Only 26% of the money in pump.fun's $100K wins was real trading profit. The rest went to hand-offs and wash loops

A $4,223 trade that turned into $4.92 million

Here is a pump.fun win that would sit near the top of any profit leaderboard. One wallet turned a few thousand dollars into almost $5 million in under an hour. The coin was called Arena. We traced where the money came from.

Almost all of that money came from one other address. That address did 99% of the buying while the winner was selling, and it never sold. In the two days before, the winner had sent that buyer more than 10,000 SOL, about $1.4 million. Everyone else who bought Arena put in about one percent of what the single buyer did.

Arena's creator was in on it from the start. It sent the winner 32 SOL, then opened a trading pool for Arena with just 0.01 SOL in it. The winner bought in the very next transaction. Six seconds after the last sale, the creator emptied the pool, and the winner sent 480 SOL back to the creator.

So one of the most impressive trades of the past ten months was a wallet selling to a buyer it had helped fund. Traders call this wash trading: trading with yourself so a coin, or a trader's record, looks better than it is.

We found the same pattern behind most of the biggest pump.fun wins. Then we checked every win of six figures or more, more than three thousand of them. Only about a quarter of the money they show was real trading profit.

How a pump.fun win gets counted

Pump.fun is a Solana app where anyone can launch a memecoin in a minute. New coins trade on a simple price curve. Coins that take off move to PumpSwap, pump.fun's own exchange, where they trade in a pool. Anyone can also open a PumpSwap pool for a coin made somewhere else.

Profit trackers and leaderboards score each wallet on each coin. They take what an address got from selling and subtract what it paid to buy. That works for a normal trader. It falls apart when the seller never paid for its coins, or when the buyer on the other side belongs to the seller.

We used that same score to find every wallet that made six figures or more on a single coin. Then we asked three questions about each win. Who bought from the winner? Where did the winner's coins come from? Did any of the wallets involved pay each other?

Real trading profit came to about a quarter. Wallets that were handed their tokens took half, and wash loops took 13%. The last 10% we could not prove either way.

Trick one: the wash loop

A wash loop needs a coin, a thin pool and two wallets with one owner. The seller gets in first: in nine of every ten loops we found, the seller's first buy came within ten seconds of the coin's first trade. Then the buyer soaks up most of the coin, while the seller sells into that buyer and books what looks like a huge profit. Somewhere along the way, money moves straight between the two wallets or through the coin's creator.

We counted a win as a wash loop when two things were both true. One address did at least half of all the buying while the winner held the coin. And that buyer had sent money to, or received money from, the winner or the coin's creator. 278 wins passed both tests. In those coins, the dominant buyers paid in more than four times as much as all other buyers together.

How many loops you count depends on where you draw the line. If one buyer must do three quarters of the buying, the count falls by about half. But in 13 of the 15 loops among the 25 biggest wins, one buyer did more than 90% of the buying.

The loops come in clusters. They run through 46 groups of linked wallets. In the largest group, one buyer took the other side of 38 separate wins. All 278 loops had just 60 different buyers between them.

The loops also favor one shortcut. Coins listed straight on PumpSwap skip pump.fun's launch page, so whoever creates the coin can open its pool with almost no money in it. Those coins make up 12% of the big wins but 25% of the wash loops. Arena was one of them.

Trick two: the hand-off

The second trick needs no fake buyer. Someone holding a big bag sends tokens to a second address. The second address sells them into the market. On a profit tracker, that address paid nothing and made a fortune.

Half the big wins work this way, and they hold half the money. We followed each seller's tokens one step back.

Most hand-offs, 1,104 of them, lead back to the people who made the coin. Half the time, the creator's own address sent the tokens. In most of the rest, the creator sent them to a middle address first, which passed them on. A smaller group came from someone who had bought the tokens on the market, so the sender and the seller together made one ordinary trade, split across two wallets. The rest came out of accounts run by code, such as vesting locks, from exchanges and bridges, or from wallets we could not trace past one step.

Some hand-offs add an extra layer. In 423 creator-side hand-offs, the transfer was signed by an address other than the sender and the seller. On Solana, every wallet that moves tokens in a transaction has to sign it. One transaction in a coin called UOTF shows how this works. A single address and four throwaway wallets sign it together. Each throwaway receives nearly eight million tokens, passes them straight to a seller, and is then closed, all in that one transaction. Only someone holding all five keys, or a group working together, can sign that transaction. One of those sellers later sold once for $1.59 million.

Hand-offs took over in July

Both tricks surged this summer. Wash loops came in two bursts, one last November and a far bigger one in August, with none at all from January to June. Hand-offs stayed under 100 per month until July, then took over. In August they made up seven in ten of the big wins.

Big wins kept coming after August. Another 2,943 six-figure wins ended in the weeks after our transfer records stop on 30 August. That is almost as many again as this whole study, and none of them is in any number here.

Who really got rich on pump.fun

Only 967 wins pass every check. The winner bought the tokens it sold. No single buyer carried the buying. No transfer ties the winner to the other side of its trades.

The best real trade in our ten months of data is SOULGUY. A trader bought into a coin that was already more than two months old and made $3.30 million over about two and a half days. It traded roughly once a minute, so software probably did the clicking.

Two more names on the list are high-speed trading bots, with tens of thousands of trades or more. Four others were in and out in under four hours. Fast trades like these often come from snipers, or from launch bundles, where a coin's creator packs buys into its first block. We covered both in our look at pump.fun launch bundles and snipers. Passing our checks means the trade was real. Whether the winner was an outsider, rather than someone close to the launch, is a separate question.

Big real wins are rare. Most pump.fun buyers never see one: our study of Solana memecoin buyers found that about three in four lose money.

Why stage a win at all?

The chain records what these wallets did. Their reasons are a guess, and none of the ones below is proven here. An address with a giant PnL attracts copy traders and paying followers. Big wins and big volume push a coin onto trending lists. Selling through other wallets keeps a coin's "dev sold" warning from lighting up on trading apps. And passing money through a coin can blur where it came from.

The two tricks share people. 22% of the wallets in wash loops also sold, sent or signed hand-offs. We have seen the same moves on other chains, in our wash trading study across six chains.

How to check a pump.fun win yourself

You can test any big win with four questions, using a block explorer or the tool below.

• Who bought from the winner? If one buyer took most of what the winner sold, be careful.
• Where did the winner's tokens come from? An address that received its tokens by transfer did not trade its way to that profit.
• Do the wallets pay each other? Direct SOL transfers between the winner, the buyer and the creator are the clearest tell.
• How deep was the pool? A pool that opened with almost no money in it makes a staged trade cheap.

How we counted

We started from every address that made six figures or more on one coin launched on pump.fun or listed straight on PumpSwap, over the ten months to 30 August 2026. We left out coins with fewer than 500 trades. For each win, we looked at who bought from the winner while it held the coin, where the coins it sold came from, and whether the wallets involved paid each other. We checked more than ninety cases against the chain itself. A separate rebuild of every count, written from the rules alone, matched every one. An account that pays other people's trading fees, and that our records wrongly credit with their trades, was left out.

FAQ

Who made the most money on pump.fun?

The biggest genuine win we found was $3.30 million, on SOULGUY. Bigger wins show up on leaderboards, such as Arena's, but our checks show most of the largest ones were wash loops or wallets selling tokens they were handed.

Are pump.fun profit leaderboards accurate?

They treat any address that sold for more than it paid as a winner, however it got its tokens. Across every pump.fun win of six figures or more, only about a quarter of the money was profit from real trading. The rest went to wallets that were handed their tokens, to wash loops, or to wins we could not prove either way.

What is wash trading on pump.fun?

It is trading with yourself to fake profit or volume. In the pattern we tested, one buyer takes most of what the winner sells, and that buyer has direct transfers with the winner or the coin's creator. We found 278 such wins.

What is a token hand-off?

Someone who holds a lot of a coin sends tokens to a second address, and that address sells them. Profit trackers see a seller that paid nothing and made a fortune. In most hand-offs, the tokens traced back to the wallets that launched the coin.

How much of the profit in big pump.fun wins was real?

About a quarter of it. Of every four dollars that profit trackers would show as a big pump.fun win, roughly one came from a wallet that paid for its coins and sold them to other traders.

Why does this study stop on 30 August 2026?

Our records of token transfers end on that date, and every check here needs them. Another 2,943 big wins ended later and are not included.