2026-11-14
Trench lessons — 2026-11-14
Key takeaways
- pump.fun bonding-curve program was redeployed Nov 13, 2:12 PM CT
- Holder Rewards fee on new custom-pair coins narrowed to 0.3–1%
- SIMD-0460 did not activate at epoch 1086
- A public scoring rule set for bundles and coordinated dumps
- Smaller items
pump.fun bonding-curve program was redeployed Nov 13, 2:12 PM CTevidence: A: on-chain, ~10:20 AM CT
- What: the pump.fun bonding-curve program was redeployed Nov 13, 2026 at 2:12 PM CT (programdata slot moved 458,921,440 → 469,055,660). No pump-public-docs commit since Oct 13. It came about an hour after pump.fun's 1:15 PM CT post narrowing Holder Rewards fees, so a likely reason is that new limit being enforced on-chain. That's a guess, not confirmed. PumpSwap and every other venue program were unchanged.
- Health sample: the last 1,000 curve-program txs had 598 failed (≈60%, in line with pumpdata's 62.6% for Nov 12). The top errors are the same everyday mix (Custom 6017, 301, 11, 1, 6009), with no new program-error code dominating. PumpSwap had 313/1,000 failed. Trades are landing after the upgrade.
Holder Rewards fee on new custom-pair coins narrowed to 0.3–1%evidence: B: pump.fun Telegram + pumpfyi, Nov 13
- What: new non-SOL/USDC "custom pair" coins with Holder Rewards can now set only 0.3%–1% (was 0.01%–3%). Existing coins keep their old rate, so 2–3% coins still exist. Stacked cost on a custom-pair coin is 1.25% curve fee + Holder Rewards fee (+0.45% app fee if traded in-app). New pairs this week: MAG7 (Nov 13), FRONTIER, JENSEN, OPNW, MNST, PRIVACY. PRIVACY, JENSEN and MAG7 quote assets withhold 0.50% on every transfer (Token-2022 transfer fee).
SIMD-0460 did not activate at epoch 1086evidence: A: on-chain; B: ChainReport
- What: the Nov 13 note said SIMD-0460 would switch on around 7:30–8 PM CT at the epoch-1086 boundary. It did not activate. ChainReport still shows it queued for the next epoch. Epoch 1087 starts about Nov 14 ~9:50 PM CT (03:53 UTC Nov 15). It changes how the VM lays out program memory; client transactions are unaffected.
A public scoring rule set for bundles and coordinated dumpsevidence: B: DeepCounterfeit, conyr
- What: DeepCounterfeit's scoring flags: coordinated dump = ≥3 early wallets make their first sell within 10 s of each other; bundled launch = non-creator buyers in the creation slot +2 slots linked to the creator holding ≥3%; sells after a call = linked wallets sell ≥1 SOL in the 15 min after a call at ≥2× their prior-15-min selling, then mcap −30% within 1 h. Each comes with stated false-positive risks (sniper bots, MEV round trips). conyr's write-up: link wallets by funding lineage + execution timing + exit synchrony; coordinated exits take slots, not minutes.
Smaller itemsevidence: no action
- Nov 13 day (B: pumpdata): 45,533 launched, 909 graduated (2.00%), 3.91M SOL traded, 92% of launches dead in hour two. 165 of 909 graduates (18.2%) graduated within 1 s of launch (whole curve bought in the launch block). The rest took a median of 4.7 min. Launch-block buyouts: 225, 41% under the 30-day average. Only 2 week-old coins held all day ($inuhao, $FOOD).
- Venue programs: PumpSwap, Raydium CPMM/CLMM/AMM v4, Jupiter v6, LaunchLab, Meteora DBC, and Token-2022 are unchanged from Nov 13. The pump.fun bonding-curve program changed, as noted above.
Sources
- https://github.com/pump-fun/pump-public-docs
- https://tg.me/pumpfun/304
- https://pumpfyi.com/news/2026-11-13-pump-fun-narrows-holder-rewards-fees-custom-pairs/
- https://pumpfyi.com/news/2026-11-06-pump-fun-adds-monster-mnst-custom-pair/
- https://github.com/pump-fun/pump-public-docs/blob/main/docs/PUMP_PROGRAM_README.md
- https://chainreport.co/fork-desk/solana-feature-gates/
- https://deepcounterfeit.com/manipulation/methodology
- https://conyr.ai/blog/detecting-coordinated-bundles-on-solana
- https://pumpdata.fun/stats/2026-11-13
Evidence grades: A = large dataset, B = official docs, C = vendor/thread/small sample. Research notes for education only. Not financial advice.