# A3 #1998/#2001: misses with true seed cap (11 Oct night)
Source: review/misses.csv (304 rows) joined to ppfeeds fill1996 (227 radar crossings; 28 are organic misses with seed/cap_5m).
- 23 of 28 matched misses are v3 pools with seed (first-print) cap < 12k: SEEDHI confirmed on true cap_first (earlier cap_first was radar value, 300k+).
- Their cap at +5 min: of 24 matched: 9 in 300-600k, 6 below 300k, 9 above 600k: a 300-600k band entry at +5 min is the entry window; ones >600k (ZEVIO, SWARM, hoodcat 17M) were already past it at +5 min -> need a seed-triggered earlier entry (cap<=300k within 2-3 min), not the mc5 band.
- 3 old pools (SRX, CAPYBARA, SAI; age 2-4.5k s, cap_5m 2.5k-58k) = slow risers; age>2000 s rule: nothing to catch at +5 min.
- Pons grad (OP, WOJAK) and BMBFI (seed 2.6M born high) are born-at-price copies: not catchable by seed.
- Caveat: sample is runners only (misses list); ceiling is hit share among seed<12k crossings still needs population base (seed<12k n=81 of 201 crossings, all ath unknown).
Action: A3-SEEDHI keeps testmgr #1785 arm; add early (+2-3 min, cap<=300k) variant to the idea, no new request.