On-chain investigationLighter · Ethereum · Hyperliquid16 November 2026

One wallet pulled $10.4 million of USDC out of Lighter and swapped it to ETH in four minutes. The ETH is now leaving in payments of 160, each one bound for Hyperliquid.

Posts on X put the loss at $15.5 million across two market maker accounts. We followed the money that left Lighter, from the withdrawal on Ethereum to ten accounts on Hyperliquid.

At a glance
Lighter is a perp DEX that settles on Ethereum. On 16 November 2026, traders on X reported that someone had drained two market maker accounts there. Lighter says it is looking into a security incident that involves one outside party, and it has paused withdrawals. We followed the money on-chain. One wallet withdrew 10.39 million USDC from Lighter and swapped it to ETH in four minutes. It then sent 1,600 ETH through ten new wallets to Hyperliquid, and each one bought XMR1. The other 4,165 ETH was still in the wallet when we last read the chain. A second amount, reported at $5.09 million, had not left Lighter when we last read the chain.
$10.39M
USDC withdrawn from Lighter to one wallet
4 min
To swap $10.36 million of it to ETH on CoW Swap
4,165 ETH
Still in the wallet after the tenth payment
5,583
XMR1 held by ten Hyperliquid accounts
Key findingsEach line has its own share link
  1. $10.39M

    One withdrawal from Lighter's contract on Ethereum, 10,393,057 USDC, at 10:11 UTC on 16 November 2026.

  2. $10,200

    What the wallet sent into the Relay bridge before the withdrawal. Ten minutes after it, the wallet deposited the same amount into Lighter.

  3. 4 min

    To swap 10.36 million USDC to 5,752 ETH in five orders on CoW Swap. The wallet kept no USDC.

  4. 72%

    Share of the ETH still in the wallet after the tenth payment, at 11:36 UTC: 4,165 ETH, about $7.5 million.

  5. 5,583

    XMR1 held on Hyperliquid by ten accounts, bought for $2.88 million. All ten still held it at 12:08 UTC.

  6. 12 sec

    Between the wallet's first 160 ETH transfer and the first fake copy of it from an address poisoning bot.

01 · What happenedThe Lighter incident, in one morning

On the morning of 16 November 2026, traders on X began to post about a large loss on Lighter. LIT, its token, fell by about a fifth.

Lighter is a perp DEX, an exchange for perpetual futures. It runs on its own rollup, and it settles on Ethereum. Users deposit USDC into a contract on Ethereum, trade on the rollup, and withdraw through the same contract.

Later that morning, the trader 0xTria shared a breakdown from a Korean analyst. It said the protocol itself held up. The API keys of two large market maker accounts had leaked, and the attacker used those keys to empty both accounts. The post put the loss at about $15.5 million.

Minutes later, Lighter posted that it was looking into a security incident that "appears to involve a single compromised external party". It paused all withdrawals and the creation of new pools.

Trades inside Lighter happen on its rollup, and Bitquery does not read that rollup. But every dollar that leaves Lighter passes through its contract on Ethereum, and Bitquery reads every transfer there. So we started at the contract and followed the money out.

02 · The reportsWhat has been reported so far

This part comes from Lighter's own post and from posts on X. We could not check it, because it happened inside Lighter's rollup.

Lighter's statement points away from the exchange itself. It says the incident appears to involve one compromised outside party. It has not said that its contracts or its rollup failed.

The posts on X agree on the outline. They say the API keys of two market maker accounts leaked, and that someone used those keys to take the accounts' USDC. Lighter has not said how the funds were moved, so we do not describe a method here.

What the posts on X reportFigure
Taken from account #714710$10.39M
Taken from account #714723$5.09M

The first of those numbers can be tested on Ethereum.

03 · The withdrawalOne wallet took 10.39 million USDC out of Lighter

It shows up in one wallet.

The wallet first received money early that morning, about $50 of ETH for gas. Three minutes later, it sent $30 of that ETH into Relay, a bridge. Two hours on, it received about ten thousand USDC through Across, another bridge, and sent most of it into Relay as well. We could not see where Relay delivered either payment. More gas, about $120, arrived at nine.

Shortly after ten, Lighter's contract paid the wallet 10.39 million USDC in one transfer. That matches the figure reported for the first account.

Ten minutes later, the wallet sent the same small amount again. This time it went straight into Lighter's contract as a deposit.

The wallet on Ethereum, 16 November 2026 (UTC)What moved
06:44 · First gas arrives0.028 ETH
06:47 · Out, into the Relay bridge0.017 ETH
08:45 · In, through the Across bridge10,526 USDC
09:00 · More gas arrives0.067 ETH
09:35 · Out, into the Relay bridge10,200 USDC
10:11 · In, from Lighter's contract10,393,057 USDC
10:13 to 10:17 · Swapped on CoW Swap10,360,000 USDC
10:21 · Out, deposit into Lighter10,200 USDC
10:46 · Last USDC swapped23,183 USDC

No second withdrawal followed. Within the hour, Lighter had paused withdrawals. When we last read the chain, the contract had paid this wallet once. If the money left the second account, it had not left Lighter.

We also checked who else Lighter's contract paid that morning. Three other payments were above half a million USDC, and they went to two wallets. One, of 2.4 million, went to a wallet 96 seconds before the big one. That wallet had withdrawn 9.4 million the day before, and it forwarded both amounts to the same address. It looks like a regular large user. We saw no transfer between it and the wallet in this story. The other two, of 0.5 million and 1.9 million, went earlier that morning to a wallet that Bitquery labels as a solver for the Relay bridge. Solvers front money for bridge users and collect it back later.

04 · The swapTen million dollars became ETH in four minutes

The wallet did not hold USDC for long. Circle, the stablecoin issuer behind USDC, can freeze it in any wallet. ETH has no issuer.

Ninety-six seconds after the withdrawal, the wallet placed its first order on CoW Swap, a DEX that fills orders in batches. Five orders went through in four minutes. A sixth order cleared the last of the USDC half an hour later.

Orders on CoW Swap (UTC)USDC sold → ETH bought
10:13:231,000,000 → 555.71
10:14:232,000,000 → 1,111.26
10:15:115,000,000 → 2,775.37
10:16:112,000,000 → 1,109.43
10:17:23360,000 → 199.87
10:46:2323,183 → 12.84
Total10,383,183 → 5,764.47

The books close to the cent. The wallet received USDC in two transfers. It sent the two small payments to Relay and to Lighter, and swapped the rest. It holds no USDC now.

05 · The exitTen wallets took 1,600 ETH to Hyperliquid

Two minutes after the last large swap, the wallet began to pay out. It sent 160 ETH to a new wallet, then the same to another. It did this ten times in 77 minutes. The payments came in pairs, about twenty minutes apart.

Each new wallet did the same thing. Within two minutes, it sent all of its ETH into LI.FI, an app that routes money across chains. About 16 to 18 minutes later, an account with the same address bought XMR1 on Hyperliquid.

Wallet · 160 ETH in (UTC)XMR1 bought on Hyperliquid
0x63d09d9bd23c3ed507771db8bdd65bf557fe968c · 10:19552.20 for $287,552
0x5c239e7607975be5cf44ec71d2cf1c66d878f746 · 10:20559.28 for $287,770
0x8b3a5786d4503faeffa78e30c31a57bfda93b121 · 10:39555.45 for $288,515
0xd8427872a4e9ddf00ee07f02c644447aa6b99926 · 10:40557.12 for $288,296
0xd1fd67c883466947e1f7dbf06f284cb2785b8c77 · 10:57556.56 for $288,848
0x26f3cc7d0c4fa5566f33cdbf8569b96f11cfc74e · 10:58554.06 for $288,537
0xb57c2a0a4b20d7ec73395dc23fe182e9e6830ab1 · 11:17567.21 for $287,613
0xed762776faa72631083f02b0dc91593426634c14 · 11:17561.15 for $289,248
0x5f75b9a476d0012ffbb0af69bf43ddd03beb8bcd · 11:36563.56 for $288,862
0x2e10f4d519e3414fdf7a9baabdc396219276ddb4 · 11:36560.73 for $288,633
Total5,587.32 for $2,883,874

XMR1 is a token on Hyperliquid named after Monero, a privacy coin. We did not check what backs it. At the last snapshot we read, the ten accounts held almost all of it. Nine also held a few dollars of USDC, and one held about $1,800. None of the XMR1 had moved.

The same exit showed up two days earlier. In the Kava hack, the proceeds also went through LI.FI, split across fresh wallets, and those wallets also bought XMR1 on Hyperliquid. Is there a link between the two? The chain shows the same route and the same token. It shows no transfer between the two sets of wallets. A shared route is weak evidence, since anyone can copy it.

The rest of the ETH is still in the first wallet. After the tenth payment, at 11:36 UTC, it held 4,165 ETH, about $7.5 million. That is 72% of what the swaps bought. Nothing more had left it when we last read the chain, about 45 minutes later.

An earlier version of this article said 83% of the ETH had not moved. That was true after the sixth payment. Four more followed.

You can open this trace in MoneyFlow Lite and follow each wallet hop by hop. The graph runs through the swap and bridge contracts, which other users share, so some of its flows are larger than this wallet's own. You can also ask the Bitquery MCP server where the money went.

06 · The botsScam bots found the wallet in 12 seconds

One more group was watching the wallet.

The wallet sent its first 160 ETH just before 10:20 UTC. One block later, a token contract logged a second transfer of the same size from the same wallet. The token was a fake with a copied name. The receiver was a new address that began and ended with the same characters as the real one.

This is address poisoning. A bot plants a look-alike address in a wallet's history and hopes the owner copies it for the next transfer. Here the bots hoped the next payment would come to them.

Within the hour, the wallet's history held fifteen fake copies of its transfers. Ten went to look-alike addresses. Five named the real receivers but moved a fake token. About two dozen dust payments came in from look-alikes too. Each of the first six real receivers had at least one twin.

It did not work. Every real payment went to the real wallet. But anyone who reads this wallet's history should filter by token contract first. A list sorted by symbol shows more than twice as many ETH transfers as the wallet really made.

07 · The limitsWhat the chain cannot tell us

Nobody has named who runs the wallet. Its money came in through bridges and left through one.

We could not see inside Lighter. The account numbers and the leaked keys are reports from X. Lighter has not published its findings, and it may describe the cause differently.

We did not see where Relay delivered the first small payment. It may have gone to Lighter, since the amount matches the later deposit. That is our inference.

We traced the gas one step back. Two wallets sent it, and one of them funds the other. We did not go further.

We do not name the market maker. Posts on X name a firm. Neither that firm nor Lighter has confirmed it.

08 · For usersWhat the Lighter incident means for users

Lighter says the incident appears to involve one compromised outside party. It has not reported a wider loss of user funds, and we saw one withdrawal tied to the event on Ethereum.

If you run a trading bot on any exchange, rotate its keys after news like this, and lock each key to your server's IP address where the exchange allows it.

Withdrawals were paused when we wrote this. Follow Lighter's own channels for the time they reopen.

For anyone who tracks stolen funds, most of this money can still be watched. One Ethereum wallet holds the ETH. Ten Hyperliquid accounts hold the XMR1. The Hyperliquid accounts are the more urgent ones, because whoever redeems that XMR1 for real Monero ends the visible trail.

09 · The traceHow we traced the Lighter withdrawal

We started with the wallet that several posts on X named, and checked it against Lighter's contract. Bitquery's address labels mark that contract, the CoW Swap contract, and the Across, Relay and LI.FI routers.

We read every transfer in and out of the wallet on Ethereum, then every transfer of the ten wallets it paid. On Hyperliquid, we read each account's fills and its balance in the latest snapshot. Dollar values use Bitquery's price for ETH at the time of each transfer.

We left out the fake tokens and dust from the poisoning bots when we added up the money.

For cases like this one, we also run crypto investigation services.

Limits on these findings

Ethereum data comes from Bitquery, read on 16 November 2026 up to about 12:20 UTC. The wallet's ETH balance is as of 11:36 UTC, its last transfer. Hyperliquid balances come from the 12:08 UTC snapshot. Anything after that is not covered.

Lighter runs on its own rollup, which Bitquery does not read. The account numbers, the $5.09 million figure and the leaked API keys come from posts on X. We did not verify them. Lighter has not published its report and may describe the cause differently.

We tie the wallet to the event by one fact: Lighter's contract paid it 10,393,057 USDC, the amount the posts report for the first account. We attribute the wallet to nobody.

The ten wallets are tied to the first wallet by the 160 ETH each received from it. We found the LI.FI deposit for nine of them; the tenth is tied by that payment and by the XMR1 its account bought. The link from each LI.FI deposit to each Hyperliquid account rests on the shared address, the timing and the amounts. We did not read the bridge's own records.

That the first 10,200 USDC reached Lighter through Relay is our inference from the amount. We did not see the delivery.

The $2.88 million is the sum of the XMR1 fills at their fill prices. We did not check what backs XMR1 or what its holders can redeem.

The $7.5 million value of the remaining ETH uses a price of about $1,807 per ETH. The wallet may have paid out again after our last read.

The note on the Kava hack compares routes. We found no transfer between the two sets of wallets, and we do not say the same people were behind both.

Lighter's contract made three other payments above half a million USDC that morning, to two wallets. One wallet shows a repeat withdrawal pattern from the day before. The other carries a Bitquery label as a Relay bridge solver. We link neither to the event.

The counts of fake transfers and dust payments cover the wallet's history up to 11:00 UTC. We do not know who runs the bots.

Links between wallets are money links. They do not prove who controls a wallet.

10 · The recordThe addresses behind the Lighter incident

0xb3fffca23d4444ee73349825acd84f7b5020e387
0x3b4d794a66304f130a4db8f2551b0070dfcf5ca7
0x5b6f27ded0e9a66ef6fa107083e28db9914e46d7635cc5112d760b9cb2f958c9
0xe6b8e54bb04420c42b20062d6941813f02b7e09556c44cc6e27bf6abcd78ca07
0x155076049c3992e82546009c840ad2d79e311522491c401dc380c51c592dc050
0x196838a707b02403dca4c4e4f64067ec45b0bdee4c7673b9e05afd08f89213fb
0x2a5f6efd130af6900596069b4a48c74c0e1b1dc5742577482e42c59a590060d2
0x63d09d9bd23c3ed507771db8bdd65bf557fe968c
0x5c239e7607975be5cf44ec71d2cf1c66d878f746
0x8b3a5786d4503faeffa78e30c31a57bfda93b121
0xd8427872a4e9ddf00ee07f02c644447aa6b99926
0xd1fd67c883466947e1f7dbf06f284cb2785b8c77
0x26f3cc7d0c4fa5566f33cdbf8569b96f11cfc74e
0xb57c2a0a4b20d7ec73395dc23fe182e9e6830ab1
0xed762776faa72631083f02b0dc91593426634c14
0x5f75b9a476d0012ffbb0af69bf43ddd03beb8bcd
0x2e10f4d519e3414fdf7a9baabdc396219276ddb4
0x66cf2f85115cc20bf439720927c1f93de9672068
0xf70da97812cb96acdf810712aa562db8dfa3dbef
0x61cd0c28f95cbd5b67597591f9df1a250a344cd8
0x67c72adcf158dcb84a12f7e8956c70c48af300c8
Run it yourself

Follow the money in plain English

Every transfer in this story comes from Bitquery's data. The Bitquery MCP server puts that data behind an AI assistant, so you can ask where a wallet's money went without writing a query. To see the same hops as a graph, open this trace in MoneyFlow Lite. To build your own alerts, use the Ethereum API.

Follow USDC out of an exchange contract to the wallet that received itList every swap a wallet made, minute by minuteRead what a Hyperliquid account bought and what it holdsCheck an address against Bitquery's labels
Explore Bitquery MCP →Ethereum and Hyperliquid data from Bitquery. Data read on 16 November 2026.
Legal disclaimer

This article is provided for informational and educational purposes only and reflects analysis of publicly available on-chain data as of the dates indicated. It does not constitute legal, financial, compliance, or investment advice, and nothing in it is a recommendation to buy, sell or hold any token or asset, or to use or avoid any exchange, wallet, bridge, protocol or service.

The findings describe transactions and balances observed on Ethereum and Hyperliquid on 16 November 2026. Attributions come from on-chain behaviour and Bitquery's own labels; they may be incomplete or incorrect and may be revised as more data becomes available. Lighter has not published its incident report, and its findings may differ from those described here.

References to Lighter, Circle, CoW Swap, Across, Relay, LI.FI, Hyperliquid, Kava or any other named project, company or service describe what the public record shows and what those parties have said in public. They are not statements about any party's intent, security practices, solvency or conduct, and nothing here asserts that any named party acted unlawfully or negligently or knew of the events described. The wallets that received and moved the funds are described by their behaviour and attributed to nobody.

Nothing herein should be relied upon as a definitive determination of fact. Readers should conduct their own independent verification before taking any action. The authors and publisher accept no liability for any loss or damage arising from reliance on this material. All trademarks and company names are the property of their respective owners.

Reported by Divyasshree for Bitquery Research, with AI tools; every figure was checked against the raw data.

FAQ

What happened at Lighter?

Lighter says it is looking into a security incident that appears to involve one compromised outside party. Posts on X say the API keys of two market maker accounts leaked and the accounts were drained. On Ethereum, one wallet withdrew 10.39 million USDC from Lighter's contract on 16 November 2026.

How did the Lighter incident happen?

Lighter has not published its findings. It says the incident appears to involve one compromised outside party. Posts on X say the API keys of two market maker accounts leaked. We could not verify how the funds were moved, because that happened inside Lighter's rollup.

How much was taken in the Lighter incident?

Posts on X put the loss at about $15.5 million from two accounts. On Ethereum, we saw one withdrawal of 10.39 million USDC. The second amount, reported at $5.09 million, had not left Lighter when we last read the chain.

Where did the money go?

The wallet swapped the USDC to 5,764 ETH on CoW Swap. It sent 1,600 ETH through ten new wallets to Hyperliquid, where each one bought XMR1. The other 4,165 ETH was still in the wallet after the tenth payment, at 11:36 UTC on 16 November 2026.

What is XMR1?

XMR1 is a token on Hyperliquid named after Monero, a privacy coin. The ten accounts bought 5,587 of it for about $2.88 million, and held 5,583 after fees. We did not check what backs the token.

Is the Lighter incident linked to the Kava hack?

We do not know. Both used LI.FI, splits across fresh wallets, and XMR1 on Hyperliquid, two days apart. We found no transfer between the two sets of wallets. Anyone can copy a route.

Are Lighter withdrawals paused?

Yes, when we wrote this. Lighter paused all withdrawals and the creation of new pools on 16 November 2026 while it looks into the incident.

Why did fake ETH transfers show up in the wallet's history?

Address poisoning bots copied the wallet's 160 ETH transfers with fake tokens, most of them sent to look-alike addresses. The first copy came 12 seconds after the first real transfer. None of the real payments went to a fake address.

Run the next investigation in minutes

Ask any question of the blockchain. Bitquery's AI Investigations agent traces funds, attributes wallets, and writes auditable reports across every major chain. Prefer to do it yourself? Connect the Bitquery MCP server to your AI assistant, pull the raw datasets, or query the chain directly with the Bitquery API.