A Merkle-tree reserve proof shows assets, not liabilities

Inherited archive entry · retained from the earlier edition. Its factual, legal and policy claims have not all been reverified for 19 September 2026. Research scope.

What a reserve proof actually checks

Two exchanges describe similar procedures, retrieved here as they stood on 16 September 2026. Kraken's proof-of-reserves page calls it 'a procedure performed by an independent 3rd party accountant' that takes 'an anonymized snapshot of all balances held and aggregates them into a Merkle tree'. Binance's equivalent page describes a four-step version of the same idea: verify ownership of the wallet addresses holding reserve assets, snapshot user balances at a point in time, generate cryptographic proofs, and build a Merkle tree linking hashed account identifiers to balances. In both descriptions, a customer can take the small piece of data assigned to their own account and check it against the published Merkle root to confirm their balance was included in the snapshot. That is a real, independently checkable claim: this specific balance existed, attributed to this account, at this moment.

What that claim does not extend to

Kraken's own page states the limits directly. The Merkle-tree method 'cannot prove exclusive possession of private keys that may have theoretically been duplicated by an attacker', 'cannot identify any hidden encumbrances or prove that funds had not been borrowed for purposes of passing the review', and says nothing about what happens to those keys or funds after the snapshot is taken. A reserve proof, in other words, verifies an asset-side snapshot. It does not verify the liability side: how much the exchange owes customers in total, whether disclosed assets are pledged as collateral elsewhere, or whether a large depositor's coins were borrowed for the specific purpose of appearing in the count. Binance's page describes the assets and the inclusion check but, in the text retrieved, does not set out these limitations in the same explicit terms Kraken uses.

Why the omission matters

A business can hold the coins it claims and still be insolvent, if what it owes customers exceeds what it holds, or if the coins it is showing are not permanently its own. A snapshot taken once, on a chosen date, also says nothing about the hour before or the hour after. None of this means a given proof-of-reserves exercise is deceptive; it means the exercise answers a narrower question than the phrase suggests, and the narrowing is stated plainly by at least one of the exchanges publishing it.

Questions to carry forward

• Does the disclosure include a liabilities figure, not only an assets figure, and are the two compared?
• Were any of the counted assets borrowed, pledged, or otherwise encumbered around the snapshot date?
• How long after the snapshot was the report published, and could balances have moved in that window?

A reserve proof is evidence about a moment, not a guarantee about a business. Reading one well means asking what was compared to what, and on which date, before treating 'proof of reserves' as a synonym for solvency.