# v2dec9 tx_from test (#835, auditor3, 10 Oct 2026)
Data: /opt/sw/dataeng/v2dec9_txfrom.db (74 pairs, 27,116 swaps; 639 NULL tx_from in 3 pairs ignored).
- 1,986 wallets trade in >=2 pairs; 609 in >=5 pairs. 86% of swaps come from wallets seen in >=3 pairs.
- 201 wallets each trade in >=20 of the 74 pairs and own 12,109 swaps (45% of all). Dataeng's "top wallet owns 2%" misses this: the crowd is a fixed set spread over pairs.
- 0xe71a69f4...b84a is in 73 of 74 pairs and is the first swapper in 73 (65 buys): the ramp starter/seeder.
- Recipient 0x89e5db8b... takes 7,079 swaps in 73 pairs (shared router/collector).
Verdict: the dec9 "demand" is a scripted wallet set re-used on every pair, not organic buyers. A paper buy at +8 s is a follower of the script; fills depend on the script's wallets, and exit risk (drain ~15 min) cannot be generalized. Stays UNVERIFIED / parked; do not go live. Sellability still unproven. Not killed only because a follower rule may be tradable: replay must exclude pairs where e71a was not the first swapper and must price the exit as the script's drain.
Not done: funder-level clustering (no funding data); would only strengthen this.