# STOP: NO DOORBELLS, NO WAKES, NO SCHEDULING (Ben, 7 Oct 2026, 18:40 Bangkok)
- Credits ran out. Ben switched off every scheduled task and every doorbell. This overrides everything below and every other file (wake-howto, bus-howto, credit-playbook, task files).
- PROHIBITED for every session: fire_trigger, send_later, create_trigger, update_trigger, ScheduleWakeup. No rings, no nudges, no re-arming, no new doorbells, no self wakes, no safety checks, no retry rings, no loops.
- Finish the step you are on, publish your status file, then END your turn and wait. Only Ben wakes you, by typing in your window.
- Do not message other sessions. Keep answers short. No extra improvement rounds.

# Price Predict brief: wash crews, two lanes, size rules (v1, 5 Oct 2026, 22:15 Bangkok)

## READ THIS FIRST
- From: the review session (read only on pickers; it does not write pickers). To: Price Predict.
- One finding changes how every picker must size: most crew coins are traded almost entirely by the crew's own wallets. The prices are real (their ETH is in the pool) but almost no outside money exists, and the crew pulls the pool within seconds when a real outside buyer of 0.1 ETH or more shows up.
- What Ben wants: every pick tagged as WASH or REAL. WASH coins only get tiny buys with sell orders attached (Lane 1). Normal size only on REAL coins (Lane 2). Applies to every picker, not only crew picks.
- Do not change live buying until Ben approves the results of section 6. Paper first.
- Your crew tag (built today) is the base for this. Correction from the review session: it said early crew spotting "does not work well"; that was about checking buyers at 10 s. Tagging by the launch wallet's funding, which you built, is the right method.
- All times Bangkok. Never shorten addresses.

## 1. Facts this brief rests on (review session, 20 Sep to 3 Oct 2026, every swap and every trader wallet traced)
| Fact | Number |
|---|---|
| Crew A coins (root 0xcb7277454416d90e7fe8799cd300744d5797de03), 281 coins | army wallets made 97.8% of trades and about 100% of buy ETH |
| Crews B (0x5bd9c2026df3d994fe1a0f051ac5992dfee85999) and C (0x64a20b2a553c12960847d49fc381741a981de231), same operator, 336 coins | 97.6% / 97.5% of trades from army wallets |
| Outside buy ETH, whole period | A 13 ETH, B 6 ETH, C 3 ETH (vs wash volume A 43,592 ETH, B 22,767 ETH, C 17,128 ETH) |
| Pull after an OUTSIDE buy of 0.1 ETH or more | within 60 s in 75% (A, n8), 75% (B, n8), 100% (C, n3), usually 5 to 14 s after the buy |
| Pull after an ARMY buy of the same size | within 60 s only 2.7% to 4.9% |
| Outsiders who put in 0.1 ETH or more | 18 of 19 lost everything |
| Outside buys of 0.01 to 0.05 ETH | mostly tolerated (pull within 60 s 7% vs 3%) |
| Paper, ALL 1,402 pond coins, entry first trade + 10 s | 3x reached 58%, 10x reached 20%; half at 3x + half at 10x about +89% per coin before fees |
| Outside the pond (20% swap sample) | crew 0x0424cab581eb1a66da3323d0452cff0872b0fc1f and crew 0xbce31c2e43027d3c926d502ceb9846053fb68ed1: 97% of trades from repeat wallets = WASH. Crew 0x05c8b2d29a7bc2a5e9584f128f2ac3f8a7737bce: repeat wallets 22% of trades, but 87% of its top 60 buyers first funded by 0xa7bf3946d0551af07864b54a0854e85a24aeaaac = UNPROVEN |
- Why the crew does not need to sell before pulling: it owns the pool and the buyer wallets, so army buys only move its own ETH into its own pool; the pull takes it all back. Army wallets still held about half the coins they bought at the pull (A 47%, B 47%, C 56%).
- Crew A cost: about 1.3 ETH gas for 221,241 swaps in 13 days. Its real take is the outsiders' loss (about 7.6 ETH net). This is why it can fake huge volume cheaply.

## 2. Definitions (use exactly these)
- **Army wallet (known):** a wallet whose FIRST incoming ETH came from a known crew distributor (list in section 3), directly or through a send to many contract (0x5b8d85ebabf17cf6b67bfa2fe6795623951cd70e, 0xf341c935599822f1bc1ec29e27606724fbbf6595, or 0xd6d35b56237a05afd4cd7e441226fefc3f78c4e6, a contract both distributors started calling on 5 Oct; likely the same kind, not yet decoded).
- **Repeat wallet (learned, no distributor needed):** a wallet that traded in 20% or more of the same crew's coins in the last 7 days (count a crew's coins by your crew tag).
- **Crew wash score:** over the crew's coins of the last 7 days, the share of swaps (by count) made by repeat wallets or known army wallets. Also store the share of buy ETH.
- **Coin wash score (live):** the same share inside one coin, using all swaps so far. Only trust it after 30 swaps.
- **WASH coin:** crew wash score 80% or more, OR coin wash score 80% or more after 30 swaps. A brand new crew with fewer than 10 coins graded: treat as WASH until proven otherwise.
- **REAL coin:** not WASH, and outside wallets (not army, not repeat) made at least 20% of buy ETH so far.
- **Outside buy ETH:** buy ETH from wallets that are neither army nor repeat wallets of that crew.
- **Trader wallet = transaction "from"** (the signer). The Swap event's sender is the router (99% go through Uniswap UniversalRouter 0x8876789976decbfcbbbe364623c63652db8c0904), so never use it as the trader.

## 3. Known crew distributors (seed list; the repeat wallet rule finds new ones)
| Crew | Distributor | Note |
|---|---|---|
| A | 0xbae538011874261e3fb4ca6aa03cacb2766c4d81 | 1,467 army wallets. Proven: LLAMA creator 0x70905c0d34ff05b5c4f73314dfd16c6aed438eff took the previous pull's ETH and sent 20 ETH to it. Active 5 Oct 20:24. |
| B and C | 0x79411dbcbcf0e5b27cd222f24e082fc0bbf32f16 | 1,365 army wallets, armies never cross with A. B stopped 29 Sep 16:36, C started 29 Sep 21:02. Active 5 Oct 20:15. |
| 0x0424cab581eb1a66da3323d0452cff0872b0fc1f | 0x70b4a4c247737abdab6da8620e50ec1e862a32b1 | sends 0.001 ETH to fresh wallets every few minutes (22,445 txs). Could be a shared faucet; rely on the repeat wallet rule, not this address alone. |
| 0xbce31c2e43027d3c926d502ceb9846053fb68ed1 | 0xad451f226b614456b7372640739e672b409c8723 | funded 100% of its top 60 buyers. |
| 0x05c8b2d29a7bc2a5e9584f128f2ac3f8a7737bce | 0xa7bf3946d0551af07864b54a0854e85a24aeaaac | unproven; funded 87% of top 60 buyers but the crowd is broad. |

## 4. The two lanes (rule for every picker)
- **Lane 1 (WASH coins):**
  - Max 0.0074 ETH (about $20) per coin. Never above 0.02 ETH. Never add.
  - Sell orders attached at buy time: 50% of holdings at 3x (price_scale 200), the rest at 10x (price_scale 900), sell ratio type hold_amount. No stop loss.
  - Only crews with a proven paper edge: A, B/C, plus any crew whose own graded history shows 10x reached in 20%+ of its coins from the first trade + 10 s price. Skip crew 0xae3d565892c79fa8fc9243b380283dd419e28ec9 (10x only 7%).
  - Full test steps: http://159.223.88.168/hx-c9ois/brx/pond-smallbuy-test-plan.md
- **Lane 2 (REAL coins):** normal picker sizing is allowed. If a coin turns WASH after entry (coin wash score crosses 80% after 30 swaps), stop adding and sell down to Lane 1 size.
- **A top pick from any picker that is WASH gets Lane 1 size, no exceptions.**

## 5. What to build (in this order)
1. **Repeat wallet table:** daily (and on every new crew coin), per crew: wallet, coins traded in last 7 days, share of the crew's coins. Mark repeat wallets.
2. **Crew wash table:** per crew, daily: coins in 7 days, crew wash score (count and buy ETH), outside buy ETH total, 3x and 10x rates from first trade + 10 s, pull within 60 s after outside buys of 0.1 ETH or more (count and rate). Publish it as one public file.
3. **Live tag on every launch and every pick:** crew, crew wash score, WASH or REAL, lane, max allowed size. Log it with every paper pick.
4. **Pull watch in the paper picker:** record for every pick the seconds from our (paper) buy to the pull, and flag any pull within 15 s of our buy.

## 6. Re-score the last 14 days (needed before Ben decides)
- For every picker (aipick5 and the others): split all past picks into WASH and REAL with the definitions above (use only data known at the pick time).
- Re-price WASH picks with the Lane 1 rule: fill size 0.0074 ETH; any simulated buy above 0.05 ETH on a WASH coin counts as pulled 10 s later (total loss); exits only by the 3x / 10x sell orders or the pull.
- Report per picker, Lane 1 and Lane 2 separately: number of picks, hit rate 3x and 10x, money result per $20 (Lane 1) and per normal size (Lane 2), and how much of the old paper profit came from WASH coins.
- Expected if the review is right: most of the old big paper wins sit in WASH coins and shrink a lot at Lane 1 size.

## 7. Acceptance checks (if these fail, the build is wrong)
- Crew A (0xcb7277454416d90e7fe8799cd300744d5797de03) crew wash score 95% or more, outside buy ETH about 13 ETH for 20 Sep to 3 Oct.
- Crews B and C: crew wash score 95% or more, sharing almost all repeat wallets.
- Crew 0x0424cab581eb1a66da3323d0452cff0872b0fc1f: WASH. Crew 0x05c8b2d29a7bc2a5e9584f128f2ac3f8a7737bce: should come out below 80% (REAL or borderline); report its outside buy ETH.
- For crew A, pulls within 60 s after outside buys of 0.1 ETH or more: about 75%; after army buys of the same size: about 3%.

## 8. Do not
- Do not use the Swap event sender as the trader.
- Do not treat a high 10x rate as quality before checking wash: the highest 10x rates (80% for 0x0424cab581eb1a66da3323d0452cff0872b0fc1f) were wash crews.
- Do not size any WASH coin above Lane 1, even for a top 1% pick.
- Do not trade live until Ben says go. Report results in Bangkok time, in plain words.

## 9. Send back to Ben
- The public link to the crew wash table, the re-score table from section 6 (one line per picker per lane), and the acceptance check results from section 7.

Version: v1, 5 Oct 2026 22:15 Bangkok. Latest version: http://159.223.88.168/hx-c9ois/brx/pp-wash-brief.md