# #418 peak cap per seen pool (ppdata, 10 Oct)
File: /opt/ppcopy/seen_peakcap.csv (research-box; also on main later via ppcopy). 7,297 rows = every pool in bundle_fund.db `seen`.
Columns: pool, token, launch_b, bundled, quote, peak_cap (USD), cap_flag, src, created_bkk, t100k..t20m (BKK time of first crossing, blank = never).
Coverage:
- 5,238 src=e1x: peak from the chain tape (e1x-runners.csv, 21 Sep-9 Oct 07:00 UTC). cap_flag=thin (4,192) = few swaps, peak is a floor/noisy; ok (1,046).
- 35 src=runtrack_ath: DexScreener ATH from runtrack (older or after-tape pools that runtrack follows).
- 2,024 src=none, peak blank: 1,~800 pools launched before 21 Sep (blocks 61-68M, no tape; runtrack only follows coins that crossed its threshold, so blank mostly = never ran, but not proven) and 250 launched after the tape end (block 84.0M, 9 Oct 07:00 UTC).
Do not treat blank as 0 for the tier test; restrict to src=e1x for a clean sample (5,238 pools, 17 Sep-9 Oct), or ask me for per-pool getLogs on the 1.8k old pools.
Example: python3 -c "import pandas as pd;d=pd.read_csv('/opt/ppcopy/seen_peakcap.csv');e=d[d.src=='e1x'];print(e.groupby('bundled').peak_cap.apply(lambda s:(s>=5e6).mean()))"