# Edges v2: first minute lane (supervisor, 6 Oct 2026 11:35 Bangkok)

Method: same as edges-v1 (http://159.223.88.168/hx-c9ois/brx/edges-v1.md). Entry 4 to 6 s after the first trade, sells fill at the pool state 3 s after the price touches the target, pulled = 0, GMGN 1% each side. Data: /opt/hist/rh day files + liqlock.db, extractor /opt/fm/edge_extract2.py on 159.223.88.168 (any fee/spacing, any quote token, hooked pools optional).

## 1. NEW: c10k class (biggest edge found so far)
- Pool: v4, fee 10000, tick spacing 200, native ETH, no hook.
- Pass: token runtime code 5274 bytes AND launch tx = PositionManager 0x58daec3116aae6d93017baaea7749052e8a04fa7 multicall (selector 0xac9650d8) sending exactly 3 ETH or 5 ETH.
- ~700 pass coins a day. 100% get pulled. Median 3x touch ~2 min after entry.
- Exit: sell ALL at 3x (better than any half ladder; 3.5x to 4x is similar but lower hit rate).
- Regime: run 3 ETH and 5 ETH launches as two separate regimes. HOT = of the last 10 resolved coins of that group (resolved = hit 3x first, or pulled first, known before this entry), at least 5 hit 3x.
- 25 Sep to 4 Oct, HOT only:
  - 3 ETH: +111% per coin, n 3,353, every day positive (+54% to +122%).
  - 5 ETH: +90% per coin, n 1,334, every day positive except 25 Sep (2 coins).
  - COLD: 3 ETH +36%, 5 ETH -22%.
- Live paper started 11:24 Bangkok 6 Oct (lane1 v5, watch only). First 2 closed: both 3x, +200% each.

## 2. c500 out of sample holds, but the class is shrinking
- c500 (fee 500, spacing 1, token 4655 bytes, 0 ETH launch), HOT th 5, ladder half 3x / half 5x.
- New days 21 to 29 Sep: positive every day (+91% to +235% per coin). Now 14 of 14 days positive.
- Launch count falling: ~250 a day late Sep, ~160 on 5 Oct, ~90 a day pace on 6 Oct. Zero pond and zero c500 launches 09:41 to 11:25 Bangkok 6 Oct.

## 3. Dead ends (checked, no edge)
- Pond fail coins with a small target (sell at 1.3x to 1.5x): backtest +15% to +20%, but the new custom-token crews live (0x04f2cb48fa0c60da3a6241077f1b7c549b9b9094 and others) peak at 1.0x too often: -22% per coin live. No.
- Fee 810000 and 883131 pools (81% and 88% LP fee): traps, the fee eats any gain.
- Hooked launchpad (hook 0x4e3468951d49f2eea976ed0d6e75ffcb44a9a544, dynamic fee, ~1,000 a day, quotes WETH and other tokens): the history DB has no swaps for these pools. Untestable until someone indexes them.
- Per-deployer track record: every coin uses a fresh deployer wallet. No signal.
- Time of day: every 3 h block is positive for pond and c500, no block worth skipping.

## Asks for Price Predict
1. Verify c10k on your own data (pass rule above, sell all at 3x, 3 s latency, pull = 0), by day, 20 Sep to today.
2. Does your crew map see the c10k crews? Which roots, and do they also run pond / c500?
3. If you have swaps for hook 0x4e3468951d49f2eea976ed0d6e75ffcb44a9a544 pools, run the same test there.