An Exchange Shows 100% Reserves. Does That Mean Your Crypto Is Safe?

A crypto exchange can show that it holds more Bitcoin than its customers have deposited. The figures may be independently reviewed, the wallet balances visible on-chain, and the cryptographic proof available for users to check.

That still leaves an uncomfortable question: does having enough crypto at the time of verification mean customers can get all their money back?

Consider Kraken’s Proof of Reserves snapshot dated June 30, 2026. Its published figures show a reserve ratio of 102.9% for Bitcoin and 100.5% for Ethereum. Solana is reported at 100.6%, while XRP reaches 102.3%.

Each figure exceeds 100%, indicating that the assets counted in the review were greater than the corresponding covered customer balances.

Those are meaningful disclosures. They offer evidence that would be difficult to obtain from a conventional financial institution without access to internal records.

But the percentages do not represent a comprehensive assessment of the company’s financial condition.

A reserve ratio compares a defined set of assets with a defined set of obligations. It does not automatically establish whether those assets are legally available to customers, whether other creditors have claims against them, or whether the exchange will retain control of the assets after the snapshot.

These distinctions explain why Proof of Reserves, or PoR, has become both a valuable transparency mechanism and a potentially misleading measure of exchange safety.

The underlying cryptography can be sound while the financial conclusion drawn from it goes too far.

To understand where the limitation begins, it helps to examine what the reserve percentage actually measures.

What a 100% Reserve Ratio Really Measures

The basic calculation behind Proof of Reserves is straightforward.

An exchange compares the assets available within a particular verification process against the customer balances included in that process.

The relationship can be expressed as:

Reserve ratio = Verified assets ÷ Covered customer liabilities × 100

If an exchange demonstrates control over 10,000 BTC while the corresponding customer liabilities total 9,800 BTC, its calculated reserve ratio is approximately 102.04%.

That represents a surplus of 200 BTC within the defined scope.

It does not necessarily represent 200 BTC of unrestricted corporate capital.

The distinction matters because the assets being verified may be customer property held in custody, and their treatment during insolvency depends on the relevant legal and contractual arrangements.

Kraken provides a useful real-world example of how reserve ratios differ across assets.

On its official Proof of Reserves dashboard, the exchange publishes the following figures for its June 30, 2026 snapshot:

Source: Kraken’s published June 30, 2026, Proof of Reserves dashboard. The figures describe asset-specific coverage, not consolidated corporate solvency.

The differences are more informative than a single statement that an exchange is fully reserved.

For every 1,000 units of covered BTC liabilities, a 102.9% ratio corresponds to 1,029 units of verified BTC reserves.

For every 1,000 units of covered ETH liabilities, a 100.5% ratio corresponds to 1,005 units of verified ETH reserves.

The first example has an excess of 29 units. The second example has an excess of five.

This is a proportional comparison, not a conversion between BTC and ETH or a claim that one asset is safer than another.

It demonstrates what each ratio communicates: a particular amount of reserve coverage relative to a particular liability base.

Even so, it would be incorrect to interpret the smaller Ethereum surplus as evidence that Kraken faces an Ethereum liquidity problem. The published ratio alone does not establish operational constraints, withdrawal demand, or the exchange’s broader financial resources.

A larger ratio does not automatically imply superior custody either.

Reserve calculations can vary with the assets included, wallet arrangements, outstanding customer positions, and the procedures used to verify them.

There is another complication.

The reserve assets must be compared with liabilities that are denominated in the appropriate assets. An exchange holding additional Bitcoin cannot necessarily use that Bitcoin to satisfy every stablecoin withdrawal without selling it, converting it, or otherwise obtaining the required stablecoin.

This becomes especially important during market stress, when prices, liquidity, and redemption conditions can change rapidly.

A reserve ratio is therefore most useful when you clearly disclose its numerator, denominator, asset coverage, and verification date.

Without those details, even a percentage above 100% tells readers surprisingly little.

What Binance and Kraken Actually Allow Users to Verify

A Proof of Reserves system needs to address two different questions.

The first concerns assets: does the exchange control the cryptocurrency it claims to hold?

The second concerns liabilities: how much cryptocurrency does it owe the customers covered by the verification?

Proving the first without addressing the second is insufficient.

An exchange could demonstrate control over 50,000 BTC, but that information would not establish adequate coverage if its customers were collectively entitled to 60,000 BTC.

This is why meaningful PoR systems attempt to verify both sides.

Binance and Kraken approach this problem through cryptographic accounting, although their verification arrangements differ.

Binance’s Proof of Reserves documentation describes a system combining Merkle trees with zero-knowledge proofs known as zk-SNARKs.

A Merkle tree represents a large collection of account records with a cryptographic root.

Individual customers can verify that their particular account record was included in the committed dataset without requiring the exchange to publicly disclose every customer’s balance.

The mechanism is useful because revealing a complete list of exchange accounts and holdings would introduce serious privacy risks.

However, inclusion alone does not establish that the entire liability calculation is correct.

An exchange could create a cryptographically valid commitment to an incomplete list of customers.

A customer who was included could verify their record successfully while having no direct knowledge of an omitted account.

This is a limitation of the information being committed, not a mathematical failure in the Merkle proof.

Binance’s use of zk-SNARKs addresses additional accounting risks.

Its verification circuit is designed to demonstrate that the included account balances contribute to the reported totals and that accounts satisfy the specified nonnegative net-balance constraints.

This matters because improperly using negative balances could distort an exchange’s reported aggregate liabilities.

The design and purpose of these protections are explained in Binance’s technical discussion of its zk-SNARK implementation.

But even a valid zero-knowledge proof has boundaries.

It can demonstrate that a calculation follows the rules encoded in its verification circuit.

It cannot independently establish that every real-world liability was supplied to that calculation in the first place.

Kraken incorporates another layer of oversight.

Its Proof of Reserves methodology uses independent third-party accountants to review covered customer balances and the corresponding assets. Customers can verify their inclusion through the published Merkle-tree verification process.

Kraken also accounts for covered balances associated with services beyond ordinary spot trading, including specified margin, futures, and staking positions.

That broader coverage matters because an exchange’s obligations do not necessarily end with the cryptocurrency displayed in customers’ spot wallets.

The involvement of an independent accountant provides a different form of evidence from an exchange publishing wallet addresses and self-generated proofs.

It introduces external procedures intended to test the information underlying the verification.

Yet those procedures still have a defined scope.

The customer’s ability to verify one account record does not amount to an independent examination of every corporate liability. Nor does an accountant’s report on specified reserve procedures automatically constitute a financial-statement audit.

The important distinction is not that one method is cryptographic and the other is financial.

A strong system may need both.

Cryptography helps users verify mathematical relationships and data inclusion. Independent examination can test matters that cryptography cannot establish from an internally supplied dataset alone.

Neither mechanism should be credited with proving information outside the procedures actually performed.

Ravel’s $1.05 Billion Test: How 105% Can Become 91.3%

The difference between reserve coverage and broader financial solvency becomes clearer through a simple experiment.

Imagine a hypothetical crypto exchange with US$1 billion in customer liabilities.

The exchange demonstrates control over US$1.05 billion in assets available within its Proof of Reserves calculation.

Its reported coverage is:

US$1.05 billion ÷ US$1 billion × 100 = 105%

The result appears reassuring.

Customers are owed US$1 billion, and the verified assets exceed that amount by US$50 million.

But suppose the company also has US$150 million in additional obligations to external creditors.

For this experiment, assume those creditors’ claims are valid, must be satisfied from the same US$1.05 billion asset pool, and are not subordinated to the customer claims.

The total liabilities competing for those assets would rise to US$1.15 billion.

Ravel’s calculation produces a different result.

Under these assumptions, the exchange’s assets cover approximately 91.3% of the total claims.

The apparent US$50 million surplus against customer balances turns into a US$100 million deficit once we include the additional obligations.

Nothing about the original mathematical calculation needs to be false.

The 105% figure can correctly describe one relationship while failing to describe the wider financial condition.

That is the central limitation Ravel wants to isolate.

A correctly calculated reserve ratio can still be an incomplete measure of the obligations that compete for the same assets.

There is an essential qualification to this scenario.

If customer assets are legally segregated and protected from claims by the exchange’s corporate creditors, those additional liabilities may not compete against the same assets.

In that case, simply adding all corporate debts to customer balances would produce a misleading assessment of reserve availability.

The result depends on ownership rights, segregation arrangements, custody agreements, creditor priority, and applicable insolvency law.

For that reason, the hypothetical 91.3% calculation must not be interpreted as evidence that Binance, Kraken, or another actual exchange has insufficient assets.

It demonstrates the importance of examining legal claims and financial obligations beyond a selected reserve calculation.

When assets are used as collateral, the same problem gets even harder to solve.

Suppose an exchange controls cryptocurrency worth US$1.05 billion, but a portion is subject to another party’s enforceable security interest.

An on-chain wallet balance may still show the assets under the exchange’s operational control.

Whether those assets are freely available to customers is a separate legal and financial question.

Control over the wallet does not necessarily establish an unrestricted right to every asset recorded in it.

This is why knowing the location and quantity of cryptocurrency is not equivalent to knowing every claim that may exist against it.

A public blockchain can reveal balances and movements.

It cannot automatically reveal all contractual obligations attached to those balances.

A Correct Snapshot Cannot Guarantee Tomorrow’s Withdrawals

Even a reserve verification that accurately captures the relevant assets and liabilities has another limitation.

It describes a particular moment.

Kraken’s June 2026 figures refer to balances at the time of its review. Binance similarly explains that its user-balance calculations rely on dated snapshots.

Neither mechanism can freeze the exchange’s subsequent financial condition.

Between the snapshot and a customer’s withdrawal request, assets may move, liabilities may change, new obligations may arise, and security incidents may occur.

These possibilities do not mean an exchange is necessarily misusing customer deposits.

They mean that a snapshot cannot establish what has happened since it was taken.

Kraken acknowledges this limitation in the shortcomings section of its PoR methodology. The exchange explains that the procedure cannot establish whether assets were borrowed specifically for the review or identify every hidden encumbrance.

It also recognizes that proving control over an address does not guarantee exclusive control of the corresponding private keys.

Those are different risks.

If assets were temporarily borrowed to satisfy a reserve verification, they might be subject to repayment obligations not captured by the customer liability calculation.

If cryptocurrency were pledged as collateral, another party might have enforceable rights over the assets.

If private keys were compromised, a successful historical demonstration of wallet control would not prevent a later theft.

A reserve report can be correct at the snapshot date while still being unable to rule out all three situations.

Vitalik Buterin examined these challenges in his research on Proof of Solvency and centralized exchanges.

One fundamental problem is the possibility that collateral may be used in more than one financial arrangement.

An entity can demonstrate control over an asset without having to prove that the asset is free of competing claims.

Another problem involves the completeness of liabilities.

A cryptographic system can verify that the reported balances satisfy its rules, but it must still rely on a sufficiently complete representation of the obligations being tested.

These limitations help explain why Proof of Reserves should not be treated as a replacement for broader financial scrutiny.

The issue has also drawn attention from accounting oversight authorities.

In March 2023, the US Public Company Accounting Oversight Board’s Office of the Investor Advocate issued a warning about third-party Proof of Reserves reports.

The advisory distinguished reserve-verification engagements from audits conducted under PCAOB standards and identified risks involving incomplete liabilities, borrowed assets, and the limited period covered by verification.

It also emphasized that these reports do not establish the effectiveness of an exchange’s internal controls or corporate governance.

This is not an argument that PoR information lacks value.

It is an argument against assigning a broader assurance to a report than its procedures can support.

A reserve review focused on selected cryptocurrency balances may provide useful evidence of custody and customer backing.

A financial-statement audit addresses a different and much wider set of questions, including the financial statements and disclosures within its scope.

Custody arrangements, operational resilience, cybersecurity, and customer withdrawal rights require further examination.

Even those additional reviews cannot guarantee that a financial institution will never fail.

What they can provide is more evidence about risks that a reserve percentage cannot capture on its own.

For customers, this distinction is practical.

An exchange could have sufficient cryptoassets recorded at the time of a review and still experience withdrawal interruptions because of compromised infrastructure, legal restrictions, operational failures, or subsequent financial distress.

The reserve ratio does not directly measure those events.

It measures a particular relationship between assets and liabilities using a particular verification methodology.

What Would Make Proof of Reserves More Meaningful?

The strongest conclusion is not that exchange reserve reports should be ignored.

Removing them would eliminate a useful source of information about customer-asset backing.

Instead, the question is how much evidence an exchange can provide beyond a simple percentage.

A meaningful review begins with the scope of the liabilities being measured.

Which assets are included? Are relevant margin, futures, and staking obligations covered? How are customer debts and negative balances treated? Are there categories of customer claims excluded from the published totals?

Those details influence the denominator of the reserve ratio.

The asset side deserves equal attention.

Verification should establish which wallets are controlled by the exchange, the amounts held, and the methods used to demonstrate control.

But readers also need information about whether those assets are subject to liens, borrowing obligations, or other restrictions that may affect their availability.

Where assets are legally segregated for customers, that arrangement should be distinguished from assets exposed to general corporate creditors.

Independent review adds another useful layer, especially when its scope and procedures are publicly understandable.

Kraken’s use of third-party accountants and customer-level verification provides an example of combining cryptographic evidence with external examination.

Binance’s use of zk-SNARKs demonstrates how mathematical checks can strengthen the integrity of reported liability calculations without exposing every customer’s financial information.

These approaches contribute different forms of transparency.

Neither establishes that every possible risk has been eliminated.

Frequency also matters.

A verification conducted regularly provides more information over time than a single historical snapshot.

But publishing snapshots more often does not automatically solve problems involving omitted liabilities, borrowed assets, or legal restrictions.

A frequent measurement of an incomplete dataset remains an incomplete measurement.

This is why the structure of the verification process matters more than the headline number alone.

Before interpreting an exchange’s reserve ratio, a customer should be able to establish what the ratio covers, when the information was collected, how assets and liabilities were verified, who examined the results, and what material uncertainties remain outside the review.

Those questions provide a more useful assessment than comparing two exchanges solely by whether one reports 102% and another reports 105%.

A higher ratio may indicate a larger buffer within the reported scope. It does not automatically establish stronger cybersecurity, better corporate governance, or superior legal protection for customer funds.

Likewise, a lower but still fully backed ratio is not, by itself, evidence of financial instability.

The meaning depends on the structure behind the percentage.

The development of Merkle-tree verification and zero-knowledge proofs has made exchange accounting more transparent than it was when customers had little choice but to trust internal statements.

That progress is real.

A customer who can independently verify the inclusion of their balance has information that an unverified assurance cannot provide.

But the cryptographic proof answers a narrower question than the one many users ultimately care about.

It can help establish that a verified calculation included the specified customer balances and that the reported assets satisfy the conditions of the review.

It cannot, on its own, guarantee that the assets will remain available, that no other enforceable obligations exist, or that every withdrawal will succeed.

Those questions extend beyond mathematics into custody, law, accounting, and operational risk.

An exchange can prove it had enough crypto for the liabilities it measured. That is not the same as proving that every customer’s crypto will always be safe.