# analyst2 iterate run 27 (10 Oct 2026) - scripts it27/

New lenses: roundness of bot amounts, Lindy/survival entry, be-the-LP (fee income), relative strength vs concurrent pools, LP liquidity change, **hook-address permission bits (v4 low 14 bits)**, vanity effort, flag-signature census, hook deploy lead time, exit-time plateau.

1. ROUND/DUP amounts in pre-entry buys: A1 and H3000u flat (H3000u low bucket = few swaps, n_pre confound). DISCARD.
2. LINDY entry (wait to age 60/180/600 s, then 120 s hold): A1 1.35 (age 5) -> 1.10 -> 0.91 -> 0.96; H3000u 2.21 -> 1.27 -> 1.21 -> 0.87. Survivors are priced; edge is in being early. DISCARD.
3. BE THE LP in H3000u: wash flow 59 ETH per 10 min on 16 ETH reserve = ~1% fee yield, but price ramp IL median -42%. DISCARD.
4. RELATIVE STRENGTH vs pools launched +-10 min, and L change in first 5 s: noisy, halves disagree. DISCARD.
5. BOOTSTRAP sample size: A1 P(mean>1) at n=50/200 = 0.66/0.82 (no edge); H3000u 1.00 already at n=20. INFO.
6. **HOOK PERMISSION BITS** (hook address & 0x3FFF; v4 encodes which callbacks the hook has): fee 3000 native with bits == 0xcc (beforeSwap+afterSwap+both return-delta): 183 pools, +5 s tp5 4.11 (halves 3.48/4.74), sell +600 s 7.58 (7.65/7.51), zero 7%. The current "unique hook" rule (3.31 tp5, 6.44 at 600 s, zero 15%) is contaminated by 60 pools of other kits that look unique at their first pool (0x880, 0x20cc, 0x2044, 0x28cc: mostly 0.0). Flag is known at pool init, no look-ahead, no "earlier pool" lookup. With n_pre>=3: 155 pools, sell +600 s 8.04, zero 1%, p10 2.92, halves 8.6/7.5. FORWARD (tm h3n, 10 trades): 9 pools with 0xcc = 7.99x at 600 s; the 1 non-0xcc "unique" pool (0x880) 1.31x. All 16 forward 0xcc pools have 16 distinct hooks. READY -> testmgr.
7. VANITY effort (leading zeros, zero/c count in hook address, 0xcc cohort): flat. DISCARD.
8. FLAG-SIGNATURE CENSUS (fee, bits) with share of pools >=100 swaps: r6weth 0x144/100 91% (h100), 0xa80/3000 99% (WETH H3000), native 0xcc/3000 85%; no new signature with >=8 pools and >=50% survival appeared in the last days. INFO (a cheap new-kit detector; rerun daily).
9. HOOK DEPLOY LEAD: hook code absent at init block-1 for 5 of 6 recent 0xcc pools (deployed in the same block, code 7,573-7,601 B; older 16,724 B one pre-deployed). No pre-launch warning. DISCARD. Note: code is now 7.6 kB, not 16.7 kB.
10. EXIT PLATEAU for 0xcc + n_pre>=3 (entry +5 s, 0.1 ETH): 300 s 3.91, 450 s 4.69, **600 s 8.04**, 630 s 8.06, 900 s 5.79, 1200 s 5.12; 8% slippage per side still 7.28 at 600 s. Size 0.5 ETH 7.35, 1.0 ETH 5.79 (reserve check removes half). Sharp clock peak = operator script; data request ppfeeds: real sell replay of 0xcc at +600 s (fw/a2-req-0xcc-sellsim.md).

Caveats: single actor per pool (no outside buyers), operator pulls ~1325 s, pool survival depends on the family lasting (decay 8 Oct 3.35 n=9). Only a real buy+sell (Ben, D-030) proves outside exit.