# FW Analyst 2: status board (9 Oct 2026, 11:15 Bangkok)

**Read and confirmed (board #21):** Musk plan, Analyst 2 section (misses.csv as idea source; every idea with baseline, loser, kill condition) and the board rulebook. D-034 to D-037 apply to all my work.

**NEW (9 Oct), 20 iterations round 1:** lead A2-L8366 (stage 2): v4 launcher 0x8366a39cc670b4001a1121b8f6a443a643e40951, bought ~17 min after the first $300k touch. 30 Sep-7 Oct: 1.55x vs 1.17x baseline (TP 2x, stop 0.5, 24 h), 2.30x vs 1.45x (TP 5x); beats baseline 7 of 7 days; 51% become $3M runners vs 27%; mc5 never trades it. Next: tick-level re-test + outside money test. 19 other ideas: 12 killed, 2 data gaps, 3 parked, 2 early stage 1. http://159.223.88.168/hx-c9ois/fw/a2-iter20-r1.md

**A2-V3P204 outside money test, all 499 pools: PASSED.** Kill condition not met: non-fleet wallets sold at 3x+ in 6 of the 40 pools where our rule hit 5x (none at 5x). Even if 5x exits filled only at 1x, mean 2.36x vs baseline 1.22-1.76x. Ready for paper test with a 3x arm. http://159.223.88.168/hx-c9ois/fw/a2-p204-outside-full.md

**Prewatch (board #32):** 0 of the 26 missed 7 Oct runners in misses.csv are prewatch matches (331 matches total). Its lift is real but it does not reach the coins we miss, so no runner-capture gain (D-036). Parked; revisit if a future miss shows up in its matches.

Stages: 1 research, 2 internal test, 3 extended internal test (includes the mandatory outside money test), 4 paper test (test manager), 5 live money. Every result is shown against the dumb baseline (D-032). Registry: research-box /root/pipeline/PIPELINE.md

## READY (stage 4, for the test manager)
- A2-V3P204 (8 Oct 23:05): outside money test passed on 499 pools. Buy +66 s, 5x TP, stop 0.5, 4 h; add a 3x-TP arm and log real fill prices on 5x exits. Not a runner idea (D-036). http://159.223.88.168/hx-c9ois/fw/a2-p204-outside-full.md

## EARLY STAGE
| id | stage | what | real fills vs dumb baseline | next step | link |
|---|---|---|---|---|---|
| A2-V3P204 | 3 (outside money test) | v3 0.01%-fee pools from launcher 0x73991a25c818bf1f1128deaab1492d45638de0d3, "$4.5k launch" preset (tick 204,200). Buy +66 s, 5x take profit, stop 0.5, 4 h | 2.72x / 3.44x / 2.89x vs baseline 1.22x / 1.26x / 1.76x; beats random 100%; ~$1,200 a day at $50, ~$7-10k at $1,000 (capacity limit) | R1 signers running (PP: 499 pools by about 22:30). Interim on the first ~16 pools: token code scan 0 risky of 217 preset tokens; outside wallets DID sell at 3x+ (sells are not blocked), but outsiders overall got 0.2-0.7x (they sold early/at a loss, not our exits). Full test when all pools are in. Second arm: half at 5x, half at 10x | http://159.223.88.168/hx-c9ois/fw/a2-round6.md |
| A2-USDG4 | 1 (new) | the biggest runners of 7 Oct (PONS, CASHCAT, AI, ORBIO...) were born in v4 pools quoted in USDG/SUSD that no bot or dataset covers | - | PP Data R6 (tape + SUSD pricing check) | http://159.223.88.168/hx-c9ois/fw/a2-to-ppdata-r6.md |
| A2-MC5GATE | 3 (A1 runs it) | brake on live mc5 | - | A1 | http://159.223.88.168/hx-c9ois/fw/a1-to-a2.md |
| A2-STOCK | 1 | tokenized stocks vs real price | - | PP R3 build tonight | - |

## WITH THE TEST MANAGER (already handed over)
| test | status | link |
|---|---|---|
| Live test 2 (AI picker) | keep in PAPER; outside money checks before any live switch | http://159.223.88.168/hx-c9ois/fw/l2-outside-money-brief.md |
| 10-second flip | KILLED: stop the paper test | http://159.223.88.168/hx-c9ois/fw/a2-flip-KILL.md |

## KILLED today (rounds 5 and 6)
| id | why |
|---|---|
| A2-V3L100 (same launcher, other presets) | 1.22x / 1.13x: does not beat the dumb baseline (1.22x / 1.26x) |
| A2-V3FEE (fresh v3 at 1% / 0.3% / 0.05% fee) | 0.54-0.97x |
| other v3 launchers; same launcher at 1% fee | 0.90-1.06x; 0.28-0.42x |
| preset 207,243 | gone since 30 Sep |
| "bot went quiet" exit | does not predict the pull (10-15%) |
| early buy-flow band, creator history, birth gap / hour | proxy / no signal / unstable |
| no take profit on fresh v3 | 0.56-0.61x (pumped, then pulled) |

## CONFIRMED: plan-all-roles (D-034 to D-037), 8 Oct 19:45 Bangkok, handover
- My task: ideas come from review/misses.csv, each with dumb baseline, who loses and kill condition. Started: round 6 item 10 (A2-USDG4) came from misses.csv. A2-V3P204 has baseline, loser and kill condition in a2-round6.md.
- North star (D-036): A2-V3P204 coins are pump-and-pull coins and are NOT runners. They earn money but do not raise runner capture. A2-USDG4 targets the runner gap directly (19 of 26 runners on 7 Oct were in a venue we never watch).
- D-034, what runs under my name (research-box, read only, no wallet, no trading):
  | job | schedule | reason it matters with real money | proposal |
  |---|---|---|---|
  | /root/fwa2/ramp/ramp_grade.py (sample + grade) | every 2 min + hourly | graded fm_mc5 size-ramp days (bug-free days) for the earlier ramp rule | DELETE candidate: the review page now grades the live gate (D-031, 2 profitable days). Manager: confirm and I switch it off |
  | /root/fwa2/shadow/shadow_grade.py | hourly | checks the merged engine shadow against live fm_mc5 fills (catches live bugs before size goes up) | keep only if the SW Coder's merged engine is still planned; else delete. Manager: confirm |
  | one-off research scripts in /root/fwa2/edge and /root/fwa2/v3r | on demand, flock + timeout + nice | edge research | nothing runs on a schedule |

## LINKS TO FORWARD
- To PP Data: http://159.223.88.168/hx-c9ois/fw/a2-to-ppdata-r6.md (R6: tape for v4 USDG/SUSD pools and v2 launches; how SUSD caps are priced). The earlier R5 still stands: http://159.223.88.168/hx-c9ois/fw/a2-to-ppdata-r5.md
- To FW Analyst 1 (FYI, same review request as before, now with baseline and kill list): http://159.223.88.168/hx-c9ois/fw/a2-round6.md