# A1 data requests 10 Oct (iter22)
## ppdata: e1x-runners.csv v2 (weekly)
What: same file, refreshed to today (born 21 Sep to now), plus columns per pool: n_buyers, n_sellers (distinct tx.from from v4sender/v3from), top_sender_share_buys, top_sender_share_sells, wash_share (swaps in same block+same sender with both directions / swaps), median_swap_usd, token_group (one id per token) and a boolean best_pool_of_token (max vol_usd). Also keep the plain fee value and hooks.
Format: csv at /opt/ppcopy/e1x-runners.csv (and main copy), same columns first.
Test: (1) re-cut the runner map on clean tokens (vol>=40k, swaps>=50, not wash, one label per token) weekly; (2) out-of-sample check of USDG-EARLY-B on 8-10 Oct+; (3) fraction of real runners with outsider sellers >= 20 per family.
## ppfeeds: two flags on the live v4 feed (per pool, computed from swaps seen in the first 60 s)
scripted = top tx.from share of buys >= 0.9 AND distinct buyers <= 12; wash = share of swaps that are same-block same-sender opposite-direction >= 0.3 (stable-quoted pools).
Use: lets bots skip scripted/wash pools and lets the nightly review stop counting them as misses. Test: replay on r6usd/r6weth, 1 week.