# A2.35 Trail shape (Analyst 2)

7 Oct 2026, 10:01 Bangkok (server clock). Pre-registered 09:59 in the status page, before any result.

## Answer
**FAIL on all 4 cells. Kit hold is tuned, stop. Keep the live 50% trail.**

Every cell loses in half B. A tighter trail sells the biggest half B runner, 0xee049979960d5e42b389b941ec45790ca7b6e1f7, too early. In live it sells at 15.78x. Here it sells at 5.73x to 12.91x.

Closest miss: **stepped 50/35** (sell at 0.5 x peak until the peak reaches 10x our entry, then at 0.65 x peak).
- Half A: +$476 a day. It beats random 94.5% of the time; the bar is 95%.
- Half B: -$51 a day.
- It does better than live on 7 coins and worse on 2. Too small and too split to switch.

## Cells ($ a day vs live, $1k, live A +$2,870 / B +$9,631)
| cell | A | B | worst A | worst B | beats random A / B | without top 1 (A / B) | without top 2 (A / B) | coins better / worse | verdict |
|---|---|---|---|---|---|---|---|---|---|
| trail 40% (0.6 x peak) | +461 | -1,093 | -2,000 | -3,161 | 94.5% / 30% | +461 / -1,452 | +352 / -1,452 | 11 / 2 | FAIL |
| trail 30% (0.7 x peak) | +205 | -3,365 | -2,000 | -3,161 | 88% / 3% | -14 / -3,365 | -190 / -3,365 | 7 / 6 | FAIL |
| stepped 50/35 | +476 | -51 | -2,000 | -3,161 | 94.5% / 46% | +476 / -530 | +309 / -530 | 7 / 2 | FAIL (near miss) |
| stepped 50/25 | +802 | -451 | -2,000 | -3,161 | 99% / 45% | +802 / -1,223 | +540 / -1,223 | 7 / 2 | FAIL |

## What the coins say
- **Half B is decided by 3 coins plus the slot rule.**
  - Under stepped 50/35:
    - 0xee049979960d5e42b389b941ec45790ca7b6e1f7 falls from 15.78x to 12.91x;
    - 0x8e41fd25e85e624560fa50ff9f117a02cd30b8b3 rises from 9.30x to 12.17x;
    - 0xf15aab29c3d22cf2b469d4a9b0c0a72f31ba0711 falls from 11.69x to 11.38x.
  - 0x9ba4d84ac1571a0788b2bb7a6412e73529d12dc6 would rise from 10.11x to 13.34x, but both hold slots are full. In live, trail 40% and both stepped cells it is force-sold at 5x (4.69x), so that gain never counts. In trail 30% a slot is free, but that cell loses the most.
- **Half A gains are small but steady.** The tighter step catches the top of mid coins better:
  - 0xe99c8437b0b29325a4080b8c837ff4e5a6f9fc69: 9.26x to 12.09x;
  - 0xe0cd4e72769e7317891d1e49905878b7b14c67ac: 8.41x to 10.48x.
- **A flat tight trail cuts runners at their first shakeout:**
  - with the 30% trail, 0x8e41fd25e85e624560fa50ff9f117a02cd30b8b3 goes from 9.30x to 3.74x;
  - with the 40% trail, it goes to 3.39x.
- **A2.34 plus A2.35 together:**
  - looser trails lose in both halves;
  - tighter or stepped trails win only in half A;
  - so 50% is the best single setting in this data.

## Limits
- Only 15 coins (9 in half A, 6 in half B).
- 2 coins are censored at data end (0x224a794031d54e49b46919ade2c55ff255dfdd97, 0xc38c332012a9116dcade0c7f43b5d3c71799a9f8).
- Stepped 50/35 can be re-tested for free once 25+ kit coins have hit 5x (about 10 days).

## Files
- http://159.223.88.168/hx-c9ois/sw/sw-a2-35-cells.csv
- http://159.223.88.168/hx-c9ois/sw/sw-a2-35-coins.csv (per coin x and exit reason for every cell)