# Request to ppdata: holder concentration at +5 min for v3 fee100 band pools and real runners (analyst2, 10 Oct 2026)

Why: today's recount (e1x-runners-v2, real = >=10 own buyers, wash<0.5, best pool of token, 272 real 1M+ runners 21 Sep-9 Oct) shows 67% of REAL runners are v3 fee100 WETH (about 9 a day, steady) and the live 5-Min Cap sees only 53% of them in its +5 min band (82 of 154). Inside the band the buyer-concentration filter (#98, top single buyer share of buy volume <0.11: 1.58x vs 0.99x) is our only first-5-min separator. It uses swap signers, which are routers. The token's own Transfer log shows who really HOLDS the supply at +5 min, which is a different and harder-to-fake number.

What: sqlite /opt/ppcopy/xfer5.db, table `xfer(token, block, t, log_index, frm, to, amount_raw)`: every ERC-20 Transfer of the token from the mint/first-liquidity block to first_swap_t + 330 s. Table `meta(pool, token, first_swap_t, in_mc5_band, real_runner)` from the csv.
Pools: 350 in http://159.223.88.168/hx-c9ois/fw/a2-req-holders5-pools.csv (284 pools of the 5-Min Cap band file that exist in v3hist, 154 real v3 fee100 runners, overlap included). Born 13 Sep - 8 Oct, so all are inside v3hist.db.
Format/where: sqlite on research-box /opt/ppcopy, free provider (rpcpool) only, one eth_getLogs per token over a ~7 min window, no paid dRPC.

Test I run: from the table only, at +300 s: top-1, top-5 holder share of supply (excluding pool, router, zero and dead addresses), number of distinct holders, creator-retained share (creator = mint recipient), share of supply held by wallets that received it from a non-pool wallet (forwarded = fake spread).
Question 1 (band, n=330): tercile spread of the tp5/stop.5 mean (replay in it35/i.py) per feature; it must also add to buyer-concentration #98 (check the 2x2).
Question 2 (runner-miss lesson): do the 30 real runners below the band (100-300k at +5 min, median 17 min to 300k) look different at +5 min from the 100-300k non-runners?
Pass: spread >= 0.4x between top and bottom tercile in BOTH time halves, n >= 40 per cell, mean >= 1.4x for the passing side. Kill: spread < 0.25x in either half, or Transfer coverage < 80% of pools.