2026-10-23
Trench lessons — 2026-10-23
Key takeaways
- Bitquery (Oct 22): only 26% of pump.fun's $100K+ wins were real trading. 51% came from handed-over tokens and 13% from wash loops
- pump.fun $2.5M "Squads" PnL competition (announced Oct 23 ~10:03 AM CT): squads of up to 20, sign up by Mon Oct 26, top 3 by PnL after 2 weeks
- Platform watch: nothing breaks
Bitquery (Oct 22): only 26% of pump.fun's $100K+ wins were real trading. 51% came from handed-over tokens and 13% from wash loopsevidence: A: 3,170 wins, $752.5M
- What it says: Bitquery looked at every pump.fun or direct-PumpSwap win of $100K+ (Oct 30, 2025 to Aug 30, 2026, coins with ≥500 trades). It found $193.9M (26%) was real buy-then-sell-to-strangers profit. Wallets selling tokens handed to them took 51%; 1,104 of 1,579 hand-offs trace back to the creator side, and 423 were co-signed by a third address using throwaway wallets. Wash loops took 13%, and 15 of the 25 biggest wins were loops. Bitquery's loop test: one address did ≥50% of all buying while the winner held, and that buyer had direct transfers with the winner or the creator. 278 loops ran through only 60 buyers in 46 linked groups; one buyer took the other side of 38 wins. In 9 of 10 loops the seller's first buy came ≤10 s after the coin's first trade. Direct-PumpSwap (curve-skip) coins are 12% of big wins but 25% of loops, because the creator can open the pool with almost no SOL. Example (Arena): creator funds winner, opens a 0.01 SOL pool, winner buys next tx, sells into one funded buyer for $4.92M over 57 min, creator drains the pool 6 s after the last sale. Hand-offs went from under 100/month to 7 in 10 big wins in August, and 2,943 more $100K+ wins ended after Aug 30 (not studied).
pump.fun $2.5M "Squads" PnL competition (announced Oct 23 ~10:03 AM CT): squads of up to 20, sign up by Mon Oct 26, top 3 by PnL after 2 weeksevidence: B: official post
- What it is: $1.5M / $750K / $250K to the top-3 squads by in-app PnL over 2 weeks; members are fixed once it starts; "manipulation may result in disqualification or zeroed PNL".
Platform watch: nothing breaks
- Jupiter, Jito, GMGN, Axiom, Photon, Birdeye: no execution-relevant change found for Oct 21–23. No new Token-2022 incidents.
Sources
- https://bitquery.io/investigations/who-got-rich-on-pumpfun
- https://bitquery.io/investigations
- https://bitquery.io/investigations/kimchi-pumpfun-launch-snipers
- https://x.com/Pumpfun/status/2113647400057237848
- https://x.com/Pumpfun/status/2113647405941842395
- https://x.com/Pumpfun/status/2113647414573682846
- https://github.com/pump-fun/pump-public-docs/commits
- https://www.tradingview.com/news/coinmarketcal:ae3cf5d2d094b:0-pump-fun-evm-limit-orders-add-automatic-take-profit-and-stop-loss-functions-22-oct-2026/
- https://github.com/anza-xyz/agave/releases
- https://solanahub.de/en/news/meteora-dbc-transfer-gebuehren/
- https://docs.meteora.ag/developer-guides/dbc/changelog
- https://docs.raydium.io/reference/changelog
Evidence grades: A = large dataset, B = official docs, C = vendor/thread/small sample. Research notes for education only. Not financial advice.