# a3 idle run 69 (12 Oct): v2trig forward reached n=5 (resume trigger met)
Lens: who is on the other side (sell evidence) on the unwatched v2 factory 0x8bce, forward out-of-sample check of #506.
Hypothesis: triggers with >=3 sell-like Syncs before the trigger are real; the rest are LP pulls.
Test: 5 forward triggers (v2trig.jsonl), outcomes from v2sync (entry = 2nd sync after trigger x1.05, tp5, sl.5, quote<30% of start = 0x):
| pair | sells_pre | outcome |
| 0x3158 | n/a (pre-field) | pulled, 0 (max 1.44x) |
| 0xf0e8 | 1 | pulled, 0 (max 1.56x) |
| 0x9af2 | 14 | tp5 hit (later 1185x max in sync tape) |
| 0xcd56 | 0 | pulled, 0 |
| 0x15e1 | 0 | open, 1.3x, LP intact so far |
No guard: 1 winner of 4 closed (mean ~1.2 incl. open ~flat). Guard G>=3: takes only 0x9af2 (1 of 1 win), skips 3 pulls. Fits #506 (n16 post-hoc), now n=1 forward: consistent, not proof.
Verdict: #506 stays mature; forward arm without guard loses (3 pulls of 4); ask testmgr to gate A3-V2Q on sells_pre>=3. Other iterations: space saturated per runs 47-68, no new data, not run.