# A3.5: audit of Virality V5, "the 0.6x stop kills winners" (6 Oct 2026, Bangkok)

## Read this first
- **V5 CONFIRMED in direction, but the gain is about half the size V5 shows.**
  - My independent rebuild of the 58 trades: no stop gives **2.11x** (first half 2.20x, second half 2.03x), against 2.00x (2.11x / 1.90x) with the 0.6x stop.
  - V5 says 2.19x (2.20x / 2.19x). The second half is where we differ.
- **No stop beats the 0.6x stop in both halves, and still wins without the best 3 trades** (1.95x vs 1.84x). The cost is real: the median trade drops from 0.57x to 0.24x, and the worst day goes from −$937 to −$1,261 at $500.
- **Not a smooth curve.** A 0.4x stop is WORSE than 0.6x in both halves (1.93x). 0.3x is the best (2.13x). Whether a stop pays depends on a few coins that dip just past a level and come back, so the exact level is fragile.
- **0xabe99cf268cdc8bd70a71a5283fdacbcb36714fd is a coin, not a wallet.** It has 3,554 bytes of contract code and is the token side of its own v3 WETH pool. V5 is right; Analyst 1's A1.11 label "farm wallet" is the mix-up. That is 1 address. A wallet list built from token addresses could have more of these; A1 should check its list against eth_getCode.
- **Pull exits are priced correctly.** With no stop, 32 trades exit on a pull at 0.134x on average (V5: 0.13x). Each is priced at the depth left in the pool right after the pull (median $24,693), not at 0. 5 pools were emptied to zero depth, and those really are worth 0.

## Stop grid on my rebuild (58 trades, $500, range-aware depth, target 5x, half-pool pull in range only, 12 h)
| Stop | All | First half | Second half | Without best 3 (A / B) | Median trade | Worst day | Targets | Near zero |
|---|---|---|---|---|---|---|---|---|
| 0.6x (rule) | 1.998 | 2.112 | 1.897 | 1.837 (1.886 / 1.797) | 0.573 | −$937 | 20 | 5 |
| 0.4x | 1.932 | 2.085 | 1.799 | 1.768 (1.855 / 1.695) | 0.380 | −$1,120 | 21 | 11 |
| 0.3x | **2.130** | 2.219 | 2.052 | 1.977 (2.001 / 1.957) | 0.290 | −$1,212 | 24 | 11 |
| none | 2.108 | 2.199 | 2.029 | 1.954 (1.979 / 1.933) | 0.240 | −$1,261 | 24 | 12 |

## The 5 "stopped, then 5x" coins V5 names
| Coin | 0.6x stop | No stop |
|---|---|---|
| 0xabe99cf268cdc8bd70a71a5283fdacbcb36714fd | stop 0.575x | 5x target 4.773x (23 Sep 09:11) |
| 0xe99c8437b0b29325a4080b8c837ff4e5a6f9fc69 | stop 0.578x | 5x target 4.795x (26 Sep 18:58) |
| 0xf15aab29c3d22cf2b469d4a9b0c0a72f31ba0711 | stop 0.538x | 5x target 4.768x (1 Oct 23:00) |
| 0x9ba4d84ac1571a0788b2bb7a6412e73529d12dc6 | stop 0.568x | 5x target 4.754x (5 Oct 00:15) |
| 0x224a794031d54e49b46919ade2c55ff255dfdd97 | **not traded in my rebuild** (top receiver 0.083 > 0.08, see A3.1) | not traded |
- 4 of the 5 confirmed. The fifth is the knife-edge coin from A3.1. It is in Analyst 2's list but not in mine, and its 5x adds to V5's second half. That plus range-aware depth explains most of the gap (V5 2.19x vs my 2.03x).

## Verdict for the live rule
- Dropping the stop is supported, but the gain is lumpy: +4 targets, all from coins that fell 40 to 70% and came back.
- **Safer middle:** stop 0.3x is the top row and keeps a floor (median trade 0.29x instead of 0.24x).
- I'd run **no stop** and **0.3x** as two paper books before choosing. 4 extra winners in 17 days is too few to separate them.

## First principles score
- 8/10. Same independent rebuild as A3.1, with only the stop changed. Exits are priced at real post-event depth.
- The gap to V5 is explained: one knife-edge coin plus burn handling.

Version 1, 6 Oct 2026 15:10 Bangkok, Analyst 3. Latest: http://159.223.88.168/hx-c9ois/sw/sw-a3-5-v5-audit.md