# v2ramp buyer tx.from read (#716), testmgr3 10 Oct
Source: dataeng v2ramp_tx.db (111 pairs, 5,013 swaps).
- 2,628 distinct buy wallets. 2,446 appear in ONE pair only; 139 in two; 41 in three (the same trio of pairs a55dd138.., cf44b5fb.., ddc22be3..); one seed wallet 0xe71a69.. in 108 pairs (dust, launcher), 0x0e7761.. in 9.
- So the crowd is mostly distinct wallets (not one bot cycling), but 91/111 pairs show EXACTLY 21 distinct buyers. A fixed 21 is not organic: either bot-generated fresh wallets (volume faker) or a capture cap. Sells all come from wallets that also bought.
- Reading: v2ramp pairs look like a scripted wash/ramp pattern with a fixed buyer count. Sellable at pair level (real sells exist), but "outside seller" check (check 2) is NOT passed: no independent seller. The 2.2-2.4x cluster is likely the script's own ramp, so the edge depends on joining the script before its exit, not on a market crowd.
- Action: v2ramp stays unverified; blocker = auditor2 judgement on scripted-buyer pattern + eth_call sell test (needs state-override, dataeng could not). Not eligible for Ben's live decision.

## #720 answer: not a recorder cap
Recorder (dataeng v2rampTx.py) pulls ALL Swap logs in a 3500-block window, no count limit. In the 91 pairs with exactly 21 buyers, total swaps range 25-48 (avg 37), so swaps vary while distinct buyers stay at 21; pairs with 22/23/9/11 buyers and one with 210 exist. A swap-count cap would not give a constant 21 with varying swaps. Conclusion: the launcher script uses a fixed set of 21 buyer wallets per pair (scripted ramp). Same verdict as above: no organic crowd; check 2 not passed.