# 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.

DOORBELL: trig_01N8WcUHKwKNUHnBGECXFyuu (armed, 07 Oct 16:32 Bangkok)
# Analyst 3 status (auditor and pull hunter)

## READ THIS FIRST
- **07 Oct 16:37 Bangkok: v1.15 A3.15 DONE. Verdict: KILL (clean, pre-registered rule).** Age-scaled trail (50%, then wider once the coin is older) loses in half A in all 4 cells (-$411 to -$2,555 a day); chosen cell (60% after 6 h) half B -$2,896 a day; beats random 64% / 3%. After a kit runner peaks the drop keeps going, so a wider trail sells lower: 0x8e41fd25e85e624560fa50ff9f117a02cd30b8b3 29.15x to 23.49x. Lead for my queue: runners sell at about 0.4 of their peak (29x of a 74.5x peak), so the next test is a TIGHTER trail far above 5x. Fills priced 1 s after the trigger (Exec twin rule). Ledger row SW-A3.15 added. http://159.223.88.168/hx-c9ois/sw/sw-a3-15-runner-exits.md cost: Claude credits about 0.5 h of one session, API $0, 0.1 hours.
- **07 Oct 16:33 Bangkok: v1.15 A3.15 RUNNER EXITS STARTED. Rule PRE-REGISTERED (before any result).** First priced answer ETA 17:40 Bangkok (why: engine reuses A3.14, about 20 min to write, about 5 min to run, then the doc).
  - Trades: every live book + kit300k trade that touched 5x under the current rules. The change acts only on KIT HOLD trades (non-kit 5x trades still sell at 5x; holding non-kit was killed in V13).
  - Current exit (the bar): after 5x, 50% trail from the peak cap, or in-range half-pool pull, or 48 h from entry.
  - ONE exit family: AGE-SCALED TRAIL. Trail width = 50% while the coin is younger than A0 (age from its pool's first swap), then W2 once it is older. Grid on half A only, both books together, best half A $ a day: A0 in 2 h / 6 h, W2 in 60% / 70% (4 cells). Pull and 48 h unchanged.
  - Mechanism (who pays us): in the first hours a 50% drop is the launch crowd dumping; a coin that survived past that has a wider holder base and swings 60 to 70% in normal pullbacks before a second leg. A fixed 50% trail sells those pullbacks at the bottom.
  - What is new vs killed tests: A3.13 keyed on the operator's wallets (they sell before our 5x); V14 net-buy and FW A1.28 drip-seller read flow. This one only changes the trail width by age, no flow read.
  - Fills: every exit (current and new) at the block 10 blocks after the trigger (about 1 s, the Exec twin rule "crossing block + 1 s", http://159.223.88.168/hx-c9ois/sw/sw-exec-exitgap.md), full tick walk, $1k. Also shown at the trigger block.
  - Random: for the trades where the chosen cell changes the exit, 200 draws of a uniform random exit time between the 5x and the current 48 h end (or data end), same fill rule.
  - PASS = chosen cell above the current exit in $ a day in BOTH halves (both books together, live book not below) AND beats 95% of random. Else KILL.
- **07 Oct 16:31 Bangkok: A3.14 time stamp fix.** The A3.14 DONE line says 16:35; it was published at 16:31 (my clock ran ahead).
- **07 Oct 16:35 Bangkok: v1.14 A3.14 DONE. Verdict: KILL (clean, pre-registered rule).** No signal beats the current exit in half A, so nothing is frozen. S1 (one in-range Burn of 20%+) fired 0 times in 283 trades: before the stop there are only 11 in-range Burns, all under 0.3%; LP leaves in one go and the live pull exit already sells that block. S3 (3-block net sell burst) loses in half A at every X: 5% -$4,977 a day, 10% -$4,344, 20% -$199, 30% -$658 (both books). Sell bursts hit future 5x coins too: S3 5% saves $67,180 on losers but cuts 40 winners for -$158,331. Ledger row SW-A3.14 added to http://159.223.88.168/hx-c9ois/fw/lab/ledger.csv . http://159.223.88.168/hx-c9ois/sw/sw-a3-14-early-exit.md cost: Claude credits about 2 h of one session, API $0, 1.8 hours (promised 16:10, late 25 min: the Mac link dropped).
- **07 Oct 14:48 Bangkok: v1.14 A3.14 STARTED. Rule PRE-REGISTERED (before any result).** Priced result delivered 16:35 Bangkok (promised 16:10; the Mac link dropped).
  - Trades: every trade of (a) my live wide-gate book (71 trades) and (b) the A3.12 kit300k book (Kit A/B/3,587 coins at fair $300k). Same engine as A3.13: entry block as before, $1k, every fill by a full tick walk of in-range positions.
  - Current exit (the bar): 0.3x stop, 5x target then KIT HOLD (50% trail, pull, 48 h), in-range half-pool pull, 12 h time stop. Base fills at the end of the trigger block.
  - Signal window: from the block after entry until the current exit fires or the 5x is hit. After the 5x nothing changes.
  - Candidates (pick ONE on half A only, both books together, best half A $ a day vs the current exit):
    - S1 in-range LP removal: one Burn removes 20%+ of the in-range liquidity (in-range reserve after / before the Burn under 0.8).
    - S3 net sell pressure: in the last 3 blocks (this block and the 2 before), quote paid out by sells minus quote paid in by buys is X% or more of the in-range quote reserve at the start of those 3 blocks. X in 5 / 10 / 20 / 30%.
    - S2 (bundle wallet sends to pool or router) is not run: my bundle tags exist only for the 41 coins that hit 5x (A3.13); the losers have none, and tracing ~280 coins on the paid archive does not fit the 2 h answer. Also A3.13 showed bundles mostly sell before our 5x, so it would fire early on winners too.
  - Fill when the signal fires in block b: sell everything at the end of block b+1 (one block late, we see it after the block). If the current exit already fires by block b+1, nothing changes.
  - If the best half A cell is not above the current exit in half A, the verdict is KILL and half B is priced only for the record.
  - Half B: the frozen cell, $ a day vs the current exit, per book and both together. Winners lost = trades that would hit 5x but the signal sold first, each priced.
  - Random: 200 draws, same number of early exits as the signal has in that half, on trades drawn at random from that half (both books), each sold at a random block between entry and its current exit (or its 5x), same one-block fill.
  - PASS = half B $ a day above the current exit (both books together, and the live book not below it) AND beats 95% of the 200 random draws. Else KILL.
  - Different from the killed A3.3 early exit (time and price only): this keys on LP leaving or a 3-block sell burst, and only before the 5x. Killed flow exits V14 (net-buy) and FW A1.28 (drip-seller) get cited in the doc.

## Doorbell
- trig_01N8WcUHKwKNUHnBGECXFyuu (Doorbell: Analyst 3), armed 07 Oct 16:32 Bangkok. Rings retired (D-022): I nudge the Supervisor's DOORBELL with update_trigger.

## Tasks
| Task | Status | Doc |
|---|---|---|
| A3.15 runner exits, age-scaled trail (v1.15) | DONE: KILL (wider trail sells runners lower); next lead: tighter trail far above 5x | http://159.223.88.168/hx-c9ois/sw/sw-a3-15-runner-exits.md |
| A3.14 early exit for SW losers (v1.14) | DONE: KILL (S1 never fires, S3 loses half A) | http://159.223.88.168/hx-c9ois/sw/sw-a3-14-early-exit.md |
| A3.11 smart wallet cohort, early entry (v1.11) | DONE: KILL | http://159.223.88.168/hx-c9ois/sw/sw-a3-11-smart-wallets.md |
| A3.13 operator exit (v1.13) | DONE: KILL | http://159.223.88.168/hx-c9ois/sw/sw-a3-13-operator-exit.md |
| A3.12 kit at fair $300k audit (v1.12) | DONE: SMALLER, paper slot yes | http://159.223.88.168/hx-c9ois/sw/sw-a3-12-kit300k.md |
| A2.31 moonbag audit (v1.11) | DONE: KILL confirmed | http://159.223.88.168/hx-c9ois/sw/sw-a3-11b-a231-audit.md |
| A3.1 rebuild the v3 rule from scratch | DONE: 58 trades 1.915x, matches A2 | http://159.223.88.168/hx-c9ois/sw/sw-a3-1-rebuild.md (trades sw-a3-1-trades.csv) |
| A3.10 KIT HOLD under 3-open cap (v1.10) | DONE: lead holds both halves | http://159.223.88.168/hx-c9ois/sw/sw-a3-10-slot-cap.md |
| A3.9 KIT HOLD audit (v1.9) | DONE: SHIP LIVE by the rule; thin Kit B sample | http://159.223.88.168/hx-c9ois/sw/sw-a3-9-kit-hold.md |
| A3.8 ladder (b) for live (v1.8) | DONE: KEEP (a) | http://159.223.88.168/hx-c9ois/sw/sw-a3-8-ladder-b.md |
| A3.7 V7 hold-or-sell audit (v1.7) | DONE: paper only; big second-half gain, one switch only | http://159.223.88.168/hx-c9ois/sw/sw-a3-7-v7-audit.md |
| A3.6 A2.16 gate audit (v1.6) | DONE: CONFIRMED, +$1,377 / +$1,898 a day | http://159.223.88.168/hx-c9ois/sw/sw-a3-6-a216-audit.md |
| A3.5 V5 stop audit (v1.5) | DONE: V5 CONFIRMED, smaller (2.11x no stop vs 2.00x) | http://159.223.88.168/hx-c9ois/sw/sw-a3-5-v5-audit.md |
| A3.2 sell block census | DONE: no stuck trades; 1.7% of coins have sell-limiting code; real trap is instant pulls | http://159.223.88.168/hx-c9ois/sw/sw-a3-2-sell-census.md |
| A3.3 early fail exit + pull warning | DONE: keep the plain rule; best early exit +0.03x (noise); pull tell real but costs money | http://159.223.88.168/hx-c9ois/sw/sw-a3-3-early-exit.md |
| A3.4 bad day risk | DONE: worst day -$937, no losing 3-day window, max 2 open | http://159.223.88.168/hx-c9ois/sw/sw-a3-4-bad-day.md |

## Needs Ben
- None. (dRPC was out of balance at 13:18 Bangkok; Supervisor v1.2 says it was topped up around 13:30.)

## Asks for Supervisor
- Tell the Coder (C1): exclude the coin itself from its own army memory (A3.1), and the pull guard must be range-aware (A3.1 v1.3).
- Tell Analyst 1: check the A1.11 wallet list with eth_getCode; 0xabe99cf268cdc8bd70a71a5283fdacbcb36714fd is a token contract.

Version: v21, 07 Oct 16:32 Bangkok, Analyst 3. Older entries: http://159.223.88.168/hx-c9ois/sw/sw-analyst3-status-archive.md . Latest: http://159.223.88.168/hx-c9ois/sw/sw-analyst3-status.md