74% of ordinary Solana buyers lose money on memecoins

Most ordinary Solana buyer wallets lose money on new memecoins. Bitquery followed about 2.9 million ordinary wallets that bought new coins on Solana launchpads such as Pump.fun, in one launch week a month from November 2025 to August 2026. Two weeks later, 74% had less SOL than they put in, after fees.

You would not guess that from your feed. A post went viral claiming a five-agent AI trading bot turned $3,000 into $12,500 overnight by sniping new Solana coins. We are not naming the account, and we cannot check the claim. What we can check is what happens to everyone else.

This is education, not financial advice.

Finding 1: the typical buyer loses a little

Bitquery counted a wallet as an ordinary buyer if it paid its own fees, made fewer than 100 swaps on its busiest day and was not a known bot or the coin's creator. Across those wallets:

• 74% ended with less SOL two weeks after buying, after fees.
• The median buyer lost $13. Half lost more, half lost less or came out ahead.
• Together they lost about $269 million in the 10 launch weeks Bitquery followed.

A public Dune dashboard by the analyst web3frank looked at about 7 million wallets that traded Pump.fun tokens from November 2025 to May 2026. Its median wallet ended up roughly flat, with a few cents of realized profit for six months of trading, and only about 44% to 49% of its wallets were profitable. The two datasets agree that the typical trader makes roughly nothing. They measure it differently: Dune counts only realized (sold) trades, while Bitquery counts unsold coins as worthless.

Finding 2: the gains go to very few

Some buyers do win, and that is the point. Bitquery found the top 1% of ordinary buyers took 56% of the money won. Most of the rest went to the next 9%. Nine in ten buyers shared what was left, and most of them lost.

The steadiest winners were not ordinary buyer wallets at all. Sandwich bots, which trade just before and just after a buyer to take a slice of each trade, came out ahead in every launch week Bitquery checked.

Finding 3: launch insiders beat the sniper bots

A separate Bitquery investigation looked at who buys a Pump.fun coin in its first moments, for launches from October 30, 2025 to June 30, 2026.

• Sniper bots: 25,882 wallets that buy new coins within a fraction of a second. They spent $440 million, and made $31.5 million on what they sold. Counting the coins they could not sell as worthless, they made $15.2 million, and only 54% of sniper wallets came out ahead.
• Launch bundles: wallets that a coin's creator packs into the launch itself, so they buy before anyone can see the coin. Creators and their bundled wallets made at least $579 million, 18 times or more what the sniper bots made.

So even fast, automated sniping is close to a coin flip for the wallets that try it. The surest money sits with whoever controls the launch.

Why "my bot made 4x overnight" posts mislead

This is survivorship bias. You see the wallet that won, not the thousands that lost on the same coins the same night.

• The winners are real, and rare. When the top 1% take 56% of the money won, there will always be a screenshot of someone doing very well.
• The losers do not post. The median buyer lost $13, which makes no headline. The total was about $269 million across about 2.9 million wallets, much of it from big losers: buyers who lost $1,000 or more account for 46% of all losses, Bitquery found.
• One night is not a strategy. A single good run says little about what the same bot does over months, after fees and after the coins it could not sell.
• Bots are not a shortcut. Bitquery's sniper bots are exactly the kind of fast, automated buyers these posts describe, and close to half of them did not come out ahead.

The limits

• Bitquery and Dune count wallets, not people. One person can run many wallets.
• Bitquery followed each buyer for the launch week plus 7 days and counted unsold coins as worth nothing. It left out tips and trading-app fees, so it says real losses are a little larger.
• The Dune figures come from one analyst's dashboard and count realized profit only, for wallets that both bought and sold. Its tables can update, so we describe the median as a few cents rather than an exact figure.
• The two Bitquery studies cover different windows, and none of these figures predicts future results.

Sources

• Bitquery, 3 in 4 Solana memecoin buyers lose money, $269 million in 10 launch weeks. bitquery.io
• Bitquery, Pump.fun devs' bundles made $579M+, snipers just $31.5M. bitquery.io
• web3frank, Pump.fun 6 months trader analysis (Dune dashboard). dune.com