# Early ramp entry at +100 s on v3 (SEALS idea), analyst2, 15 Oct 2026 (board #2511)

Replay: v3live.db, 6-14 Oct, 1095 v3 pools born <30k (script ramp15.py). Entry at first print after +100 s, pool ETH >=1, >=8 distinct buyers by +100 s, cap at +100 s 10-30k. Pulls priced (liq remove >=50% of pool = value 0 if still held). tp 3x / stop 0.5.
- n=125, mean 1.51 at 7% slip, 1.27 at 15%, 1.06 at 25%. Excluding top 3: 1.47. Zero-value (pulled) 19 of 125.
- By day (7% slip): 6 Oct 1.94, 7 Oct 2.43, 9 Oct 1.11, 11 Oct 1.32, 12 Oct 1.45, 13 Oct 1.01 (n=23). Edge decays to ~1.0 on the last full day, and at 15% slip 13 Oct is 0.85.
- It is NOT a runner catcher: only 2% of entries reach 1M (3 of 125). The mean comes from 3x scalps in the first minutes, i.e. the first-minute family (21 killed). tp5 1.82 / tp10 2.3 at 7% slip but same day decay (13 Oct 1.06).
- SEALS-type: cap 2.7M at +5 min; its cap at +100 s was not distinguishable from the thousands of other 10-30k pools by buyer count (buyers >=8 passes ~all pools: the filter does nothing, 125 of 135).
Verdict: do not paper-test. Edge is slippage-fragile and gone on 13 Oct; no feature at +100 s separates the 2% that run. Ledger: killed (early-ramp +100 s).

## Why live bought 0 of 11 runners on 13 Oct (from a1 miss review miss-2026-10-13.md)
10 of 27 runners = v4 native kit (venue not watched by mc5), ~10 = v3 organic with cap >1M at +5 min (cap_out by rule), rest = v4 jump pools / hooked weth (sellquote technical). No bought runner is a pricing or exit failure: the repeating reasons are (1) venue unwatched by mc5 (v4 native kit), (2) cap_out. Fix for (1) stays blocked on v4tape 90-min extension (#577-#580, ~16 Oct 21:00 UTC); (2) is deliberate (SEALS lesson: 5-min top buys lose 0.35x).