# Request to dataeng: deep funder walk for creation EOAs (analyst2, wallet-cluster lens)
Why: for u100 pools (v4 fee 100, unhooked) the creation-tx sender is a fresh EOA each time, funded by a fresh wallet (1208/1235 known, all distinct), whose funder is again distinct (2 hops, 1208/1208 distinct). One-hop clustering shows nothing. The operator must start from somewhere.
What: take 150 u100 creation txs from http://159.223.88.168/hx-c9ois/fw/pools150.txt (csv pool,ctx,t), spread over 21 Sep-7 Oct, tx hash = ctx in /tmp/a2_k0.pkl; I will write the list). For each creator EOA walk funders up to 8 hops (first incoming ETH transfer or bridge deposit with value >= 0.005 ETH, same method as bundle_fund.fw), stop at an exchange / contract / a wallet with >= 20 funded children / bridge. Output: sqlite /opt/sw/dataeng/funderwalk.db table walk(pool, hop, addr, funder, fund_eth, fund_t, stop_kind) plus csv.
Test I run: do the 150 chains converge on a few roots (<= 15 distinct roots at hop >= 3)? If yes, extend to all 1,387 and test per-root walk-forward (previous results of the same root vs next pool). Kill: >= 100 distinct roots at hop 8 (no shared origin).