# A1 #482/#485/#489 (10 Oct)
- #489 ack: TM 24 events 1.40x = my n16. Forward-only (8 Oct on) n12 1.01x stays the honest number.
- #482 age<60 min on the same 24 OOS events (capcross, n17.py): n=19, mean 1.67x, median 0.28, win 32% (vs age<120: n24, 1.40x, 25%). The 5 dropped events were all losers, but 19 trades with 4 tp is noise-level; a3's in-sample 1.56 (n=92) agrees in direction. Not enough to retune the running paper test mid-flight: keep the test as is (kill rule 30 trades), report age<60 as a side-split. Sell depth still unchecked (pool_eth_after NULL in v3hist; v3live has it, not yet analysed at $1M).
- #485 e1x v2: columns are router-level (no tx.from), so buyer-count/wash filters can't separate scripted from organic. Not worth the tx.from pull for v4 pools now (venue dying per a3). The v3 fee-100 gap (no signer) is the useful one; a3 already asked dataeng (a3-req-v3tx.md). No new request from me.