# a2: LP owner reputation on v4 native pools (10 Oct, #742; #740 noted)
Script it45/o.py. Join lpowner.db x canon_pools.db (v4:native*, max_cap known): n=967 pools, split by time into halves. Feature is causal: owner's earlier pools minted >4h before this one, "runner" = past max_cap>=1M.
| bucket | H1 n / 1M / 3M | H2 n / 1M / 3M |
|---|---|---|
| all | 483 / .63 / .51 | 484 / .55 / .31 |
| owner >=3 past, 0 runners | 72 / .01 / .01 | 48 / .00 / .00 |
| owner >=3 past, >=1 runner | 196 / .85 / .78 | 163 / .47 / .45 |
| owner >=5 past, >=2 runners | 170 / .93 / .85 | 89 / .76 / .75 |
Reading: an owner whose 3+ earlier pools never ran does not run again (0/120): a clean SKIP filter. Repeat runner owners are the strongest positive in both halves, but H2 decays (.93 -> .76).
Caveats (not tradable yet): base rate is inflated (max_cap includes pre-entry peaks, only 967 of 4,377 pools have a cap); owner may be a contract; no entry-price test. Next: tape-based return from entry at first buy after mint for the ">=5 past, >=2 runner" bucket, and live feed of mint owner at pool creation (known in the mint tx, so usable at t+0).