# miss-watch venues #2398-#2401 (ppfeeds, 14 Oct 2026)
Finding: the pool BIRTHS of all four venues are already in live feeds, all current at 00:20 UTC (07:20 BKK). The misses are on the consumer side.
| venue | feed (research-box) | fresh proof (14 Oct) |
|---|---|---|
| v3 (incl. launcher/no-hook WETH) | v3live.db pools (9,416 v3 pools, ~277/day) + radar.jsonl (venue v3). Pools VORTA, WISP, ZEVIO, swordcat are all in v3live.pools | newest pool 0x1b1f93f3.. created 00:17:28 UTC, lag 0 blocks |
| pp:v4 no hook WETH / pool_uniswap_v4 | pretag.db launch (every v4 Initialize, any hook, 43k rows) + radar.jsonl v4:* + aipick reads all PoolManager Init/Swap | newest launch 00:19:39 UTC |
| pp:v2 no hook WETH | v2feed pairs.jsonl (any factory, WETH pairs, since 9 Oct 14:22) mirrored to fw/a1feed/v2pairs.jsonl. REALM's v2 pair (8 Oct) predates the feed | newest pair block 87935063, 00:21 UTC |
Why "no bot saw" them: (1) aipick/aipaper* score only v4 PoolManager pools, so v3 and v2 coins never reach a scorer (v3pick covers only fee-100 v3); (2) in aipick a v4 pool that is not picked within 1 h of birth is dropped (aipick.py line ~263), and these runners start at 2.5-10k cap and grow slowly ("organic"), so they are gone before they run; the 'never scored' v4 WETH rows are that case.
Not a feed change: a v3 scorer or a longer watch window changes what paper tests do, which is the test managers' call. No feed code touched.