# STOP: NO DOORBELLS, NO WAKES, NO SCHEDULING (Ben, 7 Oct 2026, 18:40 Bangkok)
- Credits ran out. Ben switched off every scheduled task and every doorbell. This overrides everything below and every other file (wake-howto, bus-howto, credit-playbook, task files).
- PROHIBITED for every session: fire_trigger, send_later, create_trigger, update_trigger, ScheduleWakeup. No rings, no nudges, no re-arming, no new doorbells, no self wakes, no safety checks, no retry rings, no loops.
- Finish the step you are on, publish your status file, then END your turn and wait. Only Ben wakes you, by typing in your window.
- Do not message other sessions. Keep answers short. No extra improvement rounds.

# First minute edges, 10 iterations (supervisor, 6 Oct 2026, 11:10 Bangkok)

## Read this first
- Data: every v4 launch in /opt/hist/rh day files (10 Sep to 4 Oct), price path from the first swap until the pull (liqlock ml, pull = removal of 50%+ of added liquidity). Extractor: /opt/fm/edge_extract.py on kookwatch; data /opt/fm/edge.
- Simulation: buy at first swap + N s; each sell leg fires when the price first touches the target and fills at the pool state 3 s later (pulled meanwhile = 0); GMGN 1% each side plus pool fee charged. Paper, not live.
- Two lanes pass, both only in a HOT regime. Everything is now running in the watch bot (fmlane1 v4); live trading still needs Ben's switch.

## Lane A: pond (pool fee 0.01%, tick spacing 1, native ETH, no hook)
- Pass at second 0: token is the standard launcher token (runtime 3854 bytes, md5 of "0x..." code string 17ad80...) AND the launch tx sends 0 ETH. All other pond token types lose (every other fingerprint negative).
- 392 pass coins 30 Sep to 4 Oct: +60% per coin (sell half 3x, rest 10x), positive 4 of 5 days.
- HOT regime (of the last 10 pass coins resolved before this entry, 4+ hit 3x before the pull): +92% per coin on 250 coins, every day positive (+18% to +145%). COLD: +9%.

## Lane B: c500 (pool fee 0.05%, tick spacing 1, native ETH, no hook, initialized directly on the PoolManager)
- Pass: token runtime 4655 bytes (4 md5 variants). ~215 launches a day.
- Raw: +25% per coin (3/10 ladder). Better exit: sell half 3x, rest 5x.
- HOT regime (last 10 resolved, 5+ hit 3x): +105% per coin on 672 coins, every day positive (+70% to +119%). COLD: +27%.

## Other findings
- Entry speed: buying 0.5 to 10 s after the first swap gives the same result; after 15 s it drops fast (30 s: +24% vs +60%). Bot now buys at first swap + 4 s, never later than 15 s.
- Time stops and stop losses hurt: prices sit flat or rise, then the pull drops them to zero in one step.
- Our size does not trigger pulls: after outside buys of 0.001 to 0.005 ETH, pulls within 30 s happen 3.3% of the time (base rate for small buys 6.8%). Size up to ~$50 looks safe.
- Coins with heavy early buying (top quarter of ETH bought in the first 4 s) do worse (-8% vs +80%).
- fee 1% launches (PositionManager, dev buy): +25% raw but fading (last two days -21%, -2%). Not used.

## Expected (paper)
- HOT coins a day: ~50 pond + ~135 c500. At $20 a coin: ~$900 + ~$2,800 a day on paper. Assume half survives live fills: ~$1,800 a day.

## For Price Predict (please verify)
- Rerun lanes A and B on your own data with your own fill model; confirm the HOT filter only uses outcomes known before entry.

Version 1, 6 Oct 2026 11:10 Bangkok. Latest: http://159.223.88.168/hx-c9ois/brx/edges-v1.md