Proof of Reserves: What It Proves (and What It Misses)

After FTX, "trust us" stopped working. Proof of reserves (PoR) is how a modern exchange shows users their assets are actually there — and for a new venue, publishing credible PoR is one of the fastest ways to build the trust you haven't earned yet. But PoR is widely misunderstood, and a checkmark is not the same as solvency. Here's what it does and doesn't prove.

How Merkle-tree proof of reserves works

The standard approach hashes every customer balance and combines them upward into a single fingerprint — a Merkle root — that represents total liabilities. Each user can verify their own balance was included in that root without seeing anyone else's (Phemex). Alongside, the exchange proves control of on-chain assets (via signatures or attestations). Assets ≥ liabilities ⇒ the exchange is at least 1:1 backed at that moment.

The subtle weakness — and the zk-proof fix

A plain Merkle tree proves your balance is in the total, but not that the exchange built the tree honestly — in principle it could insert negative or fake entries to understate liabilities, which a user checking a single branch wouldn't catch (FinanceFeeds).

The fix leading venues adopted is a zero-knowledge proof (zk-SNARK) layered on top: the exchange proves "the sum of all real, non-negative customer balances equals this published number" without revealing individual balances, and anyone can verify the proof. Binance, for example, implemented zk-SNARK verification atop its Merkle system (FinanceFeeds).

What operators should actually do

• Publish regularly, not once. A single snapshot is marketing; recurring PoR is a signal.
• Cover liabilities honestly — use the zk-enhanced approach so the liability side is provably non-negative and complete.
• Bring in a credible third party. Self-attestation is weaker than an independent auditor or verifiable cryptographic proof.
• Pair PoR with sound custody. Transparency about where keys and assets sit matters as much as the totals.
• Don't overclaim. State plainly what your PoR covers and what it doesn't — informed users trust honesty more than a green tick.

The takeaway

Proof of reserves — ideally Merkle-tree plus a zk-SNARK over liabilities, published regularly and independently verified — is now table stakes for earning trader trust. But it's a point-in-time, on-chain-only view that can't prove off-chain solvency. Treat it as one pillar of transparency alongside strong custody and compliance, and be honest about its limits. Trust, once broken, is the one thing a new exchange can't easily rebuild.

Sources

1. Proof of reserves crypto exchanges: how verification works in 2026 — FinanceFeeds
2. Merkle-tree proof of reserves explained — Phemex
3. Proof of reserves in crypto: what it proves, what it misses — Crypto Daily