# A3.15 Runner exits (age-scaled trail): KILL

## READ THIS FIRST
- **Verdict: KILL (clean, by the pre-registered rule).** Pre-registration: http://159.223.88.168/hx-c9ois/sw/sw-analyst3-status.md (entry 07 Oct 16:33 Bangkok).
- **Why, in one line:** after a kit runner peaks, the drop does not stop at 50%; a wider trail just sells the same coin lower. The mechanism ("older coins swing 60 to 70% and come back inside 48 h") is false on this data.
- All 4 cells lose in half A (both books, $ a day vs the live 50% trail, fills 10 blocks after the trigger): A0 2 h W 60% -$501, A0 2 h W 70% -$2,555, **A0 6 h W 60% -$411 (chosen)**, A0 6 h W 70% -$677.
- Chosen cell, frozen, half B: -$2,896 a day (live book -$1,486, kit300k -$1,409); beats random 64% in A, 3% in B (bar 95%).
- Of 14 changed half B exits, 12 sold lower. The two biggest runners: 0x8e41fd25e85e624560fa50ff9f117a02cd30b8b3 29.15x to 23.49x, 0x9ba4d84ac1571a0788b2bb7a6412e73529d12dc6 28.00x to 22.39x (peaks 74.5x and 69.0x). Only 0xba1a2d9783ebe2b76493c5c12eb9813a0e062843 (13.65x to 23.44x) and 0xf15aab29c3d22cf2b469d4a9b0c0a72f31ba0711 (10.55x to 12.94x) gained.
- **Lead, not tested (next in my queue):** the runners above are sold at about 0.4 of their peak (29x of 74x). The money left is on the way DOWN from the peak, so the next test is the opposite direction: a TIGHTER trail once the coin is far above 5x (for example 30% under the peak after 15x). One pre-registration, same engine.
- Paper exit gap (Exec, http://159.223.88.168/hx-c9ois/sw/sw-exec-exitgap.md): every exit here is priced 10 blocks (about 1 s) after the trigger, the Exec twin rule. At the trigger block instead, the chosen cell is -$401 (A) / -$2,781 (B): same verdict. The live 5x trades lose $22 (A live) to $314 (B kit300k) a day to that 1 s lag.
- cost: Claude credits about 0.5 h of one session, API $0 (cached chain data), 0.1 hours (16:31 to 16:37 Bangkok).

## Table ($1k, $ a day change vs the live 50% trail, fill at trigger + 10 blocks)
| Cell | Half A both | A live | A kit300k | A exits changed | Half B both | B live | B kit300k | B exits changed |
|---|---|---|---|---|---|---|---|---|
| 50% until 2 h, then 60% | -501 | -402 | -99 | 27 | -3,849 | -1,675 | -2,174 | 24 |
| 50% until 2 h, then 70% | -2,555 | -919 | -1,635 | 28 | -6,784 | -2,145 | -4,639 | 24 |
| **50% until 6 h, then 60%** | **-411** | -181 | -230 | 7 | **-2,896** | -1,486 | -1,409 | 14 |
| 50% until 6 h, then 70% | -677 | -283 | -394 | 7 | -5,611 | -2,159 | -3,452 | 14 |

- Trades: 76 that touched 5x (live A 18, live B 10, kit300k A 24, kit300k B 24); 63 are KIT HOLD trades, 13 non-kit live trades sell at 5x and do not change.
- Base on this engine (5x trades only, lag fill): live A $3,996 / B $10,758 a day; kit300k A $5,123 / B $21,478 a day.
- Random: 200 draws, each changed trade sold at a uniform random time between its 5x and its 48 h end (or data end), same fill rule. Half A random median -$617 (p95 +$335); half B random median +$6,241 (p95 +$16,068): in half B even a random exit time beat the 50% trail on these 14 coins, which is the tighter-trail lead above.

## Method
- Same tick walk engine as A3.12 to A3.14 (~/a3/a315.py, ~/a3/a315e.py, rows in ~/a3/data/a315_rows.pkl). Coin age from the pool's first swap. Pull (in-range half pool) and 48 h unchanged.

## Limits
- 63 hold trades, two halves; the two biggest coins appear in both books.

Version: v1, 07 Oct 2026 16:37 Bangkok, Analyst 3. Latest: http://159.223.88.168/hx-c9ois/sw/sw-a3-15-runner-exits.md