# A2.27 The exit for non-kit coins (Analyst 2)

## Pre registered rule (written 18:25 Bangkok, 6 Oct, BEFORE any cell was run)
- Live: wide gate ($1M, 3 h, 0.08, 0.015), buy at gate + 30 s, exits at the end of the next block, real tick depth, 1% fee each way, $1k, slot rule (3 normal + 2 holds). Kit A and Kit B coins keep the live kit hold. ONLY the exit of the 34 non-kit coins changes, one change at a time.
- Cells for non-kit coins: sell all at 3x, 4x, 5x (live), 7x, 10x; 5x sell 80% and keep 20% on a 50% trail from peak (or pull, or 48 h); 3x sell 50% and the rest at 5x; stop 0.4x and 0.25x (instead of 0.3x, target stays 5x).
- Before the first target, every cell keeps the live guards: stop, pull (pool quote side at or under 50% of entry), 12 h time exit.
- Slot rule: the 20% runner after 5x uses a hold slot; if both hold slots are full, it sells all at 5x (same as a kit coin). Partial sells are priced on real depth for the part sold, each from that block's pool state.
- A cell PASSES only if ALL of: (1) $ a day beats live in BOTH halves; (2) beats the random-half baseline in BOTH halves: the new exit applied to a random half of the 34 non-kit coins (200 draws), and the full cell must beat the MEDIAN draw; (3) the gain over live stays above $0 in both halves after removing the 2 coins with the biggest gain in that half; (4) worst day -$3,951 or better.

## Answer in one line (18:26 Bangkok)
Keep the live exit. Every target change LOSES money. Only stop 0.4x passes the rule, by a hair (+$81 / +$27 a day). Even that should NOT go live: kit coins show real winners do dip into the 0.3x to 0.4x band before running.

## All cells (non-kit coins only change; $1k; every friction; slot rule)
| cell | usd_day_A | usd_day_B | rand_med_A | rand_med_B | gain_wo_top2_A | gain_wo_top2_B | worst_A | worst_B | dd_A | dd_B | pass |
|---|---|---|---|---|---|---|---|---|---|---|---|
| T5 live | 3317 | 6683 | | | | | 0 | -3161 | -4257 | -4700 | base |
| T3 | 2370 | 6072 | 2846 | 6474 | -1318 | -611 | 0 | -3161 | -4257 | -4916 | False |
| T4 | 2556 | 6382 | 2952 | 6493 | -761 | -301 | -526 | -3161 | -4257 | -4700 | False |
| T7 | 3208 | 7004 | 3260 | 6803 | -387 | -94 | 0 | -3161 | -4257 | -4700 | False |
| T10 | 2041 | 7596 | 2688 | 7114 | -1929 | -94 | -4424 | -3161 | -7124 | -4700 | False |
| run20 | 3104 | 6731 | 3215 | 6726 | -213 | -47 | -8 | -3161 | -4257 | -4700 | False |
| half3 | 2894 | 6398 | 3103 | 6495 | -611 | -285 | 0 | -3161 | -4257 | -4700 | False |
| stop0.4 | 3398 | 6710 | 3358 | 6695 | 43 | 5 | 0 | -3161 | -4116 | -4651 | True |
| stop0.25 | 3279 | 6671 | 3298 | 6676 | -38 | -12 | 0 | -3161 | -4373 | -4700 | False |

rand_med = median of 200 draws with the new exit on a random 17 of the 34 non-kit coins. gain_wo_top2 = gain over live, $ a day, without the 2 biggest gains in that half.

## Targets: selling sooner or later both lose
- 3x and 4x: win rate goes up (47% in A at 3x) but the winners are cut short: -$947 / -$611 a day at 3x.
- 7x: half B up $321 a day, half A down $109; without its top 2 coins it loses in both halves. 10x: half B up $913 from 1 or 2 coins, half A down $1,276, worst day -$4,424 (breaks the floor).
- Partial sells lose too: 5x sell 80% plus a 20% runner -$213 / +$48 a day; 3x sell half plus the rest at 5x -$423 / -$285 a day.
- So for non-kit coins, 5x sell-all is the top. The money above 5x lives in the kit coins, and those already hold.

## Stop: 0.4x passes, but do not use it
- 0.4x vs 0.3x: +$3,398 / +$6,710 a day vs +$3,317 / +$6,683. It beats the random median in both halves ($3,358 / $6,695). Without the top 2 it is still up, but barely: +$43 / +$5. Worst day unchanged.
- 11 of 34 non-kit trades change. All 11 are stops that get a slightly better price (for example 0xcc4cb99ef15e9799e1d7e45058683a72daf5b022 0.055x to 0.336x, 0xc3ba8a6f8d011881ceca8ad6e2b524232a40aa26 0.110x to 0.371x). In 23 days, no non-kit winner was lost.
- Why it should still not go live: of 29 coins that reached 5x (all coins), the dips before 5x were as low as 0.308x (0x9ba4d84ac1571a0788b2bb7a6412e73529d12dc6, Kit B) and 0.357x (0xe99c8437b0b29325a4080b8c837ff4e5a6f9fc69, Kit A). A 0.4x stop kills both. They are worth about +$12k at $1k, against +$1,380 saved over all 23 days by the 0.4x stop.
- The closest non-kit winner dipped to 0.503x (0xabe99cf268cdc8bd70a71a5283fdacbcb36714fd). With only 13 non-kit winners, "no non-kit winner dipped into 0.3x to 0.4x" is likely luck. Kit coins had 2 of 16. At that rate the 0.4x stop would cost about 1 to 2 winners per 23 days (about -$4k to -$8k) to save $1,380.
- Stop 0.25x: -$38 / -$12 a day. Fail.

## What this means
- Keep the live exit: 0.3x stop, sell all at 5x for non-kit coins, kit hold for Kit A and Kit B.
- The exit is done. Neither the entry (A2.25, A2.26) nor the exit (A2.27) has money left in it. The levers left are size and getting live.

## Limits
- 34 non-kit coins, 13 winners, 23 days.
- Runner and partial sells priced from each block's pool state for the part sold (our own earlier sell is not in the later state).

## First principles score: 90%
Exit math is simple: a tighter target cuts winners, a looser one waits on coins that fade, and a tighter stop only pays if no winner dips into the band. The band test uses all 29 coins that reached 5x, not just the 13 in this cell. Rule written before the run.

Files: http://159.223.88.168/hx-c9ois/sw/sw-a2-27-cells.csv

Analyst 2, 6 Oct 2026 18:26 Bangkok.