# A3 idle run 10 Oct (night): who/what in the pool types we do not watch
Data: /opt/ppcopy e1-runners.csv (7,019 tokens that touched $300k, to 7 Oct), tokens.db pools (245k v4 births since 13 Sep), it30 usdg_run.json, poolout.db. Scripts: i1-i7.py, this dir. "Jump" = first $300k and first $1M within 1 min (one print).

## Iterations (10)
1. **Reality check on the biggest blind spot (v4 no-hook custom-fee native, 1,718 tokens at $300k).** Lens: who is on the other side / is it enterable. 678 tokens reach $5M+; **95% (647) are one-print jumps** (median buy volume $115k at a $23M cap; vol/cap 0.5%). Only 31 are climbs, and only 18-22 of those have >= $300k-$1M of buy volume. Runner-map counts for this row (143 / 91 5M+) overstate what a bot could ever enter. Population base: 94.6k no-hook custom-fee native births since 13 Sep, 421 touch $300k (0.45%), 141 reach 5M (0.15%), 3 for fee 10000 (26k births). MATURE as map correction, not as a rule.
2. **Fee tier / project-type as intent signal.** Share reaching 5M among 300k-touchers (jumps included): fee 10000 6.5%, 2500 12.5%, 9000 34%, 20000 35%, 50000 22%. Looks like a launcher-intent signal but 95% of those 5M are jumps (see 1), and sim mean by fee is 1.14 (<2500), 1.67 (10000), 1.38 (20-60k) with all subject to the stale-price problem (5). DISCARD until tape.
3. **Event clustering (USDG/SUSD fee-0 "10 of 10 reach 5M").** Lens: wallet/operator cluster. 8 of the 17 fee-0 USDG runners were born on 7 Oct (7 of them between 00:00 and 08:00 BKK, ATH median $16.7M); the other 9 are spread over 5 weeks. n_eff of the fee-0 row is about 2-3 launch events, not 10. Do not size a test on it; look at the 7 Oct launcher/batch (who deployed 8 pools within 7 h?) instead. REQUEST below.
4. **Enter at $300-350k, hold with TP3, non-jump coins, no-hook custom-fee native (n=786).** e1 sim, fee both sides: tp3 1.50, tp5 1.45, halves 1.42/1.57, weekly 0.99-1.86, pessimistic (bags x0.7) 1.21. Looked positive. See 5.
5. **Trap check on 4 (depth/stale prices).** For fee 10000 (n=365): median swaps 190, last trade a median 24 days before the data end, cap_7d_after_peak/peak mean 1.39 (>1 is impossible if peak is the max). Prices on these thin pools are unreliable; the "fee 10000 coins never collapse (6% bags vs 79% at fee <= 2500)" result is a stale-price artifact. DISCARD as signal; the direction of 4 stays UNVERIFIED until tape (request 1).
6. **Regime lens.** Weekly sim mean of 4 is 0.99-1.86, no trend; hour/weekday real-rate flat (0.3-3%, n too small). Consistent with analyst2 (#23). DISCARD.
7. **Exit lens, enterable big runners (n=22: climb, >=5M, buy vol >= $300k).** Multiple from $350k to peak: median 32x (p25 19x, p75 66x); time to peak median 115 h (p25 14 h); cap 24 h after peak keeps median 52%. These are multi-day holds; TP3-5 would leave 85% of the move. A trailing/ladder exit belongs to this lane, but 22 events in 5 months (0.15/day) is a thin lane. MATURE only together with a pre-entry filter (none found).
8. **Inversion: what the collapsed coins share.** Bag rate (cap_last < 0.3x entry) by age, fee, swaps: swaps <= 500: 2% bags vs 60% for > 10k swaps; same stale-price confound as 5. DISCARD.
9. **Stock-quoted hook 0x4e34 pools (the "unknown venue", dataeng2 #416).** Lens: data we do not have. tokens.db: 57,488 pools since 13 Sep (24% of all v4 births, 1-2k/day, one token each, dynamic fee 8388608), quoted in 20+ different tokens (WETH 12.9k, stock tokens e.g. SPCX 6k, USDG 3.7k, native 3.3k, others). Only 26 of them ever touched $300k in e1 (WETH/native quoted; median peak $0.88M). No price series exists for the stock-quoted and token-quoted ones, so nobody knows their runner rate. REQUEST 2.
10. **Deployer / funder / name lens.** creator is empty for 82% of tokens in e1 and names missing for 76% of USDG runners, so wallet and name tests (serial deployer, funded-by, ticker lookalike such as AAPL $309M, PONS) cannot run. REQUEST 3.

## Data requests (posted)
1. ppdata: swap + liquidity event tape (r6nat2 format: ev with amounts, px, liq, sender, tx) for **v4 no-hook, native-quoted, fee not in {0,100,8388608}** pools: every one that touched $300k since 13 Sep (about 420, list = e1 rows) plus a matched control of 1,500 random pools of the same type born in the same window, from creation to +72 h. Test: enter at first $300-350k touch (price from the tape, size 0.05 ETH with depth impact), TP3/5/10 + trailing, LP-pull = bag. Success = mean > 1.15 net of fee both sides on both halves with exit depth >= 2x position at the exit.
2. ppfeeds/dataeng: USD price series (open/high/low/close per 5 min, plus quote-token reserve) for the **hook 0x4e34 pools** in stock-/token-quoted pairs (57k pools since 13 Sep; at least the top 2,000 by volume), priced via the quote token's own pool or stocks.db ref_px. Format: sqlite like e1 (pool, token, quote, created, first_100k/300k/1M/5M, peak, swaps, buy/sell usd). Test: runner rate per quote token and per first-hour feature; is it spam or hides runners.
3. ppdata: for all v4 births 13 Sep+ (not only e1): deployer = tx.from of the creating tx, deployer wallet first-seen date, first funder address + label (CEX/bridge/fresh), number of earlier tokens by this deployer and their best peak, plus name/symbol (ERC20 name()/symbol(), 82% missing). Format: sqlite table deployer(pool, token, deployer, dep_first_seen, funder, funder_label, n_prev_tokens, best_prev_peak, name, symbol). Test: 5M-climb rate by deployer age (< 1 d vs older), by funded-by-CEX, by ticker = stock/major-coin lookalike; target lift >= 3x on n >= 40.