# A1.16 Cascade funders (Analyst 1, 6 Oct 2026, 16:15 Bangkok)

## Read this first
- **Yes, one common funder: the same disperse contract that funds Kit B.** All 60 traced cascade sellers (20 per coin, 3 coins) got their first ETH from `0x84157d3d24630c297849c84d49ae7f1805c1b60f` through disperse transactions (selector 0xe63d38ed), with one caller per coin:
  - `0x2a6efa703d8f1f06da07d1960711f02b1ba66f10`: caller `0x081d46bd98f0cac0270a455efb13c5c459070e0d`, 10 disperse transactions, 0.21 to 0.47 ETH per wallet.
  - `0xf5a9da666eb257c0859b47efe8c2cb6745baff94`: caller `0xb6de6a038fe9469e7172c2a3796c4dd14124106b`, 11 transactions, 0.03 to 0.40 ETH per wallet.
  - `0xfe0f309cee96912d9926b8f0786fe3347364e678`: caller `0x934ade03dc73d1c030ac83c71190ebf21b6d52e4`, 16 transactions, 0.04 to 0.44 ETH per wallet.
  - The funding comes in batches spread over 7 to 95 minutes. It has to land before each wallet can buy, and it lands 2 minutes to 5 hours before the cascade.
- **But it is not a skip signal, because the same funder sits behind a third of our winners.**
  - I checked 793 random single-buy wallets across all 70 wide-rule coins.
  - The split is all or nothing: in **23 coins essentially every sampled wallet was funded by `0x84157d3d24630c297849c84d49ae7f1805c1b60f`**, and in the other 47 none was.
  - The 23 "farm-funded" coins are 14 of the 15 Kit B coins plus 11 others. No Kit A coin is among them.
  - **Farm-funded coins did better, not worse:** half B 2.40x (20 coins, 45% hit 5x) vs 1.81x for the rest (14 coins). Half A: 1.79x (3 coins) vs 2.14x.
  - All 3 fresh-wallet cascades are farm-funded coins: 3 of 23 cascaded, vs 0 of 47.
  - **Skipping farm-funded coins would throw away about $1,974 a day in half B** (and $170 in half A) at $500. The cascade cost is about $77 a day (Analyst 2).
- **What it means, from first principles:** the farm wallets are the dev's own volume and holder count. They buy one at a time to look like a crowd, and they sell together when the dev decides. In most farm coins the dev lets the coin run first (that is where our 5x trades come from). In 3 of 23 the farm dumped while we held. Nothing in the funding tells which kind a coin will be.
- **So close cascades as a skip rule.** One cheap tag is worth keeping: `farm_funded` = more than half of a coin's single-buy wallets funded via `0x84157d3d24630c297849c84d49ae7f1805c1b60f`. Log it for tracking only.

## Method
- **Sellers:** token Transfer logs in each cascade block into SwapRouter02 `0xcaf681a66d020601342297493863e78c959e5cb2` or the pool. That gives 208, 119 and 122 sellers, matching Analyst 2. I traced the first 20 per coin.
- **First funder:** binary search on eth_getBalance for the first block with ETH, then the call trace (debug_traceTransaction, callTracer) of that block to find who sent it. In 60 of 60 the value came from `0x84157d3d24630c297849c84d49ae7f1805c1b60f` inside a disperse call.
- **Spread check:** for every wide coin, up to 12 random wallets that bought once and never bought another wide coin. Their first-funding block was checked for a transaction to `0x84157d3d24630c297849c84d49ae7f1805c1b60f` whose input contains the wallet. 793 wallets, 0 errors.
- **Money:** Analyst 2's wide trades at $500 (x_500_stop03), with days per half as in A1.14.

## Limits
- I sampled 12 wallets per coin, not every buyer. The all-or-nothing split makes that enough to class a coin, but the exact farm share per coin is not measured.
- Only 3 fresh-wallet cascades so far, so "3 of 23" is a small count.

## Asks for Supervisor
- Close cascades as a skip rule (no common-funder skip pays). Coder: add a `farm_funded` tag (tracking only).
- Analyst 2's A2.11 live count still matters: if farm-funded coins start cascading more than about 1 in 4, re-check. That is when a skip could start to pay.

First principles score: 85%. Each seller's funding comes from the call trace of its first-ETH block, and the spread check reads the disperse calls directly. Missing: full buyer coverage per coin, and more cascades.

Scripts: /root/a1/a116/sellers.py, fund.py, scan.py (review container). Latest: http://159.223.88.168/hx-c9ois/sw/a1-16-cascade-funders.md