WIRETAP

Daily investigation · pump.fun · today

Is anyone actually buying this?

Every day we investigate the 24 most-traded pump.fun coins. We follow the fees on Solana: real people pay their trading apps, bots don't. Real coins get stamped. Fakes get exposed.

Coins investigated
—
Exposed as fake
—
Humans inside
—
Play today's line-up

Opening today's files…

who's actually
paying the fees? same wallets
again?? check
CASE FILE
SubjectNo. 0000
$—

MARKET CAP —

VOLUME 24H —

FEES / TRADE —

WALLETS SEEN —

Finding: —

FAKEbot-pumped · last 24h

Today's evidence wall

Caught today

The photos are today's suspects. Every red thread ties a fact to the coins it is true for; green threads lead to real people. Point at a note to follow its threads, click a photo for the full file.

← drag the board →

Today's case files

Who's really in there

Every coin we checked in the last 24 hours. Prices are live. Open a file for the full report.

The daily game

Two suspects. One is real.

The ranked line-up: two coins at a time, one has real people inside, the other is bots. 10 seconds per case, 3 strikes, one life a day. Every 10 minutes the best run wins the round.

DEPT. OF ON-CHAIN EVIDENCETHE LINE-UPsuspects waiting

Today's suspects are being brought in…

Enter the room →

The archive

Case closed. Were we right?

Every verdict we gave on earlier days, and what happened to the coin since. We keep the ones we got wrong too.
Memo · MethodInternal use: none. Public.

How we tell people from bots

  1. Which coins. Every night from 00:20 UTC we take the 24 pump.fun coins with the most trading volume in the last 24 hours, among those that graduated in the last 14 days and still have at least $50K of daily volume, $100K market cap and $10K liquidity. A known family of copy-paste "oil fund / dividend" coins is skipped.
  2. We read the chain. From all the successful transactions of the last 24 hours on the coin's pools, we pick 300 at random and read them straight from Solana.
  3. We follow the money. People trade from apps like Axiom or Photon and pay them a commission on every trade (about 1%). Volume bots trade with themselves and pay almost nothing. For each coin we measure the fees paid per trade (network fee, tips and app commissions), the part that goes to trading apps, and how many different wallets made the trades.
  4. We stamp it. FAKE if any one of these is true: fees under 0.8 mSOL per trade; app commissions under 0.49 mSOL per trade; different wallets fewer than 12% of the trades; or, on a busy coin, fewer than 20 real trades among the 300 transactions. HUMANS INSIDE only if all three are high: fees from 1.5 mSOL, app commissions from 0.98 mSOL, wallets from 19.5% of the trades. Anything in between is too close to call and is not published. When we could read fewer than 100 trades, the FAKE bar is lower (0.6 and 0.35 mSOL), so noise does not convict a coin. Coins with fewer than 200 transactions in 24 hours are not judged.
  5. The launch. We also read the first 10 seconds: what share of the supply was bought, and by how many wallets. If the launch transaction is out of reach, we write "unknown", never a guess.

A verdict is about the last 24 hours of trading, not about the people behind a coin or its future price, and it can be wrong: every past call stays on the site with how it aged ("Case closed"). Market cap and volume on the files update live from DexScreener every 30 seconds; the verdicts do not change during the day. The Wanted posters come from our own on-chain studies (PumpScan, November 2026): an address and what it did, never a name. We never change a coin's name: a coin is called what its creator called it. We withhold a coin's picture only when it is explicit, carries a slur, or accuses a real person, and the file says so. mSOL = a thousandth of a SOL. Not financial advice.

VERIFIEDon-chain · daily