Robinhood Chain wash trading: 26,000 wallets faked $322.5 million of memecoin trades.

01 · A Wednesday on Robinhood ChainOne wallet buys, the next one sells

At 16:54 UTC on Wednesday, August 26, a brand-new wallet on Robinhood Chain received 1.91 ETH. Ten seconds later it bought 1.9 million GOOSE, a memecoin. Sixteen seconds after that it sent every one of those tokens to another fresh wallet, followed by its leftover ETH. Its whole history is those four transactions, 34 seconds long.

The second wallet sold the GOOSE 41 seconds after the first wallet had bought it. Nine seconds later it sent all its ETH to a third fresh wallet, and that one bought GOOSE too. None of the three traded GOOSE a second time in our window.

The ETH that started this came from a wallet that had sold a different memecoin, a HOOD meme token, two hours earlier. This line had crossed from one token to another.

Robinhood Chain is the blockchain Robinhood built and runs itself. Stock tokens trade on it, and so do memecoins, which anyone can launch in about a minute. For a count of fake trading on six chains, we read every DEX trade on the Robinhood Chain venues Bitquery decodes, from August 24 to September 22.

In eight memecoins we found about 26,000 wallets trading like this, $322.5 million in all. We matched the buyers to the sellers and traced the ETH between them. We also found two plainer fakes: round trips that cost next to nothing, two thirds of them in pools set to charge no fee, and one token that recorded $34.3 billion in six minutes.

Leaving that token aside, fake trading was 2.9% of Robinhood Chain's DEX volume in the pools we index, the second-lowest share of the six chains we measured. Some of it took real planning.

02 · Why a lineA crowd that is really a queue

A memecoin with thousands of different wallets trading it looks alive. Screeners count holders and traders, and a long list of names reads as real demand, in a market where thousands of new memecoins launch every day.

Wash trading means trading with yourself to make a token look busy, and the usual test for it looks for a wallet that buys and sells the same token. The line gets past that test and past the screeners. No wallet in it buys and sells the same token, and nearly nine in ten make a single trade. Every hand-off adds two new names to the token's list of traders.

In September we showed how one farm of 5,500 wallets fakes a crowd for token after token on BNB Chain. On Robinhood Chain most of the crowd is new. About 2,700 wallets did come back to trade a second token in the line, almost all of them a day or more later.

03 · The hand-offBuy, pass, sell, pass the ETH

Each step in the line we traced was the same four moves. A wallet gets ETH and buys the token. It sends the exact amount it bought to a new wallet, with the ETH it has left. The new wallet sells, then sends all the ETH to the next wallet in line, which buys.

We saw that hand-off in the transfers of six of the eight tokens. To measure it everywhere, we took every wallet that bought one of the eight tokens once and looked for a later wallet that sold exactly the same amount. We found one for 98.5% of them. The median gap between the buy and the matching sale was 41 seconds, and more than nine in ten came in under a minute. Three in four took between 40 and 45 seconds, and none took less than 30.

A gap that regular points to a script. Nobody buying a memecoin for themselves hands it to a stranger who dumps it 41 seconds later.

04 · Eight tokensThe line ran through eight memecoins

Eight tokens, one pattern. In each, most of the volume came from wallets that traded one way only, mostly once, and the buyers and sellers traded the same amounts. The line made 87.7% of the eight tokens' trading. Three more wallets traded both ways in the same tokens, at almost no profit or loss, and with them the fake trading came to 99.8%.

The names say nothing about who is behind them. The HOOD meme token is a memecoin that borrows the ticker. It is not a stock token for Robinhood Markets shares, and we found nothing linking it to Robinhood. AnsemCat borrows a name too. We found no link to anyone called Ansem. Whoever deploys a token picks its name.

05 · The moneyThe same ETH, passed down the line

Fresh wallets need ETH to trade and to pay gas. We took the first 50 wallets in the line to trade after 16:40 UTC on August 26, all of them in BLORB and GOOSE, and looked up where each got its first ETH. For 48 of the 50, it came from another wallet among those 50. The other two were funded by wallets outside our sample, and one of those was the HOOD seller from the opening.

So the line funds itself. The ETH from one sale pays for the next buy, minus a little gas at each hop. In our sample each line stayed inside its token. At least two ran side by side, one in GOOSE passing about 1.9 ETH at a time and one in BLORB passing 4 to 4.5 ETH.

Who put the first ETH into the line, and who takes it out at the end, we have not traced. The trail can be followed with the Bitquery MCP.

06 · Round tripsThe bigger fake costs next to nothing

The biggest fake in our totals is a simpler one. A wallet buys and sells the same amount inside one transaction and ends where it started. When the pool charges no fee, that costs next to nothing, and it can make a pool's volume as big as anyone likes. We counted $947 million of these round trips.

Two thirds of it ran in three pools created on September 11 by one address, e4147aae. All three were set to charge a fee of zero. Two trade ETH against USDG, a dollar stablecoin, and one trades a GOOGL stock token. In the biggest, ten wallets made their first round trip in the same second and recorded $446 million in under nine hours.

The stock-token round trips came to $346 million, led by NVDA, GOOGL and SPY. We found nothing that ties the tokens' issuers to these trades. Anyone can create a pool for any token, and the GOOGL pool above was created by e4147aae.

In September we found a hook on the same chain that shows a 0% fee and takes 1%, a cut the trade record never shows. A hook is a small contract that runs on every swap, and the three pools above have one each. In the round trip we checked in each pool, the wallet paid only dust. In the biggest, it bought 191 ETH with 473,679 USDG, sold it straight back and paid 0.000002 USDG. One wallet we checked also sent up to about $20 of ETH with some of its round trips, and our data cannot show where that ETH went.

07 · The $34 billion tokenSix minutes, 0.001 ETH

On September 2 one wallet created a pool for a token called COBIE, with no fee, and put in 0.001 ETH and 0.001 COBIE as liquidity. Five minutes later it began trading against it: 701 transactions between 09:23 and 09:29 UTC, each a buy and a sell, recorded at $34.3 billion.

That is more than two thirds as much as all the other DEX trading we recorded on the chain in our 30 days. Uniswap V4, the DEX these pools run on, settles what a wallet owes only at the end of each transaction. A buy and a sell inside the same transaction cancel out, so the wallet never needed the ETH its trades recorded.

We set COBIE apart from every total, fake and real, because one number that size would drown out the rest. Why anyone would do this, we do not know. A volume ranking or a listing rule would fit, and nothing on-chain proves either. The token's name belongs to whoever deployed it and says nothing about anyone called Cobie.

08 · BuyersWho else was trading

Almost no one else touched the eight tokens. Outside the line and its three helper wallets, 353 wallets traded them in our 30 days, about $117,000 between them. Next to the tokens' $367.7 million, that is close to nothing.

We usually check whether buyers of tokens with fake trading did worse than buyers of similar tokens. The eight ring tokens were left out of that test. Their buyers are the line. Across the rest of Robinhood Chain the result was mixed. Buyers who sold out of tokens with fake trading lost money more often than buyers of comparable tokens (60% against 54%), but the typical result was about the same, a loss of about 2%.

09 · How to spot itThree signs of a line

Each of these can be checked from public data.

• The buyers and sellers pair up. One-trade buyers and one-trade sellers in matching numbers and matching amounts are one crowd passing tokens along.
• Each wallet was funded by the last one. When one-trade wallets keep getting their first ETH from other one-trade wallets in the same token, it is one group moving the same money.
• The timing is too regular. Real trades do not pair up 41 seconds apart, time after time. A pool's trade history shows it.

10 · How we measuredWhere the numbers come from

We read every DEX trade on the Robinhood Chain venues we decode, from August 24 to September 22, and valued each from its ETH or stablecoin side. A token counts as a ring token when its one-trade buyers and one-trade sellers match in number and in the amounts they traded. In each token we then matched each one-trade buyer to the first later one-trade seller of exactly the same amount. The funding check and the traced line come from transfer data, read on September 23.

Round trips count as near-free when a wallet buys and sells the same amount and the trade records show no loss on the ETH or stablecoin side. Bitquery's own DEX data records all of these trades like any other; the figures here come from pulling them apart. The fees of the four pools we name come from their creation events. COBIE is set apart by the rule we use on every chain for trades at a price no one else paid. The venues we decode carry about 92% of the volume DefiLlama reports for the chain.

11 · The recordThe addresses behind the story