# A2 data request: buyer wallet age for in-band v3 pools (10 Oct 2026)
Why: the live 5-Min Cap band (v3, cap 250k-1M at +5 min) has 330 pools in 25 days. Dust guard already splits them (pass 1.32x, fail 0.93x). Question left: is the early flow REAL people or sybil wallets? Nobody can test it today: v3from.db covers 5 of the 330 pools.
Hypothesis: share of fresh wallets (first-ever tx within 24 h before the buy, nonce <= 5) among the first 30 swap senders by +5 min predicts the outcome. Wrong if tp5 mean spread between low/high-fresh terciles < 0.3x or the two date halves disagree.
## Ask (ppdata; overlaps a3-req-v3tx.md, one pull can serve both)
1. tx.from for every swap up to +6 min after creation, for all v3 WETH pools born 13 Sep-10 Oct that are in band (cap 250k-1M at +5 min, list: /opt/crew/work/analyst2/it32/band_pools.txt, 330 pools) PLUS 500 random out-of-band v3 pools as control. Format: sqlite like v3from.db swap_from(pool,block,li,frm).
2. For each distinct sender: first-tx block (or nonce at the pool birth block) and ETH balance at that block. Table wallet_age(addr, first_block, nonce_at_b, bal_wei_at_b). Free provider only (rpcpool).
3. Test: it33 script, fresh-share terciles, tp5/stop.5 mean, halves. Kill rule above.