# Analyst2 idle run 26 (10 iterations, scripts it26/)
Baselines (native, +5 s entry, pos 0.1): A1 fee100 unhooked tp3 1.09 n=1268; H3000u unique hook tp5 3.31 n=243.
1. EXITS trailing stop / 50-50 ladder: A1 1.05-1.22 (time exit 120 s still best 1.29), H3000u 2.0-4.3 with unstable halves (2.3/5.5). DISCARD.
2. KYLE LAMBDA / pool L / swap acceleration (first 5 s): flat or non-monotone. logL low quartile H3000u 0.91 is the n_pre rule again. DISCARD.
3. NAME FEATURES / SUPPLY: length, words, digits, case flat. Supply 1e27 1.32 (n=633), 1e28 0.18 (n=118), 1 token 0.0 (n=50): identical to the init-price dead classes (cross-tab). DUPLICATE; totalSupply via eth_call is an equal way to read the same kit.
4. WALLET CREW (forward v4tape + txshared, 90 A1 pools): 30 wallets total, 3 in 51-70 of 90 pools: outside early money is a fixed bot crew. Crew count 1..5 -> hit3 0.12/0.16/0.26/0.32/0.25, halves unstable. MATURE -> dataeng #310 (fw/a2-req-crew26.md).
5. POOLS PER TOKEN: 1 everywhere. DISCARD.
6. CROSS-CHAIN NAME COPY (GeckoTerminal trending/new, 6 chains, vs 5,400 RH names): 3 A1 matches, 0 H3000u. Operator names are generated. DISCARD.
7. SCHEDULER CLOCK (minute-of-hour, second-of-minute): flat (A1 bucket 1.40 vs 0.96-1.09 not repeated in halves). DISCARD.
8. MINUTE-SCALE REGIME (ETH/BTC 5 m, 30 m, BTC-ETH divergence, ETH 15 m vol): max spread 0.2 (A1), none consistent (H3000u). Both families are operator-driven and insulated from market mood. KILLED, with hour-scale result.
9. LIVE mc5 UNIVERSE (v3mc5.db) vs BTC 1/6/24 h: only 34 bought, 5 of 43 with 2 h outcome; no signal readable. INFO: Ben's streak needs all-candidate outcomes, not buys; no request (organic regime already flat in run 10).
10. logL vs n_pre in H3000u: subsumed by n_pre. DISCARD. Request #310 (see 4).
Net: nothing to testmgr; 1 request out.