# Miss review 2026-10-21 (analyst1)
Source review/misses.csv (born 21 Oct), deduped by token: 25 runners (20M+ 2, 10M 2, 5M 3, 3M 9, 1M 9). Technical failures: 0. No paper test bought any.

| group | n | mc5 status | cause | would catch |
|---|---|---|---|---|
| v4:native organic (cap_first 9.7-10k) | 17 | NOT SEEN (venue not watched by mc5) | aipaper scored 12 of them 0.24-0.68 (below line), 5 never scored. Serial names: Kimchi x6 (2.5-17.3M), FINLEY x3 (1.3-6.8M), CUMROCKET x2, Dacity 11.5M | v4:native feed into mc5 radar, entry at first low cap |
| v3 thin (cap_first ~820) | 2 | MISSED (0x800d6f 60M "in band but skip"; 0x97811f cap $51k at 5 min) | thin seed, pull risk | thin arm with pull sim |
| v3 organic / jump | 4 | MISSED (in band but skip x2; cap over $1M at 5 min x2) | rules | none |
| v4 hook (0xb63d07 32M, CREORA 3.5M) | 2 | cap $7.3M at 5 min / NOT SEEN | hooked | none |

Same picture as 18-20 Oct. Existing ledger covers it (#608, #642 SERIAL-SYM, #698, #681, #684, #699, #694, #561, #722).
New angle this time: Kimchi/FINLEY/CUMROCKET names repeat on EVERY relaunch; of the 6 Kimchi pools, 5 reached 2.5-5.8M. A same-symbol rule (2nd+ pool of a symbol that already ran >=1M in the last 48h) is lookback-only, no look-ahead.
## One change
Add v4:native organic to mc5's watch (feed) with a symbol-repeat gate: buy at first low cap only when the symbol already ran >=1M before. Covers ~13 of 17 native misses (Kimchi, FINLEY, CUMROCKET, Dacity, Pokemon) without buying one-off names. Needs pull sim (earlier #698/#681 died on outside-sell).