Crypto wash trading on six chains: $144 billion of DEX trading failed our tests.

01 · A Thursday on SolanaBoth sides, one transaction, eight dollars

At noon UTC on Thursday, August 27, a Solana wallet paid 10.45 SOL for 20.2 million units of a token called UMIA, in a PumpSwap pool. In the same transaction it sold 20.1 million units back to the same pool for 10.37 SOL. The wallet signed both halves. It spent about 0.08 SOL, roughly $8, and our data recorded about $2,170 of trading.

The wallet belongs to a group of 50 look-alike wallets from our Solana story. In that one minute, 44 of them made 116 of these round trips in UMIA. Over about a day the group made 164,419 of them in the token, almost all the UMIA trading we priced. Who created UMIA we did not trace.

Nobody else has to be on the other side. The trading still counts as volume.

02 · The questionWhat volume is, and how much of it was not real

Volume is the dollar value of every trade. Trackers and screeners rank tokens by it, so a token with big volume looks busy and easy to trade. Wash trading means trading with yourself, or between wallets you control, to make a market look busier than it is. Our investigation of Base explains why people pay for it.

We tested every DEX trade we could price on six blockchains over the same 30 days. Between August 24 and September 22, by our tests, $144.1 billion of the trading in the DEX pools we index was not real. Solana had $117.7 billion of it, 58.4% of its volume. On the other five together the share was 19.0%, or $26.5 billion. About four-fifths of the total was the move in that UMIA transaction.

On Solana we could run fewer of our checks than on the other five, so its share is the least tested.

03 · By chainFrom 58.4% to 1.5%

Pooled, the six come to 42% of the $340.3 billion we index. That average leans on Solana, the network with the fewest checks, and our index holds 1.7 times the volume DefiLlama reports for the same six. It does not measure crypto trading as a whole, so the figure for each network is the one to use.

04 · The movesSix moves, and where we found them

The one-transaction round trip made most of the dollars, almost all of it on Solana. The other moves were smaller. Not every test ran everywhere. We looked for contract fleets and points farming only on BNB Chain, and could not test for buying from one's own pool on Solana.

On BNB Chain the biggest single piece was one fleet of wallets sending trades through one contract, from our BNB Chain story. Points farming on one Binance Alpha token was another large piece. Robinhood Chain had the hand-off line of one-trade wallets, a move we found nowhere else. On Base most of it was wallets buying from pools they had filled themselves.

05 · WhoOne group can make most of it

On Base, Arc and Arbitrum, one wallet or a small group made most of the not-real trading. On BNB Chain and Robinhood Chain the biggest group made about a third, and on Solana about a sixth.

On Arc the two wallets are the same two as in our story on Arc's first days. On Arbitrum the one wallet is a bot trading the USDai stablecoin, and we have no evidence linking it to the stablecoin's issuer. The BNB Chain contract also carried the $33.8 billion wallet we set apart, which is not in its share. We do not add these groups up. Our counts are of signing wallets, and one person can run many.

06 · Set apart$73.5 billion left out of both sides

Our totals leave out $73.5 billion of recorded trading altogether. We count it neither as real nor as not real. It is loops far bigger than the pools they pass through, and trades at a price no one else paid. Counting either would drown out everything else.

Two wallets made 93% of it. One is the $33.8 billion loop wallet in our BNB Chain story. The other recorded $34.3 billion in six minutes on Robinhood Chain in a token named COBIE. The name says nothing about anyone called Cobie, and our Robinhood Chain story covers the token.

07 · Trade-count botsOne key, three chains, 3.9 million buys

Some bots pad the number of trades and leave the dollars alone. We found 123.6 million such trades, 90% of them under $5 and 87% of them on Solana. They add almost nothing to the dollar totals above.

One address led these bots on BNB Chain and Arc and was second on Robinhood Chain. On the days it led, it made 3.9 million dust-sized buys in 11 tokens, a quarter of all such trades outside Solana. It signs its own transactions, so one key runs it on all three. A second address took turns with it on some of the same tokens. We do not know who holds the key, and nothing shows it created or owns any of the tokens.

08 · BuyersNo single answer on who lost

Outside buyers fared differently from one network to the next, and by the kind of bot. Where trade-count bots pumped a token, sellers on BNB Chain and Arc did worse than in comparable tokens. Where bots churned a token back and forth, they did about as well at the median, though on Arc more of them lost money. On Robinhood Chain, across all flagged tokens, more sellers lost money too.

These results cannot be added into one loss for all six. They cover only buyers who sold out, and many had not sold by September 22.

09 · Against DefiLlamaOur index is bigger, and the rest comes close

DefiLlama reports $203.1 billion of DEX volume for the same six networks and days. Our index holds 1.7 times that, mostly because of Solana. Take out what our tests flag and $196.2 billion remains, about 3% below DefiLlama's total.

The match is partly chance. On Solana DefiLlama counts venues we do not index and drops pools we keep, and on Base, Robinhood Chain and Arbitrum we index less volume than it does. Network by network, what remains runs from 0.8 to 1.7 times DefiLlama's figure.

10 · What we cannot sayThe limits

• A share of all crypto trading. The base is the pools we index, which hold more than DefiLlama counts.
• A trend. We measured one 30-day window.
• A combined loss to buyers. The results above do not add up into one figure.
• Who is behind it. We name addresses only, and do not add wallet counts together.
• Solana in full. Some tests could not run there. Counting routed swaps as real would bring Solana to about 57%, and counting round trips our rules missed would lift it to about 62%.
• Optimism. It was not part of this count. We measured it separately, from raw pool events, in our Optimism story.

11 · QuestionsWhat readers ask

What is wash trading in crypto? Trading with yourself, or between wallets you control, so a token looks busier than it is. Our count also takes in bots that churn one token all day and wallets farming exchange points.

Is most crypto trading fake? On five of the six networks, well under half of the volume we index was flagged. Solana was the exception, at 58.4%.

Which chain has the most? Solana, in dollars and as a share of its volume.

Does DefiLlama count this trading? Some of it. DefiLlama drops some of the pools where this happens, and our index is bigger than its figures on Solana, BNB Chain and Arc.

How can I spot it? Each move leaves its own tell, and each story in this series shows one.

Our Base investigation lists four more signs that a volume number is inflated.

12 · How we measuredWhere the numbers come from

We checked every DEX trade on Solana, BNB Chain, Base, Arc, Robinhood Chain and Arbitrum from August 24 to September 22 that we could put a dollar value on. A trade counts as not real when a wallet buys and sells the same token at bot speed and gets back about what it paid, when a few such wallets make most of a token's trading, or when a wallet trades against its own liquidity. On BNB Chain every trade by the fleet of wallets working through one contract counts, and on Robinhood Chain so does the hand-off line our story there describes. Trades on token launch curves we left out on BNB Chain and capped on Solana and Robinhood Chain, and we dropped duplicate records. Bitquery's own DEX data records these swaps like any other trade; the figures here come from pulling them apart. We use one term, not real by our tests, for all of it, though some of our earlier stories call their parts fake.

Our earlier pieces measured narrower patterns. The $108 billion in our Base investigation counted one kind of no-money round trip on Base and Arbitrum, almost all of it before this window. Our counterfeit SOL story priced counterfeits against each other, which this count leaves unpriced. Our Arc story found two wallets behind about half of Arc's volume on September 20. They are the two in our table, and across the whole window their trading came to about 7% of Arc's volume, almost all of which fell in the window's last eight days.

13 · The recordThe addresses behind the story