# 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_01PUBUd2pZhS2d9ECGr2cG1z (armed, 17:27 Bangkok)
# DC Analyst 1 status

## READ THIS FIRST
- 7 Oct 17:33 Bangkok: **Correction to Q5b (12:45):** /opt/sw/v3hist.db pools.creator is empty for all 7,584 pools, so Q5b ran A3's ruler without the creator and creator-funder exclusion. The sign flip is large (fired coins 4.49x vs 2.34x), so the verdict stands, but the counts there are "outside plus creator-funded" wallets. Q5c used Engine's creator column, so it is not affected. Edge Hunter: cards D-A1-1 and D-A1-2 in cons/ideas.csv; night script research-box /root/dc-hunt/q5c_final.sh (final Q5c score, result in /root/dc-hunt/q5c_final.done).
- 7 Oct 17:27 Bangkok: **Q5c INTERIM: V1 fails the random bar.** $ added A +$13,559 / B +$12,686 (n 150 fires), but the 50% trail on random open positions does as well: beats random 86% (A), 60% (B), bar 98%. Without the 323 partial paths: B +$3,448, 39%. Money comes from the trail, not the holder signal. V2 / V3 still unscored (LP class for v3 / Pons, Engine V3 columns). Doc: http://159.223.88.168/hx-c9ois/dc/dc-a1-7-q5c.md
- 7 Oct 17:27 Bangkok: role update taken: DC EDGE HUNTER (Consigliere 17:23). After Q5c: idea cards in cons/ideas.csv, long test as a script on research-box, report to cons/inbox.md.
- 7 Oct 16:07 Bangkok: 16:03 add taken: the 17:30 score will be shown with and without the 323 time-dropped coins. Those rows have no path in Engine q5c_paths.csv (status uncovered, all price columns empty), so the Engine must add partial paths cut at the tape end (ask 0b). Stamps now come from `TZ=Asia/Bangkok date` on the droplet; the "16:02" line above was written at 15:59.
- 7 Oct 16:02 Bangkok: **Q5c reconcile done; interim score at 17:30.**
  - Engine q5c_paths.csv: 2,608 rows = Data's 2,593 coins + 15 coins with a second pool. 2,254 rows scored ok = 2,240 coins.
  - The 353 dropped coins (2,593 - 2,240): 323 uncovered (entry + 48 h is past the tape end, a time rule), 22 no fill, 8 no reserve. All are data rules fixed before outcomes; none use the result. 5x share in the drop: 60 of 353 (17%) vs 527 of 2,240 (24%) kept.
  - Prereg "RULER NOTE" line at 15:57: one uniform window for every coin (balances from first sellable $300k - 1 h), because the logs start at different points for winners and losers and would fake a signal otherwise.
  - Growth reads done for 789 coins so far (533 5x, 256 non-5x). Scorer is ready; it waits for more non-5x logs.
  - V3 needs a price path from the Engine (70% trail armed at 2x, window stop kept): asked below. V2 needs an LP class for v3 / Pons pools.
- 7 Oct 13:16 Bangkok: **Q5c prereg line is in** (dc-prereg.md "7 Oct 13:16 | A1 | Q5c"). V1 plain trail-hold and V2 with A2's LP gate (burned or real locker) in one line. Unseen data only: DA.4 coins minus my 203, plus FW / SW positions opened after 13:15 (Coder shadow). Next: build the scorer now, run it when DA.4 lands (16:00). ETA first score 17:00 Bangkok.
- 7 Oct 12:45 Bangkok: **Q5b done. KILLED, and the sign flips.** Selling SW / FW coins on A3's holder-growth death sign (+528 in 50 min) loses -$12,065 (A) and -$42,268 (B), beats random in B 1%.
  - Coins that fired went on to a mean 4.49x exit vs 2.34x for the rest. Early growth after our entry is the run itself. A3's growth at cross + 1 h is the late exit wave. The coin's stage decides the sign.
  - Lead (not pre-registered): growth >= +528 after entry as a HOLD LONGER / SIZE UP overlay. Proposed as Q5c.
  - xfer.db now covers 203 coins (SW 70, FW mc5 176 rows, FW live 4): 2,697,380 rows, 0 gaps.
  - Doc: http://159.223.88.168/hx-c9ois/dc/dc-a1-6-q5b.md
- 7 Oct 12:10 Bangkok: **Q5 v1 done. Both overlays KILLED by the prereg bars.**
  - (a) SELL NOW on bundle sells: A -$2,121 (-$152 a day), B +$696, beats random in B 79%.
  - (b) HOLD past 5x on non-kit coins: A -$4,335, B +$519. Selling at 5x stays right.
  - Why: at SW entry the launch bundle holds a median 0% of the float (it held 70% at cross + 1 h). The insider lever is gone by gate time.
  - **xfer.db ready** for Data and A2: /opt/tw/an1/xfer.db on research-box. Transfer logs for 74 coins (70 SW + 4 FW live), mint to gate + 50 h, 1,688,214 rows, 0 gaps.
  - v1 fix stated: the mint was being read as the first market transfer, so the bundle window was empty. Fixed.
  - Doc: http://159.223.88.168/hx-c9ois/dc/dc-a1-5-q5.md
- 7 Oct 11:35 Bangkok: **Q5 v0: overlays blocked by a data gap.** 67 of 70 SW wide-trade tokens have no Transfer logs in pp3_raw, so the bundle features are empty and nothing fired.
  - Ceilings (perfect foresight, $500): sell-now about $1,300 a day (A) and $2,200 a day (B). Hold-past on non-kit 5x exits is large, but blind holding loses -$6,964 in A.
  - Asks 1 to 3 in the doc: one-off transfer fetch for the 70 tokens, DA.1 v3 LP class, coordination with SW A3.13.
  - Doc: http://159.223.88.168/hx-c9ois/dc/dc-a1-5-q5.md
- 7 Oct 11:26 Bangkok: **Q5 started.** Prereg line "Q5 overlays" written, with a time-fix line because it says 11:30. Fix for the repeated clock slips: every A1 stamp now comes from the server clock (date), never typed.
  - **Blocked:** research-box is out of memory (load 53, 130 MB free). 20 workers of PP's E1 job e1exact.py use about 11 GB, and liqbuild2.py uses 4 GB. My ceiling run was killed by it. I will not add load. Next try at 11:26 + 30 min with a small indexed extract (no pickles).
  - **Overlap note:** SW A3.13 (operator exit when bundle wallets sell) is OPEN and close to my overlay (a). SW V14, A2.29 and FW A1.28 already killed nearby exits. So my overlay (a) is scored on the same 70 trades for a direct comparison, and lever 1 (v3 LP holder) waits for Data DA.1.
  - ETA first cut: 13:30 Bangkok if the box frees up.
- 7 Oct 11:20 Bangkok: **Q-FLOAT and T4-ALIVE done. Both KILLED by their pre-registered bars.**
  - Q-FLOAT: buy all -$506 / -$158 a day. The A-chosen cut (insider share in the bottom quarter) gives -$69 / -$113 and beats random in B only 11%. Insider share AUC flips: 0.62 / 0.35.
  - T4-ALIVE: -$253 / -$277 a day, beats random in B 54%.
  - Fact: at cross + 1 h, the launch bundle still holds a median 70% of wallet-held supply, and 55% of the pools still standing get pulled within 48 h.
  - **Push back:** Lane D should switch from new entries to Q5 HOLD MORE / SELL for coins the First and Second Window already own.
  - Clock check: all my times are chain time, no launch-file times. The Lane F bug does not apply to DC1.1, DC1.2, Q-FLOAT or T4-ALIVE.
  - Doc: http://159.223.88.168/hx-c9ois/dc/dc-a1-3-qfloat.md
- 7 Oct 10:58 Bangkok: DC1.2 done, every T4 variant killed. Doc: dc-a1-2-t4.md
- 7 Oct 10:50 Bangkok: DC1.1 done. Doc: dc-a1-1-trigger.md. The v3 lead is pre-registered as DC1.1-V3F (forward).
- Next: lever 1 (v3 LP class) when Data DA.1 lands. Idle until a new task.

## Doorbell
- `Doorbell: DC Analyst 1` = trig_01J9g2776Jah7qHpuJgdeB9f (no self wakeups)

## Asks for Supervisor
0b. Engine: fill BASE and V1 columns for the 323 uncovered rows, cut at the tape end (last sellable print), with a flag column, before 17:30 if possible.
0. Engine: add V3 columns to q5c_paths.csv (after the read: keep the 0.3x stop, drop the 5x take-profit, keep the 48 h cap, and once the peak reaches 2x exit on the first print at or below 0.3 x peak). Data: LP class (burned / real locker / creator) for v3 and Pons pools at entry, for V2.
1. Q5 and Q5b are closed. Approve Q5c (holder growth after entry as HOLD LONGER / SIZE UP on FW mc5 and SW, own prereg line)?
2. A2: compare your funder-root float ruler with my first-30-block bundle (dc-a1-3-qfloat.csv.gz has F1 per coin).
3. Engine v1.2: please join dc-a1-3-qfloat.csv.gz (token, F1, F2) for an official rerun.
4. PP (open from earlier): unique buyers over time on all venues; v4 tape from 6 Oct; a liqlock refresh on research-box (it ends 5 Oct 17:39); Pons graduation LP events; v2 pools.

## Needs Ben
- none

## Cost
- X 0, API $0, research-box CPU only.

Latest: http://159.223.88.168/hx-c9ois/dc/dc-analyst1-status.md