Proof of reserves, explained: verifiable custody without trust

Proof of reserves is cryptographic evidence that a custodian actually holds the assets it says it does. Instead of a statement that asserts a balance, it is a check you can run against the blockchain itself.

The idea matters because a balance on a statement is only a record. It is true if the issuer is honest and solvent, and false the moment they are not. Proof of reserves turns the question from "do you trust the custodian's books" into "can you verify the assets exist."

This post explains the trust problem it solves, how it works, what it cannot do, and what to ask a custodian for.

The trust problem it solves

For most of crypto's history, knowing whether an exchange or custodian held your assets meant trusting its word. You logged in, saw a number, and assumed coins sat behind it.

Several large failures showed how fragile that assumption was. Customers saw healthy balances on their screens while the assets behind them had been lent out, spent, or never fully held. The accounting and the assets had quietly separated, and nobody outside could tell until a wave of withdrawals forced the gap into the open.

Proof of reserves attacks that directly. If reserves are provable on-chain, an asset shortfall cannot stay hidden behind an internal ledger, because the assets either exist on the chain or they do not. (Proving the liabilities side is harder — more on that below.)

How it works

Proof of reserves has two sides, and they are not equally easy.

The assets side

This is the tractable half. The custodian demonstrates control of on-chain holdings, for example by proving it can sign for the addresses holding the assets. Anyone can then confirm those holdings exist on the ledger.

On a confidential chain, where amounts are hidden by default, the custodian does not have to broadcast balances to the world. Selective disclosure through a blinding key lets it prove a specific balance to an auditor, regulator, or client without exposing the amount publicly. You get verification without turning the custodian's book into public information.

The liabilities side

This is the hard half, and any honest explanation says so. Proving you hold assets is not the same as proving you hold enough. To show solvency you also have to prove total customer claims, the liabilities. A custodian could show large reserves while owing even more, and a naive assets-only proof would miss it. Robust schemes address liabilities with techniques that let each customer confirm their balance is included in the total without revealing other customers' balances, but this side is harder to do well and easier to game.

Proving the assets exist is the easy half. Proving they cover everything you owe is the half that actually demonstrates solvency.

Two more limits are worth stating plainly:

• Point-in-time. A proof captures one moment. Assets can move the next block. Frequent or on-demand proofs matter more than a single annual snapshot.
• Control is not the same as clean title. Proving you can sign for assets does not by itself prove they are unencumbered. Segregation and a no-rehypothecation architecture are what close that gap.

How Liquid does it

The Liquid Network is an open-source Bitcoin sidechain whose Confidential Transactions hide amounts and asset types by default. That confidentiality is what makes institution-grade proof of reserves practical, because verification does not force public disclosure.

• Confidential Transactions plus selective disclosure. A holder can prove a balance to an auditor or regulator through blinding-key disclosure, demonstrating what is held without exposing it on a public explorer.
• Per-client UTXO segregation.Because each client's assets sit in their own UTXOs rather than a commingled omnibus pool, reserves map to specific clients. The proof is not just "the custodian holds X in total" but that your assets are distinctly held.

Blockstream Enterprise, the HSM-based custody platform built on Liquid, combines this with 2-of-2 or 2-of-3 multisig and HSM signing, so the assets being proven also cannot be moved unilaterally. The proof shows the assets exist; the architecture shows they cannot be quietly reused.

What to ask a custodian for

Proof of reserves is only as strong as its specifics. Press on these:

• Can you prove reserves on-chain, on demand, rather than in a periodic report?
• Does the proof cover liabilities as well as assets, so it demonstrates solvency and not just holdings?
• How often is it produced, and can I verify it myself rather than relying on a third party's summary?
• Are my assets segregated, so reserves map to my position rather than a shared pool?
• Can you prove my balance without exposing it publicly?

If a custodian can answer these with on-chain evidence, you are relying on verification. If the answers come back as attestations and assurances, you are still relying on trust.