# r60 (round60 arm Q) pre-buy pull rule replay (research, 16 Oct, #2648)
Data: 63 on-time forward arm-Q trades (/root/tm/round60/trades.csv, mean 1.47x, 13 winners >=4x), v4tape pools/liq + /opt/ppcopy/lp_remove.db. Scripts: research/rug4/r60.py, r60b.py.
- LP owner check is useless: all 63 pools were funded by the same sender 0x58daec31 (shared launcher/position contract), and it also does 52/52 of the removals. No deployer/LP-owner signal exists on-chain at buy time.
- Creator pull history: 53 creators for 63 pools, 0 had any earlier pool in the tape. Creators are one-shot wallets, so "deployer pulled N times" blocks 0 trades (also nonce>=10: 0). Same for "pause after 2 pulls in 2 h" per creator: never triggers. Funder history not tested (needs per-wallet RPC; fresh-wallet funders are likely as one-shot).
- Pool already pulled before entry: only 2 of 63 (the existing "liquidity not under half its peak" check at +63 s covers it).
- Pull after entry: 52 of 63 pools lose liquidity; first removal relative to our entry: p10 ~11 s, median ~230 s, 10% after 2 h. 13 winners hit tp5 before the pull; 37 of 63 exits are silent pulls booked 0.
Verdict: no pre-buy pull rule blocks anything on r60; the rug risk is intrinsic (as in rug-12oct, seed-gate 16 Oct). Do not set rug_gate true on a rule that blocks nothing. Real lever: exit side (tp5 fills before pull, 20 min hold is irrelevant after ~4 min median) and shadow modelling pulls at these timings; judge r60 on dry n>=20 with pulls booked 0 (as built). Ben's decision then rests on mean after pulls, not on a gate.