# analyst3 idle run 19, 10 Oct 2026 (16:00-16:20 UTC). Scripts k1-k6.py. 4 lenses; ideas space saturated, no new arm.
Base: v3 mc5 band 300k-1M at +300 s, tp5 sl.5 4h (n350, mean 1.07, hit5 ~11%), enter next print after +305 s, 7% slip.
1. Forward BREADTH (kitenrich buyers300, 186 rows since 9 Oct): only 9 band pools matched; buyers300>=30 are 7 of them, all lose (0.5-ish). n too small, no signal; keep forward arm running. dsboosts still 6 boosts: boosted-at-T test impossible.
2. Order-flow liveness (NEW lens, microstructure): age of the last print at +300 s. Causal version: >10 s old -> 0/90 reach 5x (mean 0.80, halves .78/.83) vs 13.8% hit5, mean 1.17 fresh. Looks strong but it is a volume proxy: stale pools are 101 of 106 with vol5<49k (existing VOL5 flag). Inside vol5>=49k only 5 stale. DISCARD (already covered by VOL5/A3-CAPS arm).
3. Other tape-shape features on the band (quartiles, halves): top-3 print share of the 5-min climb (1.19 .. 0.95, h2 weak), sell fraction by count (non-monotone), signer round-trip share (non-monotone), direction of last 5 prints (non-monotone), last-60 s change (non-monotone). DISCARD all.
4. Unwatched pools: nothing new to test; v4 nohook native/WETH/USDG remain jump/ceiling sets per ledger #337 #351 #315, and the tape (#547/NH100) is the open piece.
Verdict: nothing ready for testmgr. Next run: wait for forward results of SERIAL, SEED, VOL5, DIP, NH100, MOMDD, WIDE, BREADTH instead of new lenses.