# A2: miss review 9 Oct, second view (#358) + r6 data check (#362), 10 Oct 2026

Input: review/misses.csv, 66 runner tokens born 7-9 Oct (ATH $1M+), 231 rows.

## What the missed runners share (by venue, count of tokens)
| group | tokens | note |
|---|---|---|
| 7 Oct "?" venue | 23 | PONS, CASHCAT, AI, SUSD, ORBIO ... = USDG/SUSD-quoted v4 (A2-USDG4). 5 of the 6 biggest of the three days |
| 9 Oct USDG v4 no hook | 4 | AAPL $309M, BONER $42M, STANDARD, V4 |
| v4 WETH "no hook" | 14 | **11 of 14 have a second pool of the same token: fee 3000, hook address ending 0xa80 (the H3000 kit)**. ATH $1.4M-7M. The "no hook" label is the dummy sibling pool. Others: Clanker (fee100 hook 0x144), BOZ/GRAPHITEBOOM (fee 100 no hook, 4-5k swaps) |
| v3 launcher WETH | 8 | known launchers (ClapCat = our 0x73991a launcher, 1.0M) |
| pons_v2 / v3 pool / v2 / bankr / flap / letscash | 13 | one-offs, each venue 1-3 tokens |

Time: the 0xa80 runners were born across the whole day (07 Oct 21:28 ... 09 Oct 15:04), no clock effect. Hook family census: 121 pools of fee-3000 + 0xa80 since 26 Sep, 5-11 a day, median 10k swaps per pool. Peak from +66 s entry: median 7.9x, 99 of 117 reach 5x (peak only, not fillable P&L).

## Answers
1. The live bots (mc5/fml2/aipaper) miss these because they are not "v4 ETH pools fmmc5 watches" or never scored. Not a scoring problem: a coverage problem. 
2. Two groups carry the value: (a) USDG/SUSD v4 (biggest runners, no bot or test covers it), (b) H3000/0xa80 WETH family (already paper-tested as A2-H3000, fw/tm-h3k.md: arm A 2.60x n13, E 14.7x n12). The 11 misses are exactly what that test is for: **no new test needed for (b)**; they confirm the family produces a $1M+ runner roughly daily. Live gap is only that no live bot runs it (Ben's call, D-030).
3. Yesterday's proposals: the 0xcc native kit (arms E/F, run 27) and H3000 stay as sent; this view adds nothing that contradicts them.

## #362 data check (r6weth/r6nat/r6tape to 83.96M): covers the gap, schemas as before. First use: USDG/SUSD v4 backtest
Script /root/fwa2/it28/usdg_bt.py, rule: first swap +66 s entry, stop 0.5, tp 5x, 4 h, 3% each side, price from tick only (no depth gate, no sell simulation), 5,466 pools.
- All USDG/SUSD pools: mean 1.35x (halves 1.38/1.33), 14% reach 5x. This IS the dumb baseline. 
- Sub-groups by fee/hook: some 1.5-1.8x (fee 800000, 40000, 9000) but halves disagree or n<110, and none is separated from baseline by a pre-entry rule. Hooked pools are worse (1.17).
- Verdict: the universe is more profitable than random even unpicked, which is suspicious (dust pools, tick-only prices, unfillable). NOT READY, no testmgr arm. EARLY STAGE 1.
- Next (own work, next iterate run): add depth gate (liquidity in USD at entry), size 0.04 ETH impact, then separate PONS/AAPL-type runners from the 5,466 by pre-entry features, outside-money test.