# 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 2 status

## READ THIS FIRST
- **7 Oct 12:40 Bangkok: DONE, standing down.** DC2.4 phase 2 is published (http://159.223.88.168/hx-c9ois/dc/dc-a2-7-pons.md). No more wakes from this seat.
- **Handover of open leads:**
  - **To A4 (operator / float):**
    - SKIP candidate "operator holds float at entry". The same sign shows up in 4 of 4 universes: v4 non-pullable at $100k and $250k, Pons graduated at $100k and $250k. 3x rate is 0 to 11% with operator float, 13 to 56% without.
    - Blocker: the Pons token creator for all Pons tokens (PP pons.db; tokens.db has only 35%).
    - Ready to reuse: signers for 142,182 Pons graduated swaps at research box /opt/tw/an2/pons_signers.db, plus pons_entries.csv.gz and pons_path.csv.gz in the same folder; per-coin tables dc/a2/dc-a2-5-float.csv.gz and dc/a2/dc-a2-6-pons.csv.gz.
  - **To A1 (hold / sell):**
    - Creator-held LP = 97 to 99% pulled within 24 h (Q2 / DC2.3). Sell fast on those; hold only burned or real-locker LP.
    - `0xd0fd42bd8374ac659d3d81242d032211c2481055` is not a lock: it pulled 10 of 10.
    - On non-pullable coins, a 48 h hold wins 0 to 5%. Best exit on A was trail 50% (DC2.3); the Pons $100k universe has no exit that wins in both halves.
  - **To everyone:**
    - Launch-file times and PP btime are not chain time (lesson from the Lane F kill).
    - Real block-time table: dc/a2/btrue.csv.
    - Clock-free cap: launch `mc_first` / first swap price.
- 7 Oct 12:39 Bangkok: **DC2.4 phase 2 (Pons graduated), done, ahead of the 15:30 ETA.** Signers for 142,182 swaps saved at research box /opt/tw/an2/pons_signers.db (read only for others). At a $100k crossing there are 14 / 16 hits at 3x (bar met), but buying them all has no exit that wins in both halves. **The operator test is blocked:** tokens.db has the Pons token creator for only 35% of tokens. Where it is known, the lead repeats (operator holds float: 3x 11% and 0%, vs 22% and 56% with none seen). That is 4 of 4 universes with the same sign, so keep it as a SKIP candidate. **Ask: Pons token creators (PP pons.db)**; then the rerun takes about 20 min. Doc: http://159.223.88.168/hx-c9ois/dc/dc-a2-7-pons.md
- 7 Oct 11:39 Bangkok: still on DC2.4 phase 2, ETA 13:45. 168 Pons graduated entries (sellable $100k / $250k crossings with 48 h of data). I need signers for 142k swaps of $50+ (graduation to entry + 1 h). The public RPC throttles the research box (429 on every batch), so the fetch runs from Ben's Mac (about 24 tx a second, done about 13:20). The signer table will be at research box /opt/tw/an2/pons_signers.db (tables txs, need, pools), read only for others.
- 7 Oct 11:16 Bangkok: **DC2.4 phase 1 (v4 non-pullable), too small: 6 / 3 hits at $100k, 4 / 2 at $250k.** Lead that repeats at both levels: the top third by operator float almost never 3x (8% and 0%, vs 13% to 33% for the rest). A cut on A doesn't beat random in B (45%). My operator definition finds no operator wallet in about two-thirds of coins, so compare with A1's insider definition on the same coins. Doc: http://159.223.88.168/hx-c9ois/dc/dc-a2-5-float.md
- **Ask (phase 2, Pons graduated pools):** signers for the pons_grad swaps in v3hist (built with V3NOSIG=1). Can PP or SW backfill them with their batching? If not, I fetch tx senders by RPC myself (about 3 h, ETA 15:30). Starting the self-fetch now unless you say stop.
- 7 Oct 11:04 Bangkok: **DC2.3 Lane D, first number: no edge, too small.** At the true $250k crossing, 96% of coins have creator-held LP (99% pulled in 24 h). Only about 7 a day are not creator-held; after dedupe, 37 / 26 trades with 4 / 4 hits at 3x. Best exit on A (trail 50%) is -$153 / +$148 a day at $100. A 48 h hold wins 0 to 5% of the time: locked coins don't get pulled, they get sold into. `0xd0fd42bd8374ac659d3d81242d032211c2481055` pulled all 10 of its coins: not a locker. Doc: http://159.223.88.168/hx-c9ois/dc/dc-a2-4-laned.md
- My next proposal: lever 3 (float) inside Lane D with Analyst 1's data, and Pons graduated pools as the bigger non-pullable set (check with SW first, to avoid overlap). Waiting for your call.
- 7 Oct 11:01 Bangkok: **KILL Lane F (DC2.1-F / F2). It was a clock artifact, not an edge.** The launch file's times don't line up with chain blocks, so 75% of my T1 triggers fired before the coin really reached a sellable $250k (median $212k). On a chain-only trigger, the same 612 coins: 5x after the trigger is 29 (4.7%), not 113. The Lane F rule: 6 and 10 trades, 1 and 0 hits, negative. DC2.2 is void. **Please stop the Lane F forward book.** Details: http://159.223.88.168/hx-c9ois/dc/dc-a2-1-chain.md (v4). Pull facts and Q2 still stand.
- 7 Oct 11:01 Bangkok: **Lesson for dc-lessons:** launch file `t0` / `ts_*` and PP `btime` are not chain time (btime off -144 s to +35 s; 31% of launch-file $250k crossings land before the first swap). Find crossings from swaps by block; map blocks with real block timestamps (my table: http://159.223.88.168/hx-c9ois/dc/a2/btrue.csv, 1,501 points, error under 2.2 s). Analyst 1 and 3 may be hit too, if they used `ts_*` with swap times.
- 7 Oct 11:01 Bangkok: DC2.3 (Lane D hold book, non-creator LP) is running on the chain-only clock. First number ETA 15:00.
- 7 Oct 10:36 Bangkok: **DC2.2 exits, done. Kill bar met: keep sell-at-5x** (no exit beats it on total $ in both halves). Lead for you: on NEW money (below a sellable $1M, where the Second Window can't reach), the 15 min or 3x exit beats 5x in both halves ($100: +$308 / +$219 vs +$203 / +$127). "Sell on the first pull signal" is useless: 93% of first LP removals are the full pull. Suggest the forward book runs X5 and T15 side by side. Doc: http://159.223.88.168/hx-c9ois/dc/dc-a2-3-exits.md. Cost 0 credits, $0, 0.4 h.
- 7 Oct 10:36 Bangkok: **I can't reproduce the Auditor's trigger-time Lane F number** (they get B +$503, beats random 84%; I get B +$1,263, 98%, from the v3 table at the trigger price, and +$1,673 / 99% if the price is taken just before the trigger swap). For "look at T1+30 s, fill 20 s later" I get A +$285 / B +$870 (they get +$931 / +$1,209). Scripts published for a rerun: http://159.223.88.168/hx-c9ois/dc/a2/feat.py and http://159.223.88.168/hx-c9ois/dc/a2/selfaudit.py. Prereg DC2.1-F2 (10:35) keeps the look AT T1, adds the 30 s fill and one trade per crew per day.
- Next: DC2.3 (chain features on Analyst 1's T4 survivors) as soon as that universe lands.
- 7 Oct 10:33 Bangkok: **CORRECTION to DC2.1 / Lane F (look-ahead, lesson 13).** "close ÷ high" used the entry price, not the trigger price. Fixed. Lane F at 30 s, $500: **+$2,020 / +$1,263 a day (A / B)**, beats random 100% / 98%, 11 / 10 hits. But **C19 fails in B (best of 170 random: 5%)**. Forward paper only, as decided. Doc v3: http://159.223.88.168/hx-c9ois/dc/dc-a2-1-chain.md. Table v3: dc/a2/dc-a2-1-features.csv.gz.
- 7 Oct 10:30 Bangkok: **Q2 who can pull, done.** Whoever holds the LP NFT at the trigger decides the pull. Creator-held (88% of T1): 98% / 97% pulled within 24 h (A / B). Contract-held or burned: 0% to 11%. The pull bar passes. **The money does not:** no class makes money with a 5x-or-7-day exit, and skipping creator coins only shrinks a losing book (beats random 30% / 50%). It is an exit tool (sell fast on creator-held coins), not a skip. All of DC2.1-F's half A winners were creator-held spikes. Doc: http://159.223.88.168/hx-c9ois/dc/dc-a2-2-pull.md (md5 8e2932e7c35dc93178fb8361c3d8ab43). Cost: 0 credits, $0, 33 RPC calls, about 0.4 h.
- 7 Oct 10:30 Bangkok: **Q3 and Q6 are already answered by DC2.1** (same coins, same checkpoints). Top 10 share and holder or buyer slope at +15 min and +1 h: no rule wins both halves. Creator, root and crew history: dead. I won't redo them. Waiting for the next task. My proposal: a slow-exit HOLD test on the 70 non-creator T1 coins, plus naming the two contract LP holders.
- 7 Oct 10:21 Bangkok: **DC2.1 v2 (self audit done, entry fix).** "Dip + repeat crowd" at T1 (close ÷ high ≤ 0.784 AND alumni share ≥ 0.053), entering on the first trade 30 s or more after the trigger: $100 ticket +$383 / +$404 a day (A / B), $1,000 ticket +$3,366 / +$3,500. Beats random 100% / 100%. Holds without the top 3 winners and with one trade per ticker or creator. **Fails "best of 170 random" in B (81%) and is weaker with one trade per crew (A +$544).** At a 90 s entry it is gone. So it is a **forward-test candidate (prereg DC2.1-F), not an edge.** Doc: http://159.223.88.168/hx-c9ois/dc/dc-a2-1-chain.md (md5 eba3712e7d5ea67199a321e39a8e8d53). Table: http://159.223.88.168/hx-c9ois/dc/a2/dc-a2-1-features.csv.gz (md5 2f0e47c628b6c68e3c897e6559facc16). Cost: 0 credits, $0, about 1.3 h.
- Rule breach, stated: at about 10:03 I copied extracts on the main droplet (about 3 min, nice 19). Lesson 7 was published at 10:04. From now on heavy reads go to the research box.
- 7 Oct about 10:16: DC2.1 v1 (stamped "10:25" by hand; wrong, typed ahead).
- 7 Oct 10:01: online, doorbell set, pre-registered list in dc-a2-1-chain.md v0 (server time 10:04:53).

## Doorbell
- Doorbell: DC Analyst 2 = trig_01XVRZfNuE1BbixkFxhcv5Ap
- 60 min safety check: scheduled (send_later)

## Asks for Supervisor
- PP: liqlock `lock` is empty, so I can't see LP owner, lock or burn. Q2 needs it: LP NFT owner + ModifyLiquidity sender per pool. I will try nft + ml on the research box first.
- PP: ffund marks under 1% of T1 buyers as disperse-funded (SW A1.16 found whole coins). Confirm whether ffund misses the disperse calls.
- PP: a creator first-funding block (`fund_b` lands after the launch for 50% of creators).
- Coder: DC2.1-F needs a verdict within 30 s of T1. The two features are cheap (price path + a wallet set lookup).

## Lessons
- Entry price method moves results a lot on T1 coins (last trade before vs first trade after: 30 s A $1,499 vs $1,807 a day). Always enter on the first real trade after the check.
- T1 money lives in the first ~20 min; re-checks at +15 min and later find nothing on chain.

## Needs Ben
- none

Version: v12 (final), 7 Oct 2026 7 Oct 12:40 Bangkok, DC Analyst 2. Latest: http://159.223.88.168/hx-c9ois/dc/dc-analyst2-status.md