2026-10-31

Trench lessons — 2026-10-31

Key takeaways
  • Raydium CPMM creator fees are live on mainnet: quote each pool's round-trip cost, not just price impact
  • Binance withdrawals now targeted for Nov 1; expect a burst of fresh Binance-funded wallets after
  • Oct 30 launch-block buyouts ran to $15–16M then went to near zero: the CATBOT shape again
  • A deployer whose graduations jump from 0 to 10 in a day is a better skip signal than raw launch count

Raydium CPMM InitializeV2 is deployedevidence: B: mainnet RPC read + Raydium docs

  • Raydium redeployed its CPMM program on Oct 30 at 9:37 PM CT (slot 463,522,946), and the deployed bytecode contains the InitializeV2 instruction. Any pool creator can now turn on a creator fee. Raydium CLMM (Oct 22), the pump.fun curve and PumpSwap (Oct 19) and Meteora DLMM (Jun 3) were not redeployed.
  • The creator only picks on/off and which side. The rate comes from the fee tier's creator_fee_rate, which defaults to 0, so two pools on the same tier can now cost different amounts. Raydium tells routers to quote from each pool's own enable_creator_fee flag, never from the tier.
  • Lesson: a price-impact cap measures impact, not fees. A fee-on pool can pass an impact check and still cost up to ~3% round trip. Quote a buy-then-sell round trip before entering, and check the pool's creator-fee flag when the route goes through CPMM.

Binance withdrawals: target moved to Nov 1evidence: B: Binance statements via news; C: news

  • Binance now targets restarting withdrawals on Sunday, Nov 1, with a reserve report Nov 2 and the full proof of reserves around Nov 3. Coverage of stolen assets is up to 90%. SOL on the Solana network is still paused ("wallet maintenance"), while SOL via BNB Smart Chain stays open. TRM Labs and Chainalysis attribute the hack to North Korea's TraderTraitor group. Binance says the reopening still depends on security testing.
  • Lesson: treat the market as an abnormal regime until Solana-network withdrawals actually reopen. Expect a burst of fresh Binance-funded wallets right after, and don't read them as linked just because they share Binance as a funder.

Oct 30 pump.fun: launch-block buyouts to $15–16M, then near zeroevidence: B: pumpdata.fun recorder

  • Activity stayed weak: 39,183 launches (−19% vs the 30-day average), 872 graduations (2.23%), 3.73M SOL traded (−20%).
  • The busiest coin, $USOE, was bought out in its launch block, peaked at $16.1M and closed at $1.7K on 50,548 SOL of volume. $UDTF went from $15.5M to $2.0K the same way. $HYPERCHAN, also bought out in its launch block, closed 87% under its $3.0M peak.
  • One deployer went from 3 lifetime launches and 0 graduations to 61 launches and 10 graduations in a day, which looks like a coordinated batch rather than talent.
  • Lesson: launch-block buyout plus huge volume plus near-zero close is the CATBOT shape again. Market-cap band and reprice gates catch it. A sudden jump in a deployer's graduation count is a sharper skip signal than raw launch count, since 500+-launch spammers rarely graduate anything.

Smaller itemsevidence: C unless noted

  • Serial creators collapse more often: Before It Rugs traced 7,715 new coins and found that wallets behind 3+ coins saw 48% collapse within a day, against 29% for one-time creators. It's a vendor dataset, but it agrees with the Oct 30 batch above and supports a "deployer has 3+ launches" weight in rug scoring.
  • Helius preconfirmations (B, vendor docs via news): preconfSubscribe is now on every paid Helius plan, with signer and instruction filters. Watching a wallet's buys at the preconfirmation stage could cut copy-trade latency.
  • Agave v4.4.0 recommended for mainnet Oct 30 (B): feature activation resumes ~Nov 9. RPC now accepts minContextSlot on getTransaction / getSignatureStatuses (error −32016 when the node is behind). Handle that error rather than treating it as a failed transaction.
  • pump.fun sent Eliza Labs' founder a cease-and-desist over a claimed 97.6% reverse-engineered copy of its contracts. If it ships, expect pump-clone programs with different IDs; parsers should only trust the real program IDs.
  • pump.fun keeps adding non-Solana trading (Arbitrum Oct 28, Polygon Oct 30, Hyperliquid perps, Elysium Nov 9). Some pump.fun-app flow and attention moves off Solana memecoins.

Sources

  1. https://docs.raydium.io/reference/changelog/2026-10-30-cpmm-initialize-v2.md
  2. https://docs.raydium.io/reference/changelog
  3. https://www.tokenpost.com/news/business/44892
  4. https://cryptobriefing.com/binance-hack-recovery-withdrawals/
  5. https://cryptoticker.io/en/binance-withdrawals-solana-xrp-paused-after-hack/
  6. https://pumpdata.fun/report/2026-10-30
  7. https://pumpdata.fun/stats/2026-10-30.json
  8. https://beforeitrugs.com/reports/
  9. https://solanahub.de/en/news/
  10. https://solanacompass.com/news/anza-tells-solana-mainnet-validators-to-upgrade-to-agave-v440-three-days-ahead-of-schedule
  11. https://www.tokenpost.com/news/regulation/44182

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