Proof of reserves, a report showing the assets a platform holds measured against what it owes its users, has become the crypto industry's standard trust signal. It got there fast. When FTX collapsed in November 2022, CoinDesk reported nine major exchanges announcing within a single day that they would publish reserve certificates in the month ahead. Nearly four years on, the technique is everywhere, and so is the confusion about what it actually proves.
This piece explains how proof of reserves works, how to read one as a user, and the specific things a PoR report cannot tell you. That second half matters more, because the gap between "this platform publishes PoR" and "this platform is solvent" is exactly where the 2022 failures lived.
What a proof of reserves report contains
A proof of reserves report makes two claims that only work together. First, that the platform controls a set of assets, shown through published wallet addresses, signed transactions or statements from the venues holding the funds. Second, that those assets cover the user balances the platform owes. Dividing the first number by the second gives the reserve ratio, and a ratio of 100% or more means assets matched or exceeded user balances at the moment the report was taken.
The user-balance side usually relies on a Merkle tree. Every balance is run through a hash function, which turns it into a short fixed-length fingerprint, those fingerprints are combined in pairs and hashed again, and the process repeats until one root fingerprint represents every balance on the platform. The structure, explained well in Chainlink's education hub, lets you verify your own balance was counted without seeing anyone else's. The platform gives you a verification path, you check that your balance leads to the same published root, and if the platform had left your account out or altered it, the root would not match.
The asset side is more straightforward. Addresses on public blockchains can be checked by anyone through a block explorer, the public search engines for on-chain data. Balances held at another venue, such as an exchange account, need a statement from that venue, which shifts the question to whether you trust the statement.
Why everyone started publishing PoR in 2022
FTX did not fail because its reserves were slightly short. It failed because customer funds had flowed to Alameda Research, its affiliated trading firm, a fact users discovered only when they tried to withdraw and the money was not there. In the weeks after, proof of reserves went from a niche practice (BitMEX had demonstrated one back in 2021) to the favored answer for exchanges needing to reassure customers that deposits were really there.
The rush had a telling postscript. Mazars, the accounting firm behind several early reserve reports, suspended its crypto work a month later, with coverage citing concern over how the public was reading the reports. People were treating narrow snapshots as clean bills of health, and the reports were never that.
What proof of reserves does not prove
This is the part most PoR pages skip, though not all. Kraken's page, per its own disclosures, lists several of these limits explicitly.
A PoR report is a snapshot of one moment. It shows the position on the report date, not the day before or the day after. Assets can be borrowed for the snapshot and returned afterward, and the cryptography cannot distinguish borrowed coins from owned ones. The method also cannot rule out encumbrances, assets that appear in reserves but are already pledged as security for a loan or other obligation somewhere else.
The balance tree has a subtler gap. Your inclusion proof shows your balance is in the published set. It cannot show the set is complete. A platform carrying debts outside the tree, a loan from another company, a legal claim, an obligation to a large counterparty, can publish a clean Merkle tree that simply does not include them. How platforms reuse deposited assets, the practice known as rehypothecation, is its own subject, covered in our rehypothecation explainer, and a reserve snapshot does not settle it.
Who checks the report matters too. Some reports are verified by an outside firm, a limited review called an attestation. Some are self-attested, meaning the platform publishes and stands behind the numbers itself. Neither is a full audit of the company's finances. A SOC 2 report, which platforms also like to cite, is an examination of how a company runs areas like security and data handling, and says nothing about whether its assets cover what it owes. Researchers are still working on making these reports easier for ordinary users to verify and harder to game, a useful signal of how unsolved the problem remains.
None of this makes proof of reserves worthless. It makes it a starting point rather than a complete assessment. A platform publishing regular PoR with named wallets gives you something concrete to check, which is categorically better than a trust-us page. The collapses of Celsius and BlockFi both ran on balance sheets users could not see at all, and narrowing that blindness is precisely the job a reserve report does.
How to actually read a PoR page
A practical checklist, in rough order of weight. When was the last report, and is there a history of regular snapshots rather than a single one published under pressure. Is the reserve ratio at or above 100%, and which balances and obligations were counted in the calculation. Are wallet addresses published and checkable on-chain. Who stands behind the report, the platform itself or an outside firm, keeping in mind that a platform-published report is evidence the platform chose to show, not independent assurance. And does the page admit its own limits, because a PoR page claiming to prove solvency outright is overselling what the method can do.
Applying that checklist to a live example, EarnPark's monthly proof of reserves report showed a reserve ratio of 106.2% against user balances in its September 2026 report, with on-chain wallet addresses published for anyone to verify and a history of prior snapshots on the same page. The reports are attested by the platform itself rather than an outside firm, which the page states plainly, so a reader should weigh the checkable parts, the on-chain addresses, above the stated ones. Custody arrangements, with a significant portion of holdings held through the institutional custodian Fireblocks, are documented in the trust center, and the published roadmap lists a further on-chain verification step as planned. Regular snapshots, checkable addresses, stated limits. That is the pattern the checklist rewards, wherever you apply it.
Verdict on proof of reserves
Treat proof of reserves as a necessary signal, not a sufficient one. A platform that publishes regular, checkable reserve reports has given users a way to catch it lying about assets, a constraint FTX never accepted. A platform that publishes nothing is asking for blind trust in a market that has repeatedly punished it. Just keep the conclusion sized to the evidence. A PoR proves assets were there at the snapshot, and everything beyond that still rests on the platform's practices, disclosures and incentives.
If you are weighing where to put assets to work, the same checklist applies to yield platforms as to exchanges. Our comparison of earning platforms runs through custody and transparency for the major options, and the reserve reports described above are available to check before you start earning anywhere.
FAQ
What does proof of reserves actually prove? That the platform controlled certain assets at the snapshot date, and that your balance was included in the set of user balances those assets were measured against. Nothing more, nothing less.
Does proof of reserves prove an exchange is solvent? No. It cannot see borrowed coins, pledged collateral, debts owed outside the published set or obligations that are not user balances. It is evidence toward solvency, not proof of it.
How do I verify my own balance in a Merkle tree proof? Platforms with user verification give you a verification record in your account. You check that your balance, run through the published process, leads to the same root fingerprint the platform published for that report. If your account had been left out or altered in that snapshot, the result would not match.
How is proof of reserves different from an audit? A financial audit examines a company's complete finances, assets and liabilities, under accounting standards. A PoR is a narrow, fast, repeatable snapshot of whether specific assets covered user balances on one date. A SOC 2 report is different again, covering how the company runs its systems rather than what it holds.
How often should proof of reserves be published? There is no industry standard, so frequency is something to check yourself. Monthly or more frequent snapshots with visible history beat a single report published once under pressure.