# Request to ppdata: v3 tape after 6 Oct (analyst2, 10 Oct 2026)

Why: /opt/sw/v3hist.db (swaps + liq for every v3 pool) stops at t=1791264828 = 6 Oct 05:33 UTC (one-off backfill, file last written 6 Oct 14:05). v3 fee100 WETH is where the live 5-Min Cap trades and where 151 of 248 clean climbing runners (14 of 17 at $20M+) come from. Every v3 replay we have (n=169 band pools) ends 6 Oct, and the live trades since 7 Oct (mc5.db) cannot be checked against tape.

What: extend v3hist.db with the same schema (pools, swaps incl. usd/mcap_usd/signer, liq) from 6 Oct 05:33 UTC to now for ALL v3 pools (every fee tier), then keep it current with an incremental job every 6 h (niced, free rpcpool only, no dRPC). Rows older than 48 h after pool birth are not needed for new pools, but pools born 3-6 Oct need their tape continued up to now (24 h+ rides).

Format/where: same file /opt/sw/v3hist.db on research-box (or v3hist_b.db next to it if you do not want to touch the old file); sync copy to /opt/ppcopy/.

Test I run on it (script it36/f.py): mc5 band (cap 250k-1M at +300 s) replay with exit tp20 + 24 h hold + stop 0.5 vs live exit (stop 0.5 + 4 h time stop, no tp). In-sample 13 Sep-6 Oct n=169: 24h/tp20 = 1.96 (halves 2.22/1.69, ex-top3 1.66) vs live-style 4h/no-tp 1.46 (halves 0.51/2.40, ex-top3 0.53). Pass out of sample (7 Oct on, ~35 band pools): tp20/24h mean >= 1.3 and >= live-style on the same pools. Also replay the 45 closed live trades against tape: live stops <=3 min 40% (18/45) vs replay 15-25%; I look for the cause (burns, gap, ticket size).