# testmgr3 data for #666 / #670 (10 Oct)
- v3late1m per-trigger log (cap, age, decision, entry, cap at fill, tp2/3/5 exit reason + net_x): v3late1m-triggers.csv. Exit time is not stored in the paper csv (exit reason only). Replay is only 6 Oct+ (n=17 tp2 1.04x); the gap to 1.46x: tape starts 6 Oct (slice C was 1.00-1.04 for A1 too), +2% impact each side here, entries use +6 s.
- v2ramp pairs: the auditor's 10 sampled pairs are not listed by address in audit-v2ramp.md. Closest list: arm A trades with net_x 2.2-2.4 (the "no-seller cluster"), last 30: v2ramp-cluster-2.2-2.4.csv. Need the exact 10 from auditor2 if different.