Printing…
Reading the wallet's Pump.fun trades. Big wallets take up to a minute.
Opinion piece with sources
Our view: Pump.fun leans too hard on a handful of big accounts, while the small and everyday traders who pay the fee on every click are left to fend for themselves. Below is the evidence we could find, with its limits, and what we think should change.
Every figure links to its source. They come from third-party reports and Dune-based analyses, use different filters and dates, and are not directly comparable.
98.6%
of tokens launched on Pump.fun showed rug-pull or pump-and-dump patterns
Solidus Labs via BeInCrypto · May 2025 · 7M+ tokens, Jan 2024 to Mar 2025
Solidus's own definition of fraud signals. Dated.
<100k
of those 7M+ tokens ever held more than $1,000 of liquidity
Solidus Labs via Forklog · May 2025
Same report and window.
0.4%
of wallets realized $10,000 or more in profit (55,012 of 13.4M)
Decrypt, from Dune data · 2025
Realized profit only. The founder disputes the method.
294
wallets realized more than $1M, about 0.002% of all wallets
Decrypt, from Dune data · 2025
Realized profit only.
56.6%
of 4.26M addresses that traded 10+ tokens in six months ended with losses between $0 and $1,000
BeInCrypto, from Dune data · 2025
Cumulative realized PnL. Different filters from the stats above.
$834M+
total revenue reported for the platform, paid as trading fees
Blockworks · 2025
Figure grows over time; check the date.
Pump.fun's founder has disputed the Dune-based figures. His argument: they leave out purchases made after tokens moved to the AMM, and they ignore unrealized gains, so the true share of profitable wallets could differ.
A later CoinGecko analysis of Dune data (April 2024 to April 2026) found the share of profitable traders among those who exited was 30.1% at its low in June 2025 and rose to 73.3% by April 2026, which CoinGecko links to the retail crowd leaving the platform.
We include this because it's relevant. It changes the size of the problem, not the asymmetry below.
Sources: Decrypt · CoinGecko Research
What we could not find
No public dataset splits PnL by KOL versus everyone else, and Pump.fun hasn't published one. So we make no numeric claim about KOL performance. That gap is one of our asks.
One fee for everyone
1.25% on every bonding-curve trade, buy and sell, win or lose. A normie sizing 0.5 SOL pays it exactly like a whale.
Attention is not equal
Pump.fun signed a trader as a creative partner and its co-founder amplified a promoted token from his own account. That reach is not available to a first-time trader.
The downside lands on holders
Creators earn a fee on volume whatever happens to the price. Holders who bought the promotion carry the loss.
This section is opinion. Creator-fee figures: Decrypt.
What public reporting says happened. Allegations of coordination or profit are unproven.
Read the manifesto and sourcesTwo days earlier
Pump.fun signs the trader Kimchi as a creative partner.
The post
A poster with a Pump.fun logo and a QR code opens a brand-new token, Project Kimchi.
The repost
Pump.fun co-founder Alon Cohen reposts it from his personal account.
Minutes later
Market cap reaches roughly $14–15M.
About 30 minutes
The post and the repost disappear. The token falls 95–98%.
After
At the time of reporting, neither Kimchi nor Pump.fun had given a substantive public response.
Publish PnL by wallet class
Per-wallet outcomes for partner and large accounts versus everyone else, with the method.
Disclose partner deals
Who is paid or promoted, and a clear label whenever a partner pushes a token.
A promotion policy
Rules for staff and partners posting launches, including a pause when a promoted token is already collapsing.
Say something
When a promoted token falls 95% in half an hour, holders deserve a public answer.
Our own numbers, honestly labelled
100%
of 5 wallets checked on Pumptax have a negative real PnL.
A self-selected sample of people curious about their fees. Not representative of Pump.fun.
Check your side of it
Pumptax is independent and not affiliated with Pump.fun or anyone named here. Figures summarise linked third-party reporting and may be outdated. Allegations are unproven; opinions are labelled. Nothing here is financial advice.