# A1 iterate run 21 (10 Oct): hook contract STATE as pre-entry metadata

## Main finding: H100-S (hooked WETH fee-100 pools, hook storage slot 1 at creation)
Every hooked (…0144) fee-100 pool has its OWN hook contract (400 of 400 distinct) with 4 config slots:
slot0 = bps (0x270f or 0x2706) + shared address, slot1 = a number (wei scale), slot2 = operator address (two operators: …681c5c 372, …9e7811 194), slot3 = pool-specific address.
**Slot 1 is a free, pre-entry label** (one eth_getStorageAt on the hook at block creation+1; the free dRPC archive node answers it).

TRAP caught: slot 1 is NOT immutable. Read at "latest" it looks even better (1.1e13 group 2.88) but 3 of 3 sampled latest=1.1e13 pools had 1.1e14 at creation: survivor look-ahead. Everything below uses the value AT creation block +1 (and +600 blocks = our +60 s entry: only 4 of 566 pools differ).

Sim = /opt/crew/work/analyst1 base() (pull-aware, impact, exit-depth check, entry +60 s, reserve>=0.5 ETH at entry), h100 pools on the WETH tape 21 Sep-8 Oct, n=396 joined of 411.

| slot1 at creation | n | tp5/20min mean | zero (pulled) | halves |
|---|---|---|---|---|
| 1.1e14 (default) | 263 | 2.14 | 21% | 2.16 / 2.11 |
| 1e13 | 68 | 3.24 | 12% | 3.31 / 3.17 |
| 1e12 | 38 | 3.89 | 3% | 4.32 / 3.83 |
| 1e10 | 9 | 1.18 | 44% | (all 2nd half) |
S = {1e12, 1e13}: n=106 (~6/day, 17 days):

| exit (pos 0.02 ETH) | S | rest |
|---|---|---|
| tp3 20 min | 2.57 | 2.26 |
| tp5 20 min | 3.47 (halves 3.59/3.35) | 2.14 |
| tp10 20 min | 4.19 (4.37/4.02) | 2.09 |
| tp10 60 min | 5.34 (5.42/5.26) | 1.20 |
At pos 0.1 ETH: tp5 3.13, tp10 3.60 (20 min). At 0.3 ETH: tp10 2.56. Capacity is small: S pools start with ~0.2 ETH reserve (default 0.73) and the entry gate needs >=0.5 ETH, so entry depth is thin: keep size <= 0.1 ETH (~$250). Peak reserve median is 8-11 ETH (like default 12.7).
Why: S pools live longer (pull age median 6,951 s for 1e13, 13,896 s for 1e12, vs 1,991 s for the 1.1e14 default; p25 2,700 vs 1,330 s), so a high TP is reachable before the pull.

Kill condition for forward test: S mean (tp5, 20 min) below 1.5 after 40 forward trades, or pulled share above 25%.
Not proven: thresholds chosen on this sample (but monotone: 1e10 bad, 1.1e14 lower); only forward data can settle it. Native hooked fee-100 pools use a different hook (slot1=0 in 224 of 266) and stay dead (0.09, 85% zero), so S is WETH-only.

## The 10 blocks
See /opt/crew/ideas/analyst1.md (run 21). Scripts: hs3-hs14.py, db1-db5.py, fd1.py in /opt/crew/work/analyst1. Data: hookstate.db (latest, look-ahead, do not use), hookstate2.db (creation block, use this).