2026-10-15

Trench lessons — 2026-10-15

Key takeaways
  • LaunchTape: 2,193 pump.fun graduations, first 10 minutes
  • Platform: Meteora DBC 0.2.2 still ships tonight
  • Token-2022 transfer-fee accounting
  • Platform watch: nothing else breaks

LaunchTape: 2,193 pump.fun graduations, first 10 minutesevidence: A; Oct 10–12 UTC, corrected Oct 14

  • Outcomes (first 10 min):
  • 38.1% touched −80% (835).
  • 62.9% fell 40%+ from a running high (1,380). An earlier draft of the post called this “−80%”; corrected Oct 14.
  • 15.0% touched 2x; 5.3% touched 3x.
  • Median low in those 10 min: about −50% from T+10s.
  • Dump clock: among 894 coins with a big-holder dump, median was 2.2 min after graduation; ~48% inside the first 2 min.
  • “Fast fill = bundle = rug” failed as a −80% filter: curves filled in <5 s (1,384) touched −80% 38.4% of the time vs 37.6% for slower fills (803). Slower fills did fall 40%+ from a high more often (85.6% vs 50.0%) and touched 2x more often (26.3% vs 8.5%) — more swing both ways, not a clean skip rule.
  • Dev % at graduation is not a one-way filter: still holding >5% (1,310) → 33.6% touched −80%; already ≤5% (603) → 47.4% touched −80%, but also 28% vs 7% touched 2x. Sold-down grads swing harder both ways.
  • Honest-fill sample (2 h Oct 10 14:00–16:00 UTC, free sample): buy +15 s after grad, sell at +3 min or −20% stop, 0.1 SOL, 1% PumpSwap + 0.5% priority/slip each side, fill at end of next slot (never signal price): 67 trades, 50.7% win, median +0.3%, mean −0.9%, total −0.063 SOL. Same rules at signal price with no costs: mean +2.3%. Stops set −20% filled at median −23.9%; Odyssey stop −82.7% after one block took the pool 248→87 SOL. Full Oct 12 day with the same rules: mean −3.9% / median −1.2% / 47.6% win on 775 traded coins.

Platform: Meteora DBC 0.2.2 still ships tonightevidence: B

  • Official changelog: mainnet 0.2.2 planned October 16, 2026, 11:00 AM UTC+8 = Oct 15 ~10:00 PM CT (same wall-clock as the 10-11 / 10-14 notes; date string is Friday UTC+8).
  • Adds create_config2 with base-mint transfer fees up to 10%, quote-mint non-zero transfer fees via token badge, SDK 2.0.0 quotes with includedTransferFeeAmountIn / excludedTransferFeeAmountOut. Badge for quote mint does not stop later fee raises; migration uses the current-epoch fee.

Token-2022 transfer-fee accountingevidence: B; Raydium docs + Soltidy Oct 5

  • Raydium CPMM/CLMM: vault receives amount_in − fee_in; user receives amount_out_gross − fee_out; minAmountOut is checked against gross (what the pool sends), not net to wallet. Fee change between quote and land can eat into slippage or revert.
  • Jupiter routes fee mints through supporting venues; docs/third-party note adapter bugs that overstated net output have been reported.

Platform watch: nothing else breaks

  • Honest gaps: no new Dune, GMGN, Axiom, Photon, Bitquery investigation, or Jito-tip numbers in this window. X unusable.

Sources

  1. https://dev.to/launchtape/what-2193-pumpfun-graduations-did-in-their-first-10-minutes-and-what-it-means-for-your-bot-4f70
  2. https://launchtape.io/test
  3. https://launchtape.io/
  4. https://docs.meteora.ag/developer-guides/dbc/changelog
  5. https://docs.raydium.io/algorithms/token-2022-transfer-fees
  6. https://soltidy.com/blog/set-and-claim-token-2022-transfer-fees
  7. https://www.belphor.com/learn/where-pumpfun-coins-trade-after-migration.html

Evidence grades: A = large dataset, B = official docs, C = vendor/thread/small sample. Research notes for education only. Not financial advice.