# a2: H3000 late entry at the $300k crossing (11 Oct 2026, analyst2)
Data: /opt/sw/radar/pool_paths.jsonl (11 Oct refresh, true crossing times). Scripts: /opt/crew/work/analyst2/it70/a-c.py

Universe: v4 hooked pool, fee 3000 (H3000 family, quote WETH), born small (cap_first < 150k, median 34k), first crossing of $300k cap at age 10-60 min (t300k_s 600-3600), depth to sell into at the crossing >= $80k.
Result: P(peak >= $1M) = 92% (n=125 depth>=80k, all ages 300+ s); with age >= 10 min: n=107. P2M 68%, P600k 92%.
- Stable: quarters by birth time 68/84/100/79%; 21 Oct-10 Oct daily cells 45-100%, no dead day.
- Not a jump: 300k -> 1M takes median 2544 s (q10 1341 s, q90 3335 s), 0 of 100 under 60 s. Depth at the 1M print median $182k (q10 $164k).
- Losers (7 of 107) still peaked $500k-850k (1.4-2.3x of a 370k entry).
- Depth matters: depth $40-80k at crossing = 17% (n53, mostly fee-100 hooks); fee 100 hooked all = 9%. Only fee 3000 + depth>=80k works.
- Concurrent crossers in prior 60 min, hour of day, sender count: no usable split.
Caveat: peak-based (max_cap), not a fill sim. Entry fill assumed at 300-400k, exit at 1M = about 2.5-3x gross. Operator (H3000 campaign) behaviour: earlier audit saw 3 of 112 pulled before TP; sellability of outside wallets at 1M is the open question.