Bitquery Traces $1.4B of Raydium CLMM Volume, 22.7% of the Total, to One Wash-Trading Wallet

One Solana wallet produced $1.40 billion of the $6.17 billion traded on Raydium's concentrated-liquidity pools between 21 September and 20 October 2026, or 22.7% of the total, according to an investigation published by Bitquery with figures verified on 21 October. Bitquery says the wallet did it by trading back and forth with itself across 438 pools of counterfeit "major" tokens, such as fake XRP, ETH, LTC and XMR, on about 12,000 USDC and 1.3 SOL of working capital.

Bitquery reports no stolen funds, and the activity is not a hack of Raydium RAY$1.84-7.7%'s code. The finding is about measurement: how about $12,000 can make pools on Solana SOL$103.47-4.4% display billions in liquidity and generate more than a fifth of a major pool program's reported volume. Bitquery says it does not know who runs the wallet, and Raydium had not publicly responded in the coverage we reviewed.

How the Raydium CLMM wash-trading bot works

Bitquery describes a closed loop. The operator creates tokens that borrow the tickers of well-known assets, opens a Raydium CLMM pool for each one against a small amount of USDC, then buys and sells its own token in that pool. CLMM is Raydium's concentrated-liquidity market maker, and pool creation on it is permissionless: anyone can open a pool for any token pair.

Bitquery counted 108,871 swaps from the main wallet, forming 22,685 completed round trips. The median gap between a buy and its matching sell was 20 seconds, and each trip cost the bot about $1.13 plus two network fees of 0.00002 SOL, per Bitquery. Twelve helper wallets added another 1.25 million swaps worth $8.8 million.

The tickers fall into three groups in Bitquery's breakdown:

• exchange-listed coins, including XMR, LTC, HBAR, TAO, XRP, XLM, ADA, TRX, BNB, ZEC, LSK, BTC and ETH;
• stablecoins named after deposit networks, such as TRC20-USDT, ERC20-USDT and BEP20-USDT;
• recent hype names, including HYPE, PUMP, PAXG, RLUSD and TRUMP.

None of these is the real asset, and the prices do not track the real ones. Bitquery says fewer than one in eight of the fakes whose ticker matches a Binance market traded within 5% of the real price. Its fake XMR traded at $547 against $119 for XMR on Binance, about 4.6 times higher.

Bitquery says the 22.7% share "holds up because the volume is real USDC changing hands." The swaps settle in real USDC and pay real fees. Bitquery's case for calling it wash trading is that the same small pot of capital goes in and out of the pools thousands of times, with the bot on both sides.

Billions in displayed liquidity backed by about $10,000

The displayed-liquidity gap is the sharpest number in the report. Bitquery says the ten largest pools in the cluster showed $5.93 billion of liquidity on DEX Screener while holding $10,238 of USDC between them. A single fake LSK pool displayed $579 million of liquidity on $19.27 of USDC.

The gap comes from how the figure is built. In Bitquery's account, displayed liquidity counts the token side of a pool at the token's own traded price. When the bot holds a large supply and sets the price with its own trades, that side is worth whatever the bot says it is. Bitquery's measure instead values each pool by its quote-side reserve, the real USDC, and its filter picked out pools holding under $1,000 of measured reserve while printing thousands of dollars of volume an hour.

Bitquery is careful about where the fault sits. It says DEXScreener and Jupiter JUP$0.311-7.7% report accurately what they measure, and that what they measure is a number the bot controls. Jupiter's token search returned one of the fake tokens, flagged as unverified.

What Bitquery's report shows and what it does not

The pattern Bitquery describes was built over months. The oldest pools, fake ADA, XLM and ETH, date from 22 to 27 March 2026. The first bot wallet was funded on 10 June, a private router program was carrying the main wallet's swaps by 31 August, and a token factory minted 93 tokens between 28 September and 21 October.

Almost nobody else traded these pools. In the final week of the window, four wallets outside the cluster traded one of the pools once each, for about $27 in total, according to Bitquery. So the report does not identify any victims, and it does not establish a motive. Bitquery calls the cluster "an inference from shared pools, shared gas funding and direct transfers," says no statement from whoever runs it exists, and notes its figures were measured on 21 October and may change. CryptoSlate's write-up of the findings also says they do not establish Raydium's involvement.

The 22.7% applies to Raydium CLMM alone, on Bitquery's own volume measure. It is not a share of all Raydium activity, which also runs through its standard AMM, CPMM pools and LaunchLab. Bitquery is a data vendor that sells a Raydium API, and this report is part of a running series of wash-trading investigations it publishes.

What it means for reading Solana DEX volume

This is the second time in a month that Bitquery has put a large number on non-organic Solana trading. In September, Bitquery flagged 58% of Solana DEX volume as circular or bot-like, with most of that activity on PumpSwap. The Raydium CLMM finding is narrower but more concrete: one wallet, one pool program, a named set of fake tickers and a dollar figure for the capital behind it.

The more useful question is what a headline volume number can carry. Volume counts every swap, including a wallet trading with itself, and displayed liquidity trusts the price a pool reports. Neither tells you how much independent money is behind a pool. Bitquery's method, valuing pools by their real quote-side reserves, is one way to strip that out.

The timing also bears on Raydium's Q3 2026 results, where net revenue rose 264% to $7.99 million. Bitquery's window overlaps only the last ten days of that quarter, and the fees the bot paid on each round trip were real fees. Bitquery does not estimate how much fee revenue the wallet generated.

You can follow the pool program's activity over the same 30 days on our Raydium CLMM program page.