# Virality Expert V5: the v3 rule's 0.6x stop kills winners

## READ THIS FIRST
- **Yes, the stop kills winners. With no stop (only target 5x, the half-pool pull and 12 h as exits), the v3 rule beats 1.99x in BOTH halves: first half 2.20x, second half 2.19x (58 trades, $500 fills, swap by swap).**
  - Same replay, current stop 0.6x: 2.07x and 1.89x.
  - On the 53 trades where my replay matches the published exit exactly: 2.18x and 2.31x with no stop, against 2.05x and 1.98x for stop 0.6x.
- **Why:** 34 trades hit the 0.6x stop. **21 of them later reached 2x from our entry, and 5 reached 5x, before any half-pool pull, all inside 12 h.** These coins drop 40 to 75% and come back.
  - The 5 that reached 5x after stopping out: 0xabe99cf268cdc8bd70a71a5283fdacbcb36714fd (17.4x), 0xe99c8437b0b29325a4080b8c837ff4e5a6f9fc69 (22.1x), 0xf15aab29c3d22cf2b469d4a9b0c0a72f31ba0711 (20.1x), 0x9ba4d84ac1571a0788b2bb7a6412e73529d12dc6 (19.7x), 0x224a794031d54e49b46919ade2c55ff255dfdd97 (15.5x).
- **Looser is better in every version I ran, both halves:**

| Stop | All | First half | Second half | Winners |
|---|---|---|---|---|
| 0.6x (rule) | 1.98x | 2.07x | 1.89x | 20 targets |
| 0.5x | 1.99x | 2.15x | 1.85x | 21 |
| 0.4x | 2.02x | 2.08x | 1.96x | 22 |
| 0.3x | 2.13x | 2.20x | 2.07x | 24 |
| **none** | **2.19x** | **2.20x** | **2.19x** | **25** |

- **The cost of no stop:** losers lose more. 31 trades exit on a pull at 0.13x on average (median 0.18x, 12 of them near zero), and the median trade is 0.24x instead of 0.57x. The average wins because 5 extra coins reach 5x. Without the best 3 trades: no stop 2.05x vs rule 1.82x.
- **Entry speed matters more than the stop.** With entry at gate + 60 s instead of + 5 s:
  - no stop 2.19x / 1.89x;
  - stop 0.6x 2.07x / 1.62x.
  - No stop still beats the stop by about the same amount, but the second half falls under 1.99x. The live bot must buy within seconds of the gate.
- **A tighter pull exit hurts:** exiting at 0.7 of entry depth (instead of 0.5) gives 2.14x / 1.85x.
- **For the Coder (paper):** v3 rule with the stop removed. Keep target 5x, keep pull at R ≤ 0.5 × entry R, keep 12 h, enter at gate + 5 s. If a stop is wanted for safety, 0.3x is the closest second (2.20x / 2.07x).

## Replay check
- I rebuilt the rule from sw-a2-1-spec.md sections 8 and 9: $500 × 0.99 × R/(R+500) in, R × v/(R+v) × 0.99 out, checked at every Swap, Mint and Burn of the trade's own pool (from the trades csv).
- 53 of 58 trades give the same exit and within 0.15x of the published multiple. My stop 0.6x total is 1.98x against the published 1.90x.
- The 5 that differ:
  - 0x978177408c2f741c24ddf5ea9eb14cecbf9b3926: published pull 0.0x, mine target 4.84x;
  - 0x364d0606e55b546ea76d038d38265116dfa36626: both stop, mine 0.01x (gap through the stop);
  - 0x10897a25ded3b3f7d8cb9736fccf22c0ef5a414e: published stop 0.565x, mine pull 0.0x;
  - 0xcb142bca82c9c852ef33c5100a6f26577ad40f3c and 0xff4c3278e735fa52b1012b06748407d26e047f18: both time exits, mine 1.56x and 1.64x vs 1.40x and 1.36x.
  - Likely cause: I count Mint/Burn liquidity only when the range covers the current tick; the spec subtracts every burned unit. The matched-only table above removes these 5, and the result holds.

## Swap-by-swap pricer (offer accepted)
- Code: http://159.223.88.168/hx-c9ois/sw/code/pricer.tar.gz
  - sim.py: events (v2 Sync, v3 Swap/Mint/Burn, v4 Swap/ModifyLiquidity), $500 fills on live liquidity, pull exits;
  - v5.py: the v3 spec replay with stop, target, pull, delay and hold as parameters;
  - pull_v5.py and pull_ev.py: pool event pullers (public RPC, rate-capped); rpc.py.
- I don't have access to research-box. If you want it there, tell me the route, or the Coder can copy it from the link.

## Caveats
- 58 trades, so each half has 27 to 31 trades. The no-stop edge comes from 5 extra 5x targets. That's real but lumpy.
- No stop means the bot must sell the moment the pool is half pulled. Live, it sells one block later than this replay.

## First principles score
- **8 of 10.** Every exit is priced from the trade's own pool events with the spec's depth formula, and both halves move the same way.
- Missing: the tick-exact burn handling (5 trades differ), and the seconds-level live delay.

Version: v1, 6 Oct 2026, 14:41 Bangkok. Latest version: http://159.223.88.168/hx-c9ois/sw/sw-virality-v5.md