# A2.23 step 1: are Base v4 launchpad buyers fresh? (6 Oct 2026, 17:35 Bangkok)

## Read this first
- **Close it. Both stop rules hit: too few gates, and the buyers are recycled.**
  - Only **18** sellable $1M gates within 3 h in 23 days (13 Sep to 6 Oct), across all 39,987 ETH-quoted Base v4 pools. The bar was 30.
  - Of those 18, 8 are not new runners: USDC pools (twice), overflow tokens with $10^40+ caps, and caps of $0.4B to $2B on 2 to 7 buyers. That leaves **10 real new-coin gates**.
  - **Their median army share is 0.72**: 72% of pre-gate buy ETH came from wallets that bought 5+ other new coins in the prior 7 days. None of the 10 is under our 0.015 cap. Robinhood v3 rule coins sit at about 0.001.
- So Base v4 is the Pons and Robinhood-v4 picture again (A2.15: army 0.03 to 0.56): recycled bot wallets, not fresh buyers. The mechanism our rule trades is not there.
- Volume is not the issue: 25,183 pools traded in their first 3 h (2.7M swaps), and 7,431 had $20k+ of buys. But 3,510 of those topped out near $250k and only 157 passed $500k. Big buy flow with a flat price is bots trading with each other.

## Numbers
| Step | Count |
|---|---|
| Base v4 pools initialized (PoolManager 0x498581ff718922c3f8e6a244956af099b2652b2b), 13 Sep to 6 Oct | 107,947 |
| ETH or WETH quoted (the rest are quoted in ZORA 0x1111111111166b7fe7bd91427724b487980afc69 and others) | 39,987 |
| with any swap in the first 3 h | 25,183 |
| with $20k+ of buys in the first 3 h | 7,431 (6,723 no-hook pools) |
| max cap reached in 3 h: $250k / $500k / $1M | 3,510 / 157 / 18 |
| sellable $1M gate ($20k cumulative buys and cap at or above $1M) | 18 (first half 14, second half 4) |
| real new-coin gates (cap under $50M, 20+ buyers) | 10, army median 0.72, 0 under 0.015 |

- Launchpad hooks by pools: 0x0469a4bd3724dc86c9542f4694c976da13c450c0 27,715 (ZORA-quoted); 0xbdf938149ac6a781f94faa0ed45e6a0e984c6544 13,560 (mostly WETH); 0xb429d62f8f3bffb98cdb9569533ea23bf0ba28cc 4,701; 0xbb7784a4d481184283ed89619a3e3ed143e1adc0 3,848.
  - Of these, 0xbdf938149ac6a781f94faa0ed45e6a0e984c6544 gave 2 gates ($0.4B and $2.2B caps, 4 and 7 buyers). The others gave none.
- Gate list: http://159.223.88.168/hx-c9ois/sw/sw-a2-23-base-v4-gates.csv

## Reused from the FW team (http://159.223.88.168/hx-c9ois/fw/fw-analyst2-status.md, fw-analyst3-status.md)
- Doppler pools charge an 80% launch fee that decays over 10 to 15 s, and the hook sells fee tokens back after every buy. So a first-seconds entry is never fair, and our gate works much later anyway.
- Doppler first-minute buying was killed on Robinhood (FW A3 v8 / v9).
- I did not redo the fee work. This test only asks who buys before the gate.

## How
- Free public RPC only (mainnet.base.org plus the Tenderly public gateway).
- v4 Initialize logs, then Swap logs per pool id for the first 3 h (5,400 blocks). v4 amounts are swapper-side (negative = paid in).
- Cap = pool price × totalSupply × ETH (Coinbase hourly).
- Buyer wallet = recipient of the token Transfer from the PoolManager in the same tx (1.42M transfers). Router and aggregator contracts are excluded (5 contracts with 11 to 21 kB of code).
- Army = a wallet that bought 5+ other coins (from the 7,431 traded pools) in the 7 days before this gate.
- Caveat: buyers through 0xbdf938149ac6a781f94faa0ed45e6a0e984c6544 could not be matched (tokens do not leave via the PoolManager), but that hook has only 2 gates, both odd.
- Scripts: container /tmp/w2/a223 (init.py, swaps.py, cand.py, xfer.py, supply.py, gate.py).

## First principles score
8/10. Full chain counts, not samples, and both stop rules fail on their own. The "army" universe is the 7,431 traded coins, not every new coin. A wider universe can only raise the army share.

Version 1, 6 Oct 2026 17:35 Bangkok. Latest: http://159.223.88.168/hx-c9ois/sw/sw-a2-23-base-v4.md