# Lighter Volume Inflation Claim: What a ZK Proof Can and Cannot Verify
> Coinglass questioned whether Lighter's reported volume is organic. Here is what Lighter's zero-knowledge proofs actually establish about trading volume, and what no validity proof can settle.
Source: https://lighterpedia.com/privacy/lighter-volume-inflation-zk-verifiability
A data provider published a table suggesting that not all of Lighter's trading volume is organic. Lighter's whole engineering pitch is that its exchange is verifiable. Set those two next to each other and a question falls out, though probably not the one you expect. Not "who is right", which nobody outside the two parties can answer from public data. What does a zero-knowledge proof on Lighter actually establish about volume, and where does it stop?
This page reports the claim and answers the second question from primary sources. It does not adjudicate the accusation. Nothing here is financial advice.
Lighter's proofs answer **"did the operator apply the published rules to these signed transactions?"** They do not answer **"were the two sides of this trade independent parties?"** Volume quality sits outside what any validity proof can settle. That is a property of the proof system, not a defect in Lighter's implementation of it, and the same limit applies to every venue that identifies traders by key rather than by identity.
## What Coinglass Actually Claimed
On 9 February 2026, the data provider Coinglass published a 24-hour comparison of three perpetual DEXs, putting Lighter next to Hyperliquid and Aster. The figures it reported were:
| Exchange | 24h volume | Open interest | 24h liquidations |
| --- | --- | --- | --- |
| Hyperliquid | $3.76B | $4.05B | $122.96M |
| Aster | $2.76B | $927M | $7.2M |
| Lighter | $1.81B | $731M | $3.34M |
Source: [Coinglass Accuses Aster, Lighter Of Inflated Volume, Claims Hyperliquid Shows Real Activity](https://yellow.com/news/coinglass-accuses-aster-lighter-of-inflated-volume-claims-hyperliquid-shows-real-activity), Yellow, published 9 February 2026 (read 1 September 2026), and [Coinglass Ignites Perp DEX Data War Amid Hyperliquid Volume Debate](https://www.bitget.com/amp/news/detail/12560605189062), Bitget News (read 1 September 2026).
The inference Coinglass drew from those numbers, in its own words, was that the pattern may reflect "incentive-driven trading, market-maker looping, or points farming rather than organic hedging demand." It concluded that Hyperliquid displayed "stronger internal consistency across key metrics," and that the "volume quality of some competitors warrants further validation using indicators such as funding rates, fees, order-book depth, and active trader counts."
Read carefully, that is a weaker statement than the headline it generated. Coinglass did not say any exchange falsified a number. It said one ratio looked unusual and named several possible explanations, at least two of which describe activity that is entirely above board. Market-maker looping is what market makers do. Points farming is a response to an incentive program the exchange published deliberately.
The analysis drew a hostile response, and Coinglass answered it: "Coinglass simply highlighted a few discrepancies based on publicly available data. We didn't expect that a neutral, data-driven observation would trigger such hostile reactions." Critics of the analysis, quoted in the same coverage, argued a single-day snapshot cannot support the conclusion, and pointed to whale positioning, differing liquidation engines and platform-level differences in leverage ceilings as alternative explanations for the liquidation gap. Coinglass itself acknowledged that "higher leverage ceilings on some platforms could make them structurally more prone to forced liquidations."
**A note on the date.** This story is often circulated without one. The Yellow article carries a `datePublished` of 9 February 2026 in its own page metadata, which we read on 1 September 2026. Treat any version of the table you see elsewhere as a snapshot of that day, not of today.
## Did Lighter Respond?
We looked for a public response and did not find one. There is no statement from Lighter in the coverage cited above, none in [Lighter's documentation](https://docs.lighter.xyz/), and none we could locate elsewhere as of 1 September 2026. Aster does not appear to have responded publicly either.
That absence is worth stating plainly rather than filling in. It also means the rest of this page has to work from Lighter's published architecture rather than from anything Lighter has said about the accusation specifically.
## What Lighter's ZK Proofs Actually Prove
Lighter documents its architecture in [Technical Architecture: Lighter Core](https://docs.lighter.xyz/about-lighter/technical-architecture-lighter-core) (read 1 September 2026). The mechanism, in Lighter's own description:
> All exchange operations are executed deterministically via user-signed transactions. A batch of these transactions produces a new post-execution state along with a succinct cryptographic proof.
That proof is checked on Ethereum before anything becomes final. Lighter states that smart contracts on Ethereum hold the deposited assets and the canonical state root, "which encapsulates user assets and positions, public pools, order books, and other critical system components," and that each state-update proposal is published with data blobs "containing the details of the state transition (e.g., per-account changes), allowing users to independently reconstruct and verify their state."
A second guarantee sits on top of that one, and it is the one that matters if you want to check the exchange's numbers yourself. The proofs, per the same document, "attest to the validity and sufficiency of the data posted in Ethereum's data blobs, demonstrating that the posted data alone is enough to reconstruct all user states using only censorship-resistant and immutable on-chain data."
An independent review supports this reading of the scope. [zkSecurity's public report on the zkLighter circuits](https://blog.zksecurity.xyz/posts/lighter-xyz/), dated 24 April 2024 (read 1 September 2026), describes circuits that verify "that a series of state transitions were executed correctly," covering order matching, account state updates through Merkle trees, and transaction execution across transfers, withdrawals and order creation and cancellation. The report notes that "thanks to data being published on the L1, in case of emergency any L1 user is able to recompute the latest state of the L2 by replaying all of the exposed diffs." It also records its own limit: the auditors "only looked at the circuit-side of things" and did not review the L1 contract logic.
So the set of things that are cryptographically established is specific and real:
- The matching engine followed the published matching rules for the transactions in the batch.
- The margin, funding and liquidation math was computed correctly.
- The resulting state is the state that settled on Ethereum.
- The data published on Ethereum is sufficient to rebuild every account balance and position without asking Lighter for anything.
None of those four statements is about who was trading.
## Where the Proof Stops
A validity proof takes a set of signed transactions as its input and proves that the output state follows from them under the published rules. The signatures are what make a transaction admissible. A signature establishes that whoever holds a key authorized this order. It does not establish who that person is, or whether the same person also holds the key that signed the order on the other side of the fill.
Two accounts controlled by one entity, trading against each other, produce a sequence of transactions that is valid in every respect the circuit checks. The orders are properly signed. Price-time priority is respected. Margin is posted and released correctly. The proof verifies, because there is nothing about that sequence for it to fail on. The volume it generates is real in the sense that a real order matched a real order, and questionable in the sense Coinglass was pointing at.
Lighter does document a [self-trade prevention](https://docs.lighter.xyz/trading/self-trade-prevention) mechanism (read 1 September 2026): "Trades between the same account cancel the resting order (maker) instead of executing the respective trade." Lighter also notes the default behavior changes from "reduce both" to "cancel maker" as of 31 May 2026. But that rule operates on a single account. It is a fee and position-accounting protection, not an anti-wash-trading measure, and Lighter presents it as exactly that: without it, "self-trades would result in no position change and fees being paid unnecessarily."
One more structural feature belongs in the picture, and it is documented rather than alleged. Lighter runs [Public Pools](https://docs.lighter.xyz/trading/public-pools) (read 1 September 2026), in which depositors "combine funds under a designated operator who trades on their behalf," with operators whitelisted by the protocol to run strategies. Pool trading is disclosed, permitted and ordinary. It is also, by construction, professional flow rather than the retail hedging demand Coinglass used as its reference point. An exchange can therefore post a high volume-to-liquidation ratio without anything improper happening at all, simply because a larger share of its book is being worked by market makers who do not get liquidated.
This is the honest answer to the question the page opened with, and it is not a clean one: **the verifiability that makes Lighter's settlement trustworthy is the wrong tool for the question Coinglass asked.** Cryptography can prove that a computation was performed correctly. It has nothing to say about the economic relationship between two parties who both produced valid signatures. No perpetual DEX solves that today, and pointing at a ZK badge as an answer to a volume-quality question would be a category error.
For the fuller picture of what the proof system does protect, see our explainer on [Lighter's ZK verifiability](/privacy/lighter-zk-verifiability-explained) and on [whether Lighter is safe](/privacy/is-lighter-safe-verifiable).
## What You Can Check for Yourself
Start with the exchange's own numbers, because they cost nothing to get. Lighter serves market data without a key. Reading `https://mainnet.zklighter.elliot.ai/api/v1/orderBookDetails` on 1 September 2026 returned 231 order books, a summed 24-hour quote volume of about $1.34B, and open-interest notional of roughly $521M. Those are Lighter's figures, self-reported, and that is the honest way to label them. What makes them useful is that you can pull them yourself on any day and watch how the ratio behaves over a run of days instead of on one.
The settled state is the part the proofs actually secure. The state root and the data blobs live on Ethereum, and Lighter's design commits to those blobs being enough to rebuild every account. Reconstructing a full L2 state from blob data is heavy work, and most readers will never do it by hand. The property that matters is that it can be done without Lighter's cooperation.
Then look at ratios over time. The Coinglass table is one 24-hour snapshot, which was the main objection raised against it. Volume against open interest, and liquidations against open interest, tell you far more as a series. A ratio that stays odd for a month means something. A ratio that was odd on a Tuesday does not.
Last, check what incentives were live in the window you are looking at. Points and rebate programs move volume openly and by design. Lighter publishes its [Points Program](https://docs.lighter.xyz/points-program), and the [referral program](/ecosystem/lighter-referral-code) pays in fee rebates rather than points. Comparing two venues on volume also compares whatever each was paying traders to do at the time. Our [points program guide](/ecosystem/lighter-points-program) covers what Lighter documents about the current season.
What you cannot check, on Lighter or anywhere else, is the identity behind a signing key. That is the whole of the gap.
## What Remains Unresolved
Whether Lighter's volume is inflated is not established either way, and this page does not settle it. Coinglass raised a question from one day of public data and named several possible explanations. Nobody, us included, has published the funding-rate, order-book-depth and active-trader analysis that Coinglass itself said would be needed to judge volume quality.
Lighter has not responded publicly, so its own account of the ratio does not exist to report. The zkSecurity review covered the circuits and not the L1 contracts, by the auditors' own statement, so the audit trail on the settlement side is separate work by other people. And no cryptographic answer to the counterparty-identity question exists yet, on this venue or any other we are aware of.
If Lighter publishes a response, or if someone runs the longer-window analysis Coinglass asked for, this page gets updated.
Coinglass 24-hour perpetual DEX comparison, published 9 February 2026
| Exchange | 24h volume | Open interest | 24h liquidations |
|---|
| Hyperliquid | $3.76B | $4.05B | $122.96M |
| Aster | $2.76B | $927M | $7.2M |
| Lighter | $1.81B | $731M | $3.34M |
Source: Coinglass, as reported by Yellow, 9 February 2026, via Lighterpedia. https://lighterpedia.com/privacy/lighter-volume-inflation-zk-verifiability
`}
/>
## FAQ
### What did Coinglass say about Lighter's trading volume?
In a 24-hour comparison of perpetual DEXs published on 9 February 2026, Coinglass reported Lighter at 1.81 billion dollars of volume against 731 million dollars of open interest and 3.34 million dollars of liquidations, and suggested that a combination of high volume with relatively low liquidations may point to incentive-driven trading, market-maker looping or points farming rather than organic hedging demand. Coinglass did not state that any figure was falsified. It said volume quality warranted further validation.
### Do Lighter's zero-knowledge proofs prove its volume is real?
No, and they are not designed to. Lighter's proofs establish that a batch of user-signed transactions was executed according to the exchange's published rules and that the resulting state is the one settled on Ethereum. A trade between two accounts controlled by the same person is a valid sequence of signed transactions, so it satisfies the proof exactly as an ordinary trade does. Validity and economic substance are different questions.
### Does self-trade prevention stop wash trading on Lighter?
It stops one account from matching against its own resting order. Lighter documents that such a crossing cancels the maker order instead of executing. It does not address two separate accounts under common control, because nothing in a signature reveals who holds the key. That limit applies to every exchange, centralized or on-chain, that identifies users by key rather than by identity.
### Has Lighter responded to the Coinglass analysis?
We searched for a public response from Lighter and found none in the coverage or in Lighter's own documentation as of 1 September 2026. Coinglass did respond to the wider backlash, writing that it had highlighted discrepancies based on publicly available data and had not expected a neutral observation to draw hostile reactions.
### What can an outside observer actually verify about a perp DEX's volume?
On Lighter you can independently confirm that the state settled on Ethereum follows from the published data, because the proofs also attest that the posted data blobs are sufficient to reconstruct every user state. You can read the exchange's own reported volume from its public API. What no cryptographic method available today can tell you is whether the two sides of a given trade were economically independent parties.