# Runner map, 30 days (analyst2, 10 Oct 2026) for #363 and #381

Window: tokens born 13 Sep to 7 Oct (WETH/native pools, table e1-runners.csv, ATH flagged "ok", glitch prints removed) plus USDG/SUSD v4 tokens born 20 Sep to 9 Oct (r6tape swaps, ATH from median-of-5 prints, gated on >= $40k traded volume, same floor e1 runners have). Scripts: /opt/crew/work/analyst2/it30/ (map.py, usdg.py, late.py).
Not covered: tokens born 8-10 Oct for WETH/native (e1 copy ends 8 Oct) and Pons v2 / Bankr style venues outside these tables. Weekly update needs ppdata to refresh e1-runners.csv (asked below).

## 1. Runners by pool type (1M+ = 979 tokens; 3M+ 487; 5M+ 342; 10M+ 206; 20M+ 130)
| Pool type (venue / hook / fee / quote) | 1M+ | 3M+ | 5M+ | 10M+ | 20M+ | share of 1M+ | median min to $1M | we trade or test it? |
|---|---|---|---|---|---|---|---|---|
| v4 NO hook, custom fees (10000, 2500, 75, 20000, 100xx...), native | 143 | 112 | 91 | 69 | 46 | 15% | 495 (8 h) | NO. Nothing. Biggest source of big runners |
| v3 fee100 WETH | 142 | 62 | 49 | 26 | 15 | 15% | 43 | YES: live 5-Min Cap (40 trades), a1mc, P204, v3-005 |
| v4 hooked 0xa80 fee3000 WETH (H3000 family) | 122 | 55 | 25 | 5 | 0 | 12% | 64 | YES: h3k (13 trades) |
| v2 WETH | 111 | 59 | 42 | 16 | 6 | 11% | 38 | YES: V2RAMP 64 trades, V2DEC9 |
| v4 NO hook, USDG fee-any | 91 | 45 | 34 | 20 | 12 | 9% | n/a (no tape clock) | NO (research row A2-USDG4 stage 1, no rule) |
| v4 hooked, other hooks, fee custom, native | 87 | 30 | 15 | 8 | 7 | 9% | 10 | only partly (h3n covers one kit) |
| v4 hooked 0xe5e7...044 fee0 native (ONE hook, 68 runners) | 68 | 13 | 7 | 1 | 1 | 7% | 80 | NO |
| v4 hooked other fee0 native | 40 | 23 | 7 | 4 | 4 | 4% | 4 | NO |
| v4 NO hook fee100 WETH+native | 42 | 1 | 1 | 1 | 1 | 4% | 44-46 | YES: busy60, fast5, kitcode, d0fd (too few runners above 1M anyway) |
| v3 fee10000 WETH | 19 | 10 | 7 | 6 | 4 | 2% | 540 | NO |
| v4 hooked other, USDG | 16 | 12 | 9 | 6 | 4 | 2% | n/a | NO |
| v4 hooked 0xc4 (native, custom fee) | 12 | 12 | 12 | 9 | 6 | 1% | 17 | NO. 12 of 12 reach 3M+, 6 reach 20M+ |
| v4 hooked other, WETH | 11 | 10 | 9 | 7 | 6 | 1% | 1066 | NO. 9 of 11 reach 5M+ |
| v4 hooked 144 fee100 WETH | 9 | 1 | 0 | 0 | 0 | 1% | 105 | YES: h100 |
| v4 nohook fee0 SUSD / USDG | 10 | 10 | 10 | 8 | 6 | 1% | n/a | NO. 10 of 10 reach 5M+ |
| rest (33 small buckets) | ~50 | | | | | 5% | | mostly no |
Full table: it30/map.out.

## 2. What the table says
- **Where we are blind (share of $1M+ runners):** about 55%. No bot or test watches v4 no-hook custom fee native (15%), USDG/SUSD (10%), hook 044 fee0 (7%), other hooked native/WETH/USDG (about 17%), v3 fee10000 and the long tail.
- **Where we are present:** v3 fee100, v2, H3000 a80, fee100 hooked/unhooked cover about 45% of runners, but they are the small ones. Of 5M+ runners (342) the covered pools hold: v3 fee100 49, v2 42, a80 25, unhooked fee100 1, hooked 144 0 = about 117 (34%). The other 66% of 5M+ and 80% of 20M+ sit in pools nobody trades.
- **Big-runner ranking by pool type (share that goes on to 5M+ after 1M):** v4 hooked 0xc4 100%, SUSD/USDG nohook fee0 100%, v4 hooked other WETH 82%, v4 nohook custom native 64%, v3 fee10000 37%, v2 38%, v3 fee100 35%, a80 H3000 20%, 044 fee0 10%, unhooked fee100 3%.
- **Two of the families we test hard cannot make a big runner:** unhooked fee100 (42 runners, 1 ever above 3M) and H3000 a80 (0 of 122 above 20M, 5 above 10M). They are fine for the 1M tier ("first seconds, ride to 1M") but cannot serve the 5M-20M hunt.
- **Biggest single gap:** v4 NO hook, custom fee, native: 143 runners, 91 above 5M, 46 above 20M, median 8 h to reach $1M. Fee tiers are many (10000: 15, 100: 14, 2500: 10, 0: 6, 75: 4, 20000: 4, 10028/10032: 6 ...), so no fee/hook key exists; the only handle is the v4 Initialize event (the v4init feed) plus first-hours behaviour. These coins are slow (8 h to $1M): a late-hunter lane, not a seconds lane.
- Live = 40 trades, all v3 mc5. 15 paper tests sit on 7 pool types; 0 on the 8 types holding the most 5M+ runners.

## 3. Misses, what caught them (#381 part 1: late hunters)
All 5M+ runners (342): P(5M|1M)=34%, P(10M|3M)=41%, P(20M|5M)=38%, P(20M|10M)=64%. Once it passes 10M it mostly goes on.
Of all tokens that touch 300k (1,913): 15% reach 5M, 8.8% 10M, 5.6% 20M.
- **Age at 300k is the one clean separator** (first-300k time minus pool creation): under 15 min: 5M rate 9%, 10M 3%; 15-60 min: 10% / 5%; 1-6 h: 19% / 14%; 6-24 h: 48% / 35% / 24% at 20M (n=134); 24-100 h: 32% / 26%. Holders at 300k and new holders in the hour before show no signal (n=20-30 per bin, flat).
- **Caveat that matters:** for aged coins the 300k-to-1M step is mostly instant (median 0 min, p75 0 min: both touches stamp the same sample), so these are re-pricings/step jumps, not climbs you can enter at 300-400k. For young 5M+ runners the 300k-to-1M gap is median 11 min (p25 0, p75 39 min). So a 300-400k entry window exists for roughly half of the young runners, and for almost none of the aged ones. Not testable as a rule until measured on swap tape per runner (ask below).
- **Exit side:** peak is a median 6.5 h after the 300k touch (p25 2.2 h, p75 34 h). After the peak a 5M+ runner keeps a median 68% at +24 h, 36% at +72 h, 28% at +7 d. So a long hold with a loose trailing stop is not hopeless; sell-to-peak is rare to time, sell inside the first day is the zone where most of value lives.
- **Hourly-kline exit sim (inconclusive):** only 256 of 1,794 tokens have hourly klines, 99 entered at a 240-500k close, 12 reach 5M. Rules (7% round trip): hold-to-end mean 1.6x median 0.71; TP3 1.70x, TP5 2.16x, trail 50% 1.59x; 47% of holds end below 0.5x. Survivorship (kline coverage favours tracked winners) and hourly resolution make this unreliable; do not use. Direction only: a plain late entry bleeds half the time, so the late hunter needs a pre-entry filter, not just an exit.

## 4. What would catch them (next steps)
1. **v4 no-hook custom-fee native (143 runners, 8 h to $1M).** Watch via the v4init feed (every v4 birth). Candidate: enter after a base (age >= 1 h, cap >= 300k, 1-h volume rising), tp ladder with loose trail. Needs swap-level tape for these 143 plus their non-runner siblings to measure hit rate. Requested below.
2. **0xc4 hook (12 of 12 above 3M) and "other hooked WETH" (9 of 11 above 5M):** 23 tokens in 25 days, tiny n. Find out what the 0xc4 kit is (hook code, who launches) and whether the init event is distinguishable (hook flag bits in the address). Candidate for a pure "kit fingerprint" arm like 0xcc.
3. **Hook 0xe5e7...044 fee0 native (68 runners, mostly 1-3M):** 1M tier hunt only (13 above 3M). A first-minutes arm like FAST5 on this kit.
4. **SUSD/USDG fee0:** 10 of 10 above 5M at tiny n; still no pre-entry rule (see A2-USDG4).

## 5. Data requests (posted)
- ppdata: refresh e1-runners.csv to today (born 8-10 Oct) incl. USDG/SUSD tokens (e1 currently skips them: 126 runners missing), plus a quote column. Weekly.
- ppdata: swap tape (r6-style) for the 143 v4 no-hook custom-fee native 1M+ runners AND all other no-hook custom-fee native pools born 13 Sep onward (controls), to test the late-hunter entry.

## Correction 10 Oct (from analyst3 a3-blindspots-10oct.md)
The "v4 no-hook custom-fee native" blind spot (143 / 91 5M+) is overstated: 95% of its 5M+ are one-print jumps (not enterable); only ~18-22 are real climbs. USDG fee-0 "10/10" is mostly one 7 Oct batch (n_eff ~2-3). Treat both rows as non-targets until the tape arrives.