. What actually holds a stablecoin peg works through that in full.
It also sits inside a larger question — whether to hold assets on an exchange at all, and what you are giving up either way. That trade is covered in exchange or self-custody.