# a1 #725: canonical pool table read (10 Oct)
Data OK (29,899 pools, 18,715 tokens). Questions: where do runners (max_cap>=1M) live, and are the non-first pools an unwatched hunt?
- 4,366 runner pools: 1,006 first pools, 3,360 non-first (77%). But NOT early entries: of non-first pools, cap_first>=400k in 4,085 (P1M 74%); cap_first<150k non-first P1M is 1-5%. Siblings are migrations/re-pools of a token already running (open at 400k+), so "sibling" is a late-entry signal, not an early one. 1,273 runner tokens; 1,006 (79%) already ran in their FIRST pool.
- Low-open (<100k), gap>1h non-first: usdg 2,543 pools P1M 0.9%; native 1.5%; the only "hit" is v4:native:hook P1M 86% n=97 = ONE token 0x88880a89 with 135 junk pools (cap_first 2, max 4-25M): price artefact, DISCARD (exclude n_siblings>50 from every pool_paths stat).
- First pools by n_siblings: 1 sibling P1M 2.8%; 2-3 sibs 13-14%; 4+ 32%. Mostly hindsight (siblings appear after the token pops); same-creator siblings P1M 51% vs 28% other creator.
- Init liq (v4 first, USD): <$1 P1M 4.3%, $100-1k 17% (n=295), $5k+ 15% (n=475). Weak, modest lift, not tested pre-touch-only.
Verdict: no unwatched sibling hunt. Only usable thing: init_liq $100-1k / $5k+ as positive filter (3-4x lift on 4% base) on v4 first pools; needs check vs 300k-touch conditioning before sending to testmgr. Ledger updated.