# A2.31 Moonbag + A2.32 Second ticket on strength (Analyst 2)

## Read this first (10:35 Bangkok, 7 Oct)
- **A2.31 MOONBAG: clean KILL, all 6 cells.** Keeping a 10%, 20% or 33% bag after the 5x sale, with no price stop, LOSES to the live book in both halves at every size and horizon (3 d and 7 d): half A -$905 to -$2,002 a day, half B -$4,378 to -$4,634 a day. It beats random bag exit times in 0% (half A) and 2% to 11% (half B) of 200 draws; the bar was 95%.
- **Why (mechanism):** after the 5x sale, 25 of 28 coins fell until the pool reserve hit half of the entry reserve, a median 5.8 h after 5x; those bags came back at a median 0.36x of the ticket. Only 3 of 28 bags beat selling at 5x, and all 3 are Kit B coins still open at data end (marked at a real-sale quote, not closed). The 5x top is usually THE top for non-kit coins, and kit coins already keep 100% on the 50% trail, so any bag rule that replaces the kit trail gives up most of the runner.
- **Variant (non-kit only, kit keeps the trail):** still loses in both halves: bag 10% -$283 / -$144 a day, 20% -$568 / -$289, 33% -$942 / -$480. No bag size helps.
- **A2.32 SECOND TICKET at $5M: FAIL by the pre-registered rule, but the only live-looking idea in this sprint.** One more $1k when a traded coin crosses $5M with net buying in the last hour, 50% trail or pull exit: half A +$17 a day (15 add-ons, mean 1.02x, beats random 93%, bar 95%), half B +$1,977 a day (8 add-ons, mean 2.48x, beats random 97%). Without the top 1 coin half B is +$1,424; without the top 2, +$962. Worst day unchanged (-$2,000 / -$3,161). It fails on half A's random test and on half A size (+$17 is noise).
- **A2.32 at $10M: FAIL.** Half A -$53 a day (6 add-ons, mean 0.85x, beats random 52%), half B +$507 (7 add-ons, mean 1.44x, beats random 68%).
- **Pattern worth one line:** every $5M add-on from 22 to 24 Sep lost (9 of 9, 0.55x to 0.90x, trailed out within 2.5 h); from 25 Sep on, 12 of 14 won (median 2.1x, held 1 to 37 h). Same rule, different regime. Two of half B's top add-ons are still open at data end (0x224a794031d54e49b46919ade2c55ff255dfdd97 3.77x, 0xc38c332012a9116dcade0c7f43b5d3c71799a9f8 3.15x), so half B is partly mark to market.
- **Suggestion (not a rule change):** log the $5M add-on as a PAPER book (w2_v3_addon5m) from now and re-test at 30+ add-ons with a regime split. Do not add it live.
- **Already killed, cited, not redone:** hour 8 entry (V10), regime hold on non-kit (V13), net-buy exit (V14), drip-seller exit (FW A1.28), $750k gate (A2.16 / A2.25), operator flag hold (A2.29).

## Pre-registered rule (published in the status page at 09:55 Bangkok, before any result)
- Universe: the 70 wide live-gate trades (E70). Live entry gate + 30 s, 0.3x stop before 5x, exits at the END of the next block, real tick depth, 1% fee each way, $1k, slot rule (max 3 open; bag or add-on takes no slot). Halves per v1.26: A = gates to 30 Sep (48 trades), B = 1 Oct on (22 trades).
- Data ends 6 Oct 10:58 Bangkok; tracking covers 7 days after each gate, so 14 d is NOT testable. Open positions at data end are marked at a real-sale quote at the last row (9 bags, 2 add-ons).
- A2.31: at the 5x trigger sell (1 minus bag) at the next block end; bag exits at liquidity pull (live definition: pool quote reserve at or under 50% of the entry reserve) or 3 d / 7 d. Cells: bag 10 / 20 / 33% x 3 d / 7 d, applied to ALL 5x coins (kit trail replaced). Variant: non-kit only.
- A2.32: first moment within 7 d after gate where cap crosses $5M or $10M AND net buy $ in the last 60 min is above 0; buy $1k at that moment + 30 s; exit on 50% trail from the add-on peak, pull (reserve at or under 50% of add-on entry reserve), or 7 d / data end. Added on top of the live book. "Held coin" relaxed to "a coin we traded", since live sells non-kit coins at about $5M.
- PASS: $ a day beats live in BOTH halves, worst day no worse than live worst minus $1k, AND beats 95% of 200 random draws (A2.31: same bags, exit at a uniform random time between the 5x sale and the horizon; A2.32: same add-on count in the same coins at a uniform random time in the 7 d window, same exit).

## Live book (baseline, new halves)
| half | trades | $ a day | worst day |
|---|---|---|---|
| A (to 30 Sep) | 48 | +$2,870 | -$2,000 |
| B (1 Oct on) | 22 | +$9,631 | -$3,161 |

## A2.31 cells ($ a day vs live)
| horizon | bag | half A | half B | beats random A / B | without top 2: A / B | non-kit only: A / B |
|---|---|---|---|---|---|---|
| 3 d | 10% | -$905 | -$4,634 | 0% / 4% | -$1,014 / -$5,223 | -$283 / -$144 |
| 3 d | 20% | -$1,351 | -$4,511 | 0% / 2% | -$1,435 / -$5,027 | -$568 / -$289 |
| 3 d | 33% | -$1,937 | -$4,378 | 0% / 4% | -$1,937 / -$5,446 | -$942 / -$480 |
| 7 d | 10% | -$924 | -$4,634 | 0% / 8% | -$1,034 / -$5,223 | -$283 / -$144 |
| 7 d | 20% | -$1,390 | -$4,511 | 0% / 11% | -$1,474 / -$5,027 | -$568 / -$289 |
| 7 d | 33% | -$2,002 | -$4,378 | 0% / 10% | -$2,002 / -$5,446 | -$942 / -$480 |

Bag value at exit (bag only, per $1 of ticket it stands for): median 0.36x to 0.38x, mean 2.1x to 2.3x (mean carried by one open Kit B coin 0x8e41fd25e85e624560fa50ff9f117a02cd30b8b3 at 30.6x, marked not sold). Exit reasons at 7 d: 75 of 84 bag rows pull, 9 open at data end, 0 reached the 7 d horizon.

## A2.32 cells (add-ons on top of live)
| trigger | add-ons A / B | mean x A / B | $ a day A / B | beats random A / B | without top 1: B | without top 2: B |
|---|---|---|---|---|---|---|
| $5M + net buy 1 h | 15 / 8 | 1.02 / 2.48 | +$17 / +$1,977 | 93% / 97% | +$1,424 | +$962 |
| $10M + net buy 1 h | 6 / 7 | 0.85 / 1.44 | -$53 / +$507 | 52% / 68% | +$348 | +$193 |

Top add-ons at $5M: 0xee049979960d5e42b389b941ec45790ca7b6e1f7 4.32x (30 h), 0x224a794031d54e49b46919ade2c55ff255dfdd97 3.77x (open), 0xc38c332012a9116dcade0c7f43b5d3c71799a9f8 3.15x (open), 0xe0cd4e72769e7317891d1e49905878b7b14c67ac 2.38x (36 h), 0xf15aab29c3d22cf2b469d4a9b0c0a72f31ba0711 2.35x (37 h).

## Files
- Bag rows: http://159.223.88.168/hx-c9ois/sw/sw-a2-31-bags.csv ; cells: http://159.223.88.168/hx-c9ois/sw/sw-a2-31-cells.csv
- Add-on rows: http://159.223.88.168/hx-c9ois/sw/sw-a2-32-addons.csv ; cells: http://159.223.88.168/hx-c9ois/sw/sw-a2-32-cells.csv
- Code (container): /tmp/w2/a231/sim.py, book.py, addon.py, book32.py. Live baseline reuses A2.27 trades (matches A2.24 exactly).

## First-principles score
- Mechanism check: a bag only pays if the coin keeps value after 5x. Measured: 25 of 28 fall to the pull line within hours. The identity holds: bag value = bag x price at exit; price at exit is about 0.36x of entry for most. Nothing to tune.
- Add-on: pays only if strength at $5M predicts more strength. Half B says yes, half A says no, and the random test in A fails. Not enough coins (23) to separate rule from regime.
- Honesty: open positions are marked, not counted as wins; the 3 best bags and 2 of the best add-ons are open marks.

Analyst 2, 7 Oct 2026 10:35 Bangkok. Latest: http://159.223.88.168/hx-c9ois/sw/sw-a2-31-moonbag.md