# Audit: h100 (hooked fee-100) + busy60 (unhooked fee-100, control Z) - auditor, 9 Oct 2026
**Verdict: FIX / NOT VERIFIED YET** (sample 12 trades, no live overlap, sellability untested, pulls still invisible). The 3x/4x prints are real.

Data: /root/tm/h100/trades.csv (49 rows), busy60 (40). Scripts: impact.py, outside.py (research-box /opt/crew/work/auditor).
## 2. Outside sellers WITH tx.from (I fetched tx.from by RPC; tape only has router sender)
For every TP exit (h100 A 8, B 5; busy60 Z 16 exits): buyers between entry and exit 70-470 distinct wallets, 6-110 ETH bought, creator share 0%, top wallet 2-10%. Sells at >= 80% of target within +-60 s: 5-106 distinct non-creator wallets, 0.3-20 ETH (our size 0.04 ETH). 29 of 29 pass. So the pump is not the operator wash trading, and our 0.04 ETH exit has real buyers at that price. This corrects fw/tm-audit-1.md (that said "cannot be split").
## 4. Fills
- Exit fill is "first swap that crosses the target, filled exactly at target x 0.97". I walked the real V3 depth at the exit swap for a 0.04 ETH sell: impact 0.2-0.7% vs 3% assumed. So slippage is conservative, size is fine.
- Entry price = spot of the swap at +63 s (post-swap price) then 3%: ok.
- NOT rebuilt by hand from receipts.
## 3. Look-ahead
Entry uses swaps up to entry only (read h100.py). Sign test (sg) uses the whole pool but only flips direction. PASS. Hook byte size 11689 / hi-init filters come from A1's in-sample study: forward test, fine.
## Problems
- 7 of 12 h100 trades are LATE (feed lag > 30 s, live could not enter at +63 s). On-time only: A n=5 mean 2.33x, B 3.10x, C/D 3.88x. Judge by on-time only. busy60 Z: on-time n=13 mean 1.34x, late n=9 2.59x.
- Quiet exits count 0 (correct after TM fix); real pulls need ModifyLiquidity (ppfeeds #162) to be seen.
- busy60 control Z: 14 of 22 unhooked fee-100 pools reach 3x. If that is true of the whole population the edge is the universe, not the filter; arm A (n=2) cannot be judged yet. Kill rule (A mean < Z + 0.3) is the right test.
## 1. Sellability: only indirect (many real sells at the target price). Hook skim/sell block on a fresh wallet not tested -> $5 sell test (Ben decision, #138) or hook code read.
## 5. Replay vs live: no live trade on these. Open.
## To reach VERIFIED: >= 30 on-time trades per main arm, ModifyLiquidity feed, $5 sell on one hooked pool.