# A2-KITCODE (token runtime code size 1764) forward paper test (Test Manager) - 10 Oct 12:40 UTC

Feed v4tape (forward only). Universe: fee 100, unhooked, WETH/native pools whose token has eth_getCode size 1764 bytes (A2's hash 5f4a77e044c6 could not be reproduced: our keccak of the raw code differs from tokcode.db for the same size, so size is the key) (A2 #184, fw/a2-iter10-r11.md; code fetched by us via rpcpool, cached in codes.json, max 40 new per run). Entry first swap +60 s, reserve >= 0.3 ETH, 0.04 ETH, 0.01% fee + 3% slippage per side, pull (liq <= 50% of peak) = 0. Arms: A tp3 stop 0.5 4 h | B tp5 stop 0.5 4 h.
- Pools 224; token codes known 99; waiting 1 (incl. code not fetched yet); open arm-runs 0; skipped {'reserve<0.3': 13, 'other token code': 99, 'no swap after 63s': 108, 'pulled before entry': 3}; code sizes seen {3854: 98, 1939: 1}
- arm A (tp3, stop 0.5, 240 min): all n 0, mean 0.00x, <1x 0, zeros 0 | ON-TIME only (feed lag<=30 s) n 0, mean 0.00x, <1x 0, zeros 0 | late n 0, mean 0.00x, <1x 0, zeros 0
- arm B (tp5, stop 0.5, 240 min): all n 0, mean 0.00x, <1x 0, zeros 0 | ON-TIME only (feed lag<=30 s) n 0, mean 0.00x, <1x 0, zeros 0 | late n 0, mean 0.00x, <1x 0, zeros 0
- Sim expects (A2): A 1.58 (n=51, ~5/day until 3 Oct, unknown since), tp5 1.92, baseline other code 0.93. Longest loss streak arm A: 0.
- Quiet exits (pool stopped swapping before hold end) count as 0; no ModifyLiquidity in v4tape so a silent pull looks the same. Late = feed lag > 30 s: judge by ON-TIME line. Sign check agree 0, disagree 41747.
- Kill: arm A mean < 1.2x over 30 trades, or 10 losses in a row. Ready for live only after 2 profitable days, no bugs, enough trades; sellability (hook skim n/a, unhooked) still to audit; only Ben switches (D-030).