# A2-FAST5 (enter 5 s after first swap) forward paper test (Test Manager) - 12 Oct 13:10 UTC

Brief: fw/a2-brief-fast5.md (#257). Feed v4tape, fee 100, unhooked, WETH/native. ON-TIME only (entry swap emitted lag <= 2.5 s). 0.04 ETH, 0.01% fee + 3% slippage per side + buy impact size/R + sell impact size*mult/R, pull = 0. Brief says arm B "same, tp5" and "B >= 0.4 ETH": run as B = tp5 at reserve>=1.0 and C = tp3 at reserve>=0.4.
- Pools seen 556, waiting 1, open arm-runs 0, first swap by ramp bot 547/555; skipped {'A late>2.5s': 45, 'B late>2.5s': 45, 'C late>2.5s': 45, 'D late>2.5s': 39, 'E late>2.5s': 39, 'Z late>2.5s': 25, 'R late>2.5s': 45, 'Z5 late>2.5s': 45, 'U late>2.5s': 45, 'T120 late>2.5s': 45, 'T300 late>2.5s': 45, 'AF late>2.5s': 45, 'AT2 late>2.5s': 45, 'A reserve<1': 9, 'B reserve<1': 9, 'C reserve<0.4': 8, 'D reserve<1': 9, 'E reserve<1': 9, 'Z no swap after delay': 39, 'R reserve<1': 9, 'Z5 reserve<1': 9, 'U reserve<1': 9, 'T120 reserve<1': 9, 'T300 reserve<1': 9, 'AF reserve<1': 9, 'AT2 reserve<1': 9, 'R ramp<1.2': 4, 'D no swap after delay': 18, 'E no swap after delay': 18, 'A no swap after delay': 11, 'B no swap after delay': 11, 'C no swap after delay': 11, 'R no swap after delay': 11, 'Z5 no swap after delay': 11, 'U no swap after delay': 11, 'T120 no swap after delay': 11, 'T300 no swap after delay': 11, 'AF no swap after delay': 11, 'AT2 no swap after delay': 11, 'A pulled before entry': 1, 'B pulled before entry': 1, 'C pulled before entry': 1, 'D pulled before entry': 1, 'E pulled before entry': 1, 'R pulled before entry': 1, 'Z5 pulled before entry': 1, 'U pulled before entry': 1, 'T120 pulled before entry': 1, 'T300 pulled before entry': 1, 'AF pulled before entry': 1, 'AT2 pulled before entry': 1, 'Z reserve<1': 3, 'Z5 ramp>=1.2 (ON group)': 1}; sign check agree 0 disagree 96587
- arm A (tp3, enter +5 s, reserve>=1): n 5, mean 1.53x, <1x 2, zeros 2
- arm B (tp5, enter +5 s, reserve>=1): n 5, mean 2.40x, <1x 2, zeros 2
- arm C (tp3, enter +5 s, reserve>=0.4): n 6, mean 1.27x, <1x 3, zeros 3
- arm D (tp3, enter +8 s, reserve>=1): n 4, mean 1.31x, <1x 2, zeros 2
- arm E (tp5, enter +8 s, reserve>=1): n 4, mean 2.09x, <1x 2, zeros 2
- arm Z (tp3, enter +60 s, reserve>=1): n 4, mean 2.06x, <1x 1, zeros 1
- arm R (tp5, enter +5 s, reserve>=1): n 1, mean 3.98x, <1x 0, zeros 0
- arm Z5 (tp5, enter +5 s, reserve>=1): n 4, mean 2.00x, <1x 2, zeros 2
- arm U (tp10, enter +5 s, reserve>=1): n 5, mean 1.77x, <1x 2, zeros 2
- arm NA (tp3, enter +5 s, reserve>=0): n 0, mean 0.00x, <1x 0, zeros 0
- arm NA8 (tp3, enter +8 s, reserve>=0): n 0, mean 0.00x, <1x 0, zeros 0
- arm NS (tp3, enter +5 s, reserve>=0): n 0, mean 0.00x, <1x 0, zeros 0
- arm NS5 (tp5, enter +5 s, reserve>=0): n 0, mean 0.00x, <1x 0, zeros 0
- arm T120 (tp1e+09, enter +5 s, reserve>=1): n 5, mean 1.13x, <1x 2, zeros 2
- arm T300 (tp1e+09, enter +5 s, reserve>=1): n 5, mean 1.77x, <1x 2, zeros 2
- arm NT120 (tp1e+09, enter +5 s, reserve>=0): n 0, mean 0.00x, <1x 0, zeros 0
- arm AF (tp3, enter +5 s, reserve>=1): n 5, mean 1.53x, <1x 2, zeros 2
- arm AT2 (tp3, enter +5 s, reserve>=1): n 5, mean 1.53x, <1x 2, zeros 2
- Pull share (A1 #262, stop if >25%): A 0/5, B 0/5, C 0/6, D 0/4, E 0/4, Z 0/4, R 0/1, Z5 0/4, U 0/5, NA 0/0, NA8 0/0, NS 0/0, NS5 0/0, T120 0/5, T300 0/5, NT120 0/0, AF 0/5, AT2 0/5
- Arms D/E = A/B entered +8 s (A1 #262, lag cost). Fill is price of first swap at/after the delay, not block-quantised.
- Sim expects: +4 s 1.71 (n=460), +60 s 0.95. Longest loss streak arm A: 2.
- DUPNAME split arm A: dup n 3 mean 1.69x; non-dup n 0; unknown 2 (tag from 19:36 UTC 9 Oct, reliable false from 21:36 UTC 11 Oct). Kill: spread <0.15 after 60 per tag.
- DUPNAME split arm Z: dup n 3 mean 2.74x; non-dup n 0; unknown 1 (tag from 19:36 UTC 9 Oct, reliable false from 21:36 UTC 11 Oct). Kill: spread <0.15 after 60 per tag.
- A2-FAST5-RAMP (#274, fw/a2-brief-fast5-ramp.md): arm R = arm B (tp5) only if entry price / price after the pool's first swap >= 1.2; arm Z5 = B on the OFF pools. Sim ON 2.19 vs OFF 1.48. Kill: R n>=40 and (R mean < Z5 mean or R < 1.2x). Size cap in any live talk 0.1-0.25 ETH.
- A2-U100CLOCK (#279, fw/a2-iter18.md) = arm U: enter +5 s, tp10, stop 0.5, sell at first swap >= entry+300 s (else pull = 0). Sim 2.12 vs 1.84 for hold, but forward v4tape 9 Oct only 1.67 vs 1.72 (risk smoothing: win 48% vs 44%). Compare U with arm B/A. Kill: U n>=40 and mean < arm A mean, or < 1.2x.
- Native-quoted pools (#292, A2 #290, fw/a2-native-290.md): arms NA (tp3 +5 s, all native), NS (tp3, skip if ETH in range at entry > 0.6), NS5 (same, tp5). All other arms are WETH only. Sim native +5 s tp3 1.09 (62% zero), best bucket ~1.5. Kill: NS n>=40 and mean < NA mean or < 1.2x; native tp3 NA mean < 1.0 over 60 = discard.
- Time-exit arms (#293, A1 fw/a1-iter18-testmgr.md Test 2): T120 / T300 = WETH, enter +5 s, no TP, sell at first swap >= entry+120 / +300 s (stop 0.5 and pull=0 still apply); NT120 = native fee-100 same. History WETH T120 1.45 (zero 38%) vs tp3 1.31; T300 1.50; native T120 1.29 vs tp3 1.09. Kill: n>=100 per arm and mean <= arm A (WETH) / NA (native) on the same pools.
- A2 #351 arm AF = arm A but skip if >=4 distinct buy txs in the first two blocks (sim: reach3x 0.21 vs 0.37). Compare AF vs A on the same pools; keep as skip rule if AF mean > A mean over n>=60 and the skipped pools' mean < 1.3. Kill: arm A below.
- #336/#879 arm AT2 = arm A but skip if a Tier-2 flag (DEX paid order or template-site cluster>1, from dsmeta/dsorders, first seen <= entry) is known at entry; no Tier1 (X age/mutuals not available). Flags are polled minutes after launch so few are known at +5 s: read the skipped count. Verdict n>=60: AT2 mean > A mean and skipped mean <1.3.
- Kill: ON-TIME arm A mean < 1.2x over 40 trades, or first swap no longer the ramp bot. Quiet exits count 0 (silent pulls invisible). Not for live: execution latency and sell fill unproven; only Ben switches (D-030).