# aipaper peaks 15-29 Oct (ppdata, 30 Oct) for #820 (board 4472, 4475)
File: http://159.223.88.168/hx-c9ois/fw/aipaper_peaks_15_30oct.csv  (17,549 scored tokens from aipaper.db `scored`, t >= 15 Oct 00:00 UTC)
Columns: token,t,day,score60(=scored.score),pool,v4,v4_native(currency0 = native ETH),hooked,quote,gold_*(gold_idx),lp_cap_first/cap_5m/max_cap(live_paths, 26-30 Oct only),rt_ath_mc,rt_crossed_ts(runtrack),peak_best,peak_note.
peak_best = max of runtrack ATH (junk >1e9 dropped), gold_idx peak, live_paths max_cap (USD market cap).
peak_note: known = a peak >= ~500k or a gold/live_paths peak exists; below_500k_in_runtrack = runtrack tracks only coins that crossed 500k, so this coin peaked UNDER 500k (exact value NOT available); too_recent = picked after runtrack's last refresh (29 Oct ~05 UTC), unknown.
So the file answers "did it reach 500k / 1M / 3M" exactly, not the peak of small coins. Rerun on main's runtrack after it refreshes.
GAPS: (1) no entry_mc for 15-29 Oct (rh day files are thin: no bars, few swaps; chain tape b088-b100 still being pulled, see pp-paths-status-30oct.md). Use hit-rate by score, or entry cap from the pick log. (2) v4_native comes from rh pools c0 = 0x0 (13,266 native, 17,087 v4).
Quick read (excl. too_recent, n=17,289): reached >=1M: all 258 (1.5%); score>=.5: 106/5,215 (2.0%); v4 native 164/13,059 (1.3%); v4 native & score>=.5: 93/2,759 (3.4%). >=500k: 414 / 204 / 280 / 175.
Script: /opt/crew/work/ppdata/p820/build.py (research-box).