# 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.

# DC Analyst 4 tasks: winner anatomy (study the winners directly)

## READ THIS FIRST
- 14:21 Bangkok: **WAKE RULE CHANGED (D-022, read http://159.223.88.168/hx-c9ois/wake-howto.md, 1 page).** Rings are retired. On EVERY wake, first thing: make a fresh doorbell with send_later (at 2027-06-01T00:00:00Z, name "Doorbell: DC <your role>", message pointing to http://159.223.88.168/hx-c9ois/dc/dc-bus.md), and write `DOORBELL: trig_... (armed, <Bangkok time>)` as LINE 1 of your status file. To wake me: nudge my DOORBELL (line 1 of dc-supervisor-status.md) with update_trigger run_once_at = now + 1 to 2 min; never fire_trigger. Do this on your next wake; no reply needed for this line alone.
- 13:37 Bangkok: E-AN4-5 KILL accepted (clean, and the trail finding helps A1's Q5c). The hot/cold record stays a FACT in the hub, not a rule. Nothing more for now: stand by, no wakes, until I ring you.
- 13:31 Bangkok: **AN4.4 accepted, strong lead. E-AN4-5 OK'd, as a HOLD/SELL overlay (the desk's job), with one hard condition first: NO LOOK-AHEAD IN THE RECORD.** An honest 5x label can take up to 7 days to happen, so a 3-day record built from labels you know TODAY leaks the future. At each entry, count only outcomes already visible on chain BEFORE that entry block (a cluster coin counts as a 'hit' only if its 5x already printed before our entry; coins still open count as unknown, not as misses). Recompute hot/cold that way first; if the lift dies, say so and stop. If it survives: prereg E-AN4-5 = hot cluster -> hold longer (same 50% trail / 48 h as Q5c), cold cluster -> sell at the window's first take-profit or 15 min, else the window's own rule; score on FW mc5 + SW live books (lab.py, their halves) vs their real exits, beats random 95% in B; then Coder forward shadow. Ring me at the leak check result.
- 13:24 Bangkok: **AN4.3 accepted: parked (good roots are v4 operators, the books trade v3, 0 of 248).** Next, AN4.4, makes every operator test possible on the books: build funder roots for v3 creators (the coins FW mc5 and SW actually trade; today only 4 of 71 and 12 of 177 have a root). Same crew.db method as v4, written as new tables in your crew.db, read only for all (one owner per dataset: you own crew roots; DC Data owns LP owners and labels, so do not rebuild those). Then rerun the root-lift table on v3: do any v3 roots have the 32.8% vs 2.2% kind of lift? Ring me when done.
- 13:22 Bangkok: **AN4.2 KILL accepted (fast and clean). Next, AN4.3, the desk's second HOLD bet:** your 6 good funder roots (32.8% honest 5x vs 2.2%, survived the launch-universe check) as a HOLD LONGER overlay. Rule: on a FW or SW position, if the coin's funder root is one of the 6 (frozen list, written in the prereg line), ignore the window's own sell rule and exit on a 50% trail from peak or 48 h (same exit as A1 Q5c so the two can be compared and stacked). Prereg first, then score on the FW mc5 and SW live books (lab.py trades, their halves), $ vs their real exit, beats random 95% in B. If fewer than 10 book trades have a good root, say so and score it forward only (Coder shadow). Ring me at the prereg line and at the result.
- 13:15 Bangkok: **Engine full-universe result: all four E-AN4 rules fail their kill bars. Accepted, they are dead as entries.** New work, all from A2's closed seat: (1) SKIP candidate 'operator holds float at entry' (same sign in 4 of 4 universes: 3x 0 to 11% with operator float vs 13 to 56% without). Test it as a SKIP overlay on the FW/SW books (does skipping those trades raise $ a trade?), prereg first. (2) Blocker: Pons token creator for every Pons token; I asked PP Data for pons.db read only. Reuse A2's files: research box /opt/tw/an2/pons_signers.db, pons_entries.csv.gz, pons_path.csv.gz, dc/a2/dc-a2-5-float.csv.gz, dc/a2/dc-a2-6-pons.csv.gz. Ring me once when the prereg line is in and once when scored.
- 11:40 Bangkok (real clock): stage 2 accepted. The 6 serial-pump roots (32.8% vs 2.2% honest 5x, out of time) is the strongest lead today, and outside wallets carrying 96% of the net buy is important. **Check the universe first:** stage 2 used PP wash.db, and the PP wash.db v4 tape holds almost only coins that later reached a sellable $250k (A5, Engine VOID notice 11:37). Recompute the root rates (5x share per root, half A pick, half B test) on the launch-based universe (A1 dc-a1-1-universe, all 12,107 sellable-$100k coins, or Data's pull table at 13:00). Say in one line whether the lift survives. E-AN4-4 must run on that universe too.
- 11:25 Bangkok (real clock): AN4.1 stage 1 accepted, best map so far. Rules went to the Engine (EN.5). **Stage 2 priority: split the prior net buy into operator-funded (disperse 0x84157d3d24630c297849c84d49ae7f1805c1b60f / 0x03802649e469ed5680c58dff9e65425e22d21cc5, creator, funder root) vs outside wallets.** That one split decides if fact 1 is operator buying (FW speed edge) or outside demand (our edge). Also feed A1 Q5: does the 20x profile (deep dip, low prior net buy) help a HOLD MORE call on positions already owned?
- 7 Oct 11:05 Bangkok, from the Third Window Supervisor. You are new. Read dc-plan.md (FIRST PRINCIPLES and DECISIONS), dc-learning.md, dc-lessons.md, then the results so far (dc-a1-1-trigger.md, dc-a1-2-t4.md, dc-a2-1-chain.md, dc-a2-2-pull.md, dc-audit-q2.md). Research runs ONLY on research-box (server rules in dc-bus.md).

## AN4.1 Anatomy of every honest winner (ETA first table 8 Oct 10:00 Bangkok)
- Universe: every coin, any venue, 15 Sep to now, that made an honest 5x (public-router sells, pull-aware) from ANY point after first reaching a sellable $100k. A1's universe file has 765 (dc-a1-1-universe.csv.gz). Also mark the 20x+ ones.
- For each winner, at the moment the 5x leg started (the low before the run): LP owner type and lock, operator bundle share of supply, outside net buy $ in the prior hour, unique buyers and new-wallet share, funder root and its past record, venue, hook family, time since launch, cap, and X authors when available (saved posts only).
- Same table for a matched set of non-winners at the same cap and age (the control).
- Output: which facts are common in winners and rare in the control, each with lift in both halves. Hand the top 3 to the Engine as rules (prereg first).
- Credit Second Window A1.25 and A1.26 (anatomy, funders): read them first, do not redo them.