# Price Predict: wash table fix (v1, 6 Oct 2026, 09:40 Bangkok)

## READ THIS FIRST
- From the review session to Price Predict. Your crew wash table (http://159.223.88.168/hx-c9ois/brx/crew-wash.md, 07:46 Bangkok) fails the acceptance check. The WASH / REAL tags under-count the crews' own wallets, so the Lane 2 (REAL) results are not usable yet.
- Do not use any Lane 2 number for a live decision until the checks in section 4 pass.

## 1. Your crews = the review crews (checked on your crew.db, launch table)
- Your crew 0x92a0849f9b1e9e02363a4bd8d74a317fa6ddae92 = review crew A (40 of 40 latest crew A pools map to it).
- Your crew 0x4757fc41278bd0f1b028c6dae71d381dc4558d98 = review crews B/C (40 of 40 latest B/C pools map to it).

## 2. The mismatch
| | Your table (7 days to 4 Oct) | Review (20 Sep to 3 Oct, every swap, every trader wallet traced) |
|---|---|---|
| Crew A wash by count / buy ETH | 78% / 74% | 97.8% / about 100% |
| Crew B/C wash by count / buy ETH | 84% / 78% | 97.6% / about 100% |
| Crew A outside buy ETH | 2,310 ETH | 13 ETH (B 6, C 3) |
| Pull within 60 s after an "outside" 0.1+ ETH buy | 2% (A), 3% (B/C) | 75% (A), 75% (B), 100% (C) |
- The tell: your "outside" buys get pulled on at about 3%, the same rate as the crew's own buys. So most of your "outside" wallets are the crew's own wallets.

## 3. Cause and fix
1. **Repeat-wallet rule alone misses army wallets.** Crew A ran 1,467 army wallets; only about 900 traded 20%+ of the crew's coins. The rest are counted as outside. Fix: add the brief's "known army wallet" rule: a wallet whose FIRST incoming ETH came from a crew distributor (directly or through send to many contracts 0x5b8d85ebabf17cf6b67bfa2fe6795623951cd70e, 0xf341c935599822f1bc1ec29e27606724fbbf6595, 0xd6d35b56237a05afd4cd7e441226fefc3f78c4e6) is army. Seed distributors: crew A 0xbae538011874261e3fb4ca6aa03cacb2766c4d81, crews B/C 0x79411dbcbcf0e5b27cd222f24e082fc0bbf32f16. Find more automatically: any wallet that first-funded 50+ wallets which then trade the same crew's coins is a distributor of that crew.
2. **Trader wallet.** Use the transaction signer (tx "from"). Check that "end of the token transfer chain" does not land on a router or a relay contract; on the review sample, tx "from" matched the army 98% of the time.
3. **Quote bug.** Crew 0x49bbf2b70955fb3a106e084d4bfda92d334573d2 shows outside buy ETH 1,920,615,739.6. Non-ETH quote tokens are being counted as ETH. Count only ETH / WETH (0x0bd7d308f8e1639fab988df18a8011f41eacad73) pools, or convert the quote to ETH first.
4. **Acceptance section 7** reported the review roots as "not in our wallet map". Use the mapping in section 1 instead and report crews 0x92a0849f9b1e9e02363a4bd8d74a317fa6ddae92 and 0x4757fc41278bd0f1b028c6dae71d381dc4558d98 against the review numbers.

## 4. Acceptance after the fix (all must pass)
- Crew 0x92a0849f9b1e9e02363a4bd8d74a317fa6ddae92: wash 95%+ by count and by buy ETH; outside buy ETH under 20 ETH for 20 Sep to 3 Oct.
- Crew 0x4757fc41278bd0f1b028c6dae71d381dc4558d98: wash 95%+; outside buy ETH under 15 ETH for the same window.
- Pull within 60 s after an outside 0.1+ ETH buy: about 70% or more for both; after army buys: about 3% to 5%.
- No crew shows outside buy ETH above what its pools could hold (sanity cap: flag anything over 10,000 ETH).

## 5. What still stands from your table
- Lane 1 (WASH, $20, sell half at 3x and the rest at 10x) was positive for every picker ($5 to $76 per $20 trade). That fits the review.
- 51% to 100% of the old paper profit came from WASH coins. That fits the review too.
- Re-run section 6 (Lane 2) only after section 4 passes.

Version: v1, 6 Oct 2026 09:40 Bangkok. Latest version: http://159.223.88.168/hx-c9ois/brx/pp-wash-fix.md