Introduction
When FTX collapsed in November 2022, it emerged that client funds had been used for the company's own investments. Proof of Reserves became the industry standard to prevent a repeat: exchanges cryptographically prove they hold reserves equal to or greater than total client liabilities.
Not every Proof of Reserves is equal, though — frequency, technology and coverage vary widely.
What exactly is Proof of Reserves?
Proof of Reserves (PoR) is cryptographic evidence that an exchange holds sufficient assets to cover all client liabilities:
- Reserves = what the exchange holds (wallets, custody)
- Liabilities = what it owes clients (your balance)
If reserves ≥ liabilities for every asset, the exchange is solvent. PoR proves this without exposing individual client balances.
How does it work technically?
- Snapshot of all client balances per asset
- Merkle tree construction — each client becomes a hash
- Wallet addresses published
- Independent verification — anyone can check wallet balances against the Merkle root
Each client can verify their own balance without seeing anyone else's.
The three technology generations
1. Merkle Tree (basic) — client balances hashed into a tree, public wallet addresses. Weakness: negative balances could historically be hidden (as FTX did).
2. zk-SNARKs / zk-STARKs — cryptographic proof that all balances are positive; hiding negatives becomes impossible.
3. Plonky2 / recursive ZK proofs — the newest generation: faster verification, smaller proofs, enabling daily instead of monthly publication, privacy-preserving and near real-time.
What is proven — and what is not?
Proven: total holdings ≥ total liabilities per asset; your balance is included; published wallets hold the reported amounts.
Often not proven: what happens between snapshots (monthly PoR = 30 blind days); margin positions (spot-only PoR); unrealized PnL; off-chain assets; obligations to non-client third parties.
That makes frequency, coverage and scope crucial. A 30-day-old snapshot of just BTC and ETH proves little for an exchange supporting 100+ coins.
Frequency: daily vs monthly vs quarterly
- Daily — the new standard, enabled by recursive ZK proofs
- Monthly — the industry norm since FTX
- Quarterly — accepted at traditionally audited platforms
- Never — a major red flag
Comparing major exchanges
| Exchange | Frequency | Technology | Coverage |
|---|---|---|---|
| Backpack | Daily (internal: every 10 min) | Plonky2 ZK proofs (PoRv2) | All assets incl. margin and unrealized PnL |
| OKX | Monthly | zk-STARK / Merkle Tree | 22 coins |
| Bybit | Monthly | Merkle Tree | Major assets |
| Binance | Monthly | Merkle Tree / zk-SNARKs | Major assets |
| Kraken | Quarterly | Independent audit | 7 assets |
| Bitvavo | Quarterly | Independent attestation | 13 assets |
Notable: only one venue publishes daily public ZK proofs covering all assets including margin and unrealized PnL; quarterly audits leave 90 blind days between checks.
What does this mean for you?
Ask five questions: How often is PoR published? Which assets are covered? Are margin and unrealized PnL included? Is the code independently monitored (e.g. OtterSec)? Is it open source?
Red flags: no PoR at all; monthly Merkle-only without ZK; limited coverage; audits without cryptographic proof. Green flags: daily public ZK proofs, full coverage, independent monitoring, open-source code.
Frequently asked questions
Is Proof of Reserves enough to trust an exchange?
It is an important signal, not the whole picture. Combine it with licences (MiCA, MiFID II), custody structure and track record.
Can an exchange cheat a PoR?
With old-style Merkle trees, negative balances could be hidden. Modern ZK proofs rule this out.
Does PoR replace an audit?
No — PoR is a real-time cryptographic proof; audits look wider (corporate structure, off-chain assets). Both have value.
How do I verify my own balance?
Most PoR exchanges offer a verification tool where you enter your account ID and cryptographically confirm your balance is included.