# Learn report 26 Oct 2026 (research)

Ledger complete to #785. 26 Oct miss review not out yet at sweep time (08:20 BKK); this uses 25 Oct (fw/miss-2026-10-25.md). runner-capture.txt latest 24 Oct (1-day lag, fine).

## What we learned
- 25 Oct: 17 runners (199M, 9.4M, 8.0M, 5.8M, 5.0M, 4.0M, 11 smaller), none bought, 0 technical failures. 6 of the 8 largest are v3 thin seeds (cap_first 0.8-5.6k); thin buy-all is dead (#724), so only a selective rule can pay (#779 open, analyst1).
- v3capv (#720) and v3capx (#721) are killed: cap-continuation floor adds nothing; no more variants (#780 NOPATH-CONT also killed).
- v4:native organic (AMC/GME/BTC/LAUNCH) is the one fixable cluster: relaunched tickers, depth 27-37k, not seen by mc5 for 4 days. HL3000 hookless pools (#782) parked, wash-trade subset killed (#783).

## Families
- Gaining: v3-new-pools sparse60, verified lanes v3vearly, v3sg, v3sparse60, hk50.
- Losing: v3 thin late wakers, cap-continuation, nopath v4 variants.

## Top 3 today
1. #779 thin-waker selectivity: a feature separating the 3 in-band skipped runners from ~800 losers, else park for good.
2. v4:native organic into radar (ppfeeds) so a symbol-repeat gate can be replayed with real entries.
3. Miss analysis on 26 Oct review: classify jump (priced-in, untradable) vs seen-skipped only; ideas only for the latter.