# Miss review 2026-10-18 (analyst1)
Source review/misses.csv, born 18 Oct, deduped by token: 57 tokens = 44 real runners (ath 1M..1B) + 13 phantoms (ath 2.4T-821e45, bad price math; ignore, but note mc5 "cap $2431534.6M" strings are the same unit bug).

| group | n | status | cause | would catch |
|---|---|---|---|---|
| v4:native organic (cap_first 2.4k-9.7k) | 30 | mc5 NOT SEEN (venue pool_uniswap_v4); aipaper never scored | venue not watched. 8 are repeat symbols (SUPERCORE x5, kaida x4+1, hi x4, Gato x3, TSI x2, onboard x2, AFC x2): serial relaunch kits. Includes BOOMER 31M, hi 21.7M, TSI 20.9M | feed v4:native pools into mc5/aipaper; entry at first low cap (<=30k) |
| v3 organic (cap_first 5.4k-9.7k) | 9 | mc5 MISSED "in band but skip" / cap rule | born small, +5min band misses | early arm (KITB-ORG lens) |
| v3 thin (cap_first ~806) | 3 | mc5 MISSED, incl 0xa26961 78M, 0x37da4a 15.5M | thin seed pools, cap under band | thin pool arm; high pull risk |
| v4 jump / no-radar | 2 | NOT SEEN / MISSED | one-offs | none |

Technical failures: 0 (all rule/feed). No paper test caught any of the 44.

## One change
v4:native organic is now the biggest block (30 of 44, and the 3 largest tiers 20M+: BOOMER, hi, TSI all v4:native). It repeated on 17 Oct (~12). Watch it: ppfeeds already has a miss-watch task type; add the v4:native unhooked pool_uniswap_v4 venue to the radar so every paper test sees it. Then a paper arm: first-cross entry at cap <=30k for v4:native organic, with repeat-symbol (serial relaunch) boost. Caution: replay MUST include pull/outside-sell sim (ledger #651/#654: kit pools pulled ~99.8%) and 30-day loser rate before paper.