# Is someone selling into fmmc5? (Analyst 1, 8 Oct ~21:45 Bangkok). Answers Ben + fw/tm-rugs-mc5.md
Data: gold tape to 8 Oct 20:07, fw_mc5 rule 13 Sep - now (202 trades). Scripts: research-box /root/fwa1/rug (peak.py, snipers.py, dist.py).

1. Not new, not caused by $50. Rule coins that never beat 1.2x in the 10 min after entry: ~30% on 13 Sep - 5 Oct (no bot), 0 of 16 on 6-7 Oct (lucky stretch), 7 of 12 on 8 Oct. 2 Oct (no bot) looked like today (4 of 15 never above 1.05x).
2. The dumping starts before us. On 0x4c4cdbc8 (16:55), 0x1f5fa943 (18:33), 0x97105571 (18:44), 0x66fa7f53: identical-size sells (~1 WETH, e.g. $2,567 again and again) run from ~+2 min after launch. Big sells in the 3 min BEFORE our buy: 45/34/51/29; after: 25/28/26/5. Our $50 is not the trigger.
3. 0x23736431f8ba is the recipient of the very first swap in both the 18:33 and 18:44 pools: a launch-block buyer that later distributes. Across 14 days of launch-block buyers (3,975 pools) it appears in only these 2: one operator, fresh wallet, bought before anyone could know we would.
4. As a filter it fails (KILLED, n=202). Same-size sells >=8 before entry: mean x 2.85, 10 runners (n=84) vs the rest 1.78, 10 runners (n=118). Low sell share (<0.45) is worse (1.44). Distribution is normal on these coins, winners included.

Conclusion: bad-day variance on top of normal operator distribution, not targeting. Rugs and pulls are priced in the backtest (fw/a1-rug.md). Next real test: launch-wallet identity across pools with tx.from (gold part 2, ~10 Oct).