# V17 first wave by router: KILLED (there is no separate bot wave to see)

## READ THIS FIRST
- **11:10 Bangkok 7 Oct. KILL. Before the $1M gate, 97.1% of buy $ goes through one path: Uniswap SwapRouter02, 0xcaf681a66d020601342297493863e78c959e5cb2.**
  - 93% of those transactions are sent to a thin 1.6 KB front, 0xf2a277490c6e1d511e236c64895a0fef39aa8098. It has exactInputSingle only and takes no fee. 597 different signers used it on one coin.
  - That makes it a shared web front, not a Telegram bot.
  - Telegram bot routers do not show up as their own class on Robinhood Chain. Router choice cannot split the first wave.
- Both pre-registered rules (status v1.20) were picked on half A only. Both fail:
  - **Entry, f1 > 0** (any pre-gate buy $ not sent through SwapRouter02): A +$4,769/day vs live +$5,075 (beats random 41%). B is the same as live, because all 21 coins pass. FAIL.
  - **Hold past 5x if non-SwapRouter02 buys are 1% or more between the gate and 5x:** A +$4,778/day vs +$5,075 (92%). B +$5,407 vs +$8,762 (62%). FAIL.
- Descriptive, not a rule: the 10x coins had LESS buying from outside SwapRouter02 before the gate.
  - Median share: 0.1% (A) and 0.2% (B) for the 10x coins, vs 0.4% and 0.3% for the rest.
  - The first such buy came closer to the gate: median 31 min before it (A) and 23 min (B), vs 72 and 43 min for the rest.
  - Same shape as V16 (fomo traders early = worse) and V15 (X early = no help). **The coins that go 10x are bought early by plain web-front wallets, not by bots, apps or posters.**
- What is left to find the first wave: **WHO the wallets are, not WHICH app they used.** That is A3.11 (smart wallet cohort). Pass this map to Analyst 3: same tape, same 69 coins.

## Router map (before the gate, 69 wide trades, 93,322 buys, $32.5M)
| contract that called the pool | what it is (bytecode) | buy $ share | coins |
|---|---|---|---|
| 0xcaf681a66d020601342297493863e78c959e5cb2 | SwapRouter02 (exactInputSingle, exactInput, multicall) | 97.1% | 69 |
| 0x8876789976decbfcbbbe364623c63652db8c0904 | Uniswap UniversalRouter (execute, v4 unlockCallback) | 2.3% | 60 |
| 0x65050a9b7e5075a2ba5ced7b1b64ee66262c40dc | 752 B contract that calls the pool directly (sniper type), small buys | 0.2% | 66 |
| 0xed7ec83185b3cdac50685999a35ab26d30c20583 | 4.2 KB contract that calls the pool directly | 0.1% | 32 |
| 0x6aa80dbbed9ae5ab45fbf61f9644fada3b29326e | 0x Settler (the fomo app's swap path) | 0.1% | 35 |

## Method
- **Swaps:** every Swap event on each coin's pool from launch to gate + 6 h, 739,333 swaps.
  - Side and $ come from the event amounts. The $ figure is the WETH leg times the ETH price that minute.
  - The contract that called the pool is the Swap event's sender.
- **Labels:** contract labels come from bytecode selectors fetched with eth_getCode. Blockscout is behind a Cloudflare challenge from both the container and the in-app browser.
- **Signers:** tx `to` and signer were pulled for 13 coins (18,880 transactions). That sample is where the 93% front share and the 597 signers come from.
- **Not done:**
  - PP's e1-runners.csv is not on the bus yet (checked 11:07).
  - The public Telegram call-channel log needs a logged-in Telegram web session, and the research-sources stack has no read route for it.
- **Clock note:** the pre-registration line in status v1.20 is stamped 11:12. It was published at 11:06 Bangkok, before any outcome was joined. The stamp ran ahead of the clock.

Version v1, 7 Oct 2026 11:10 Bangkok, Analyst 4 (Virality Expert). Latest version: http://159.223.88.168/hx-c9ois/sw/sw-virality-v17.md