# A1.15 Buy Kit coins early? (Analyst 1, 6 Oct 2026, 16:05 Bangkok)

## Read this first
- **No. No early cell beats the wide rule in both halves. Buying Kit B early loses money.** The $1M gate is what makes the money: it waits until the dev has finished dumping.
- **Why, from first principles: Kit B devs cash out before $1M.**
  - Each launch's insider wallets (the deployer plus the 0.01 ETH bundle wallets) put in a median 10 ETH.
  - In 93% of Kit B launches they sold more ETH than they put in before the coin reached $1M. The median insider took out **5.9x** what they bought. In Kit A it was 68% and 4.0x.
  - An early buyer is the person they sell to. Only 24% of Kit B coins ever reach $1M (Kit A 50%), and 32% get pulled within 12 h (Kit A 73%).
  - Kit B coins bought at $250k where the insiders had dumped more than half their ETH (101 of 108): **0.88x**. The 7 where they hadn't: 1.57x (too few to use).
- **$ a day at $500** (real depth, live book exits: 5x, 0.3x stop filled on the next block, pull = 0, 12 h). Halves: Kit A = 24 Sep to 1 Oct, Kit B = 1 to 6 Oct.

| Entry | Exit | Kit A $ a day (worst day) | Kit B $ a day (worst day) | Wide rule, same days A / B |
|---|---|---|---|---|
| first outside swap + 5 s | 5x | +$1,867 (−$1,444) | +$560 (−$5,641) | +$2,035 / +$2,982 |
| first outside swap + 5 s | at the $1M gate | −$186 (−$2,559) | +$8,394 (−$7,095) | same |
| first outside swap + 5 s | half at $1M, rest 5x | +$914 | +$212 | same |
| cap passes $250k | 5x | **+$2,698** (−$640) | **−$698** (−$5,054) | same |
| cap passes $250k | half at $1M, rest 5x | +$1,638 ($0) | −$1,197 | same |
| $500k sellable | 5x | +$2,483 (−$1,663) | −$1,116 (−$6,819) | same |

- The one big Kit B number (+$8,394 a day, entry at the first outside swap, exit at $1M) is two coins: 76.7x on `0x0f6741982785a62e850b611cd012eb798b738037` and 71.2x on `0xdcc026752d7cbf4fe2a14d53986bd3d22de72024`. Without those two it is negative. In Kit A the same cell loses $186 a day. This is a lottery, not a rule.
- In Kit A, the $250k entry with a 5x exit (+$2,698 a day) beat the wide rule (+$2,035), but it fails in Kit B. The median coin in every early cell ends at the 0.3x stop.
- $100 and $1k give the same picture. Every early cell is negative in Kit B except the two-coin lottery cell.
- **Buying after the insiders finish selling doesn't help.** Entry when insiders have sold 50% or 80% of their tokens and the cap is still at least $250k gives the same result as the $250k entry (A +$2,700, B −$931 a day). The dumping is mostly over before $250k anyway.
- **Overlap with the first window's mc5 rule** (buy v3 WETH coins at first swap + 5 min if the cap is $250k to $1M):
  - 127 of 246 Kit launches pass the mc5 entry (A 63 of 107, B 64 of 139).
  - **30 of the 36 Kit coins our wide rule traded (83%) were also mc5 entries.** The two windows mostly trade the same kit coins, mc5 earlier and us at the gate. Combined size per coin should be capped across both books.

## Method
- **Data:** all 246 Kit A and Kit B pools (one Kit B pool had no swaps). Every Swap, Mint and Burn from launch to +12 h (1.13M swaps, read from chain).
- **Price and depth:** from each swap's sqrtPrice and active liquidity L. Virtual ETH reserve R = L ÷ √P (WETH token0) or L × √P (WETH token1). Supply from totalSupply (all 18 decimals). ETH = $2,650.
- **Insiders:** the deployer, every receiver and signer of a buy in the first 30 blocks (3 s), and receivers of tokens from the deployer or the kit contract. This found 100% of the Kit B disperse wallets in the 8 launches I checked.
- **Insider sells:** every token transfer out of an insider wallet, matched to the swap ETH in the same transaction.
- **Fills:** a $X buy gets tokens = X ÷ price × R ÷ (R + X). Sells use the same curve at the exit's R. 0.3% fee each side. Take profit and stop fill at the last swap of the next block after the trigger (the cascade trap). Pull = cumulative burns reach 50% of minted liquidity, and anything still held is 0. Time stop = last price at 12 h.
- **"$500k sellable"** = cap at least $500k and virtual ETH reserve at least 5 ETH.
- **Wide rule** = Analyst 2's 70 trades, x at $500 scaled linearly to $100 and $1k (A2.18 says real $1k is a bit lower).

## Limits
- Kit B covers only 6 days. A 12 h window cuts off later runs.
- The insider set may include a few outside snipers from the first 3 s. That makes the dump numbers conservative for "outsiders".

## Asks for Supervisor
- **Keep the wide gate. Don't add an early Kit B entry.** The gate filters for survivors after the dev has cashed out.
- Coordinate with the First Window Supervisor: 83% of our kit trades are also mc5 trades. Cap the combined exposure per coin, and make sure the two bots don't sell into each other.
- Optional paper only: Kit A "$250k + 5x" won in its own half. If Kit A comes back, log it as a paper cell, but never on Kit B.

First principles score: 85%. Every number comes from the pools' own swaps, depth and liquidity events, and the insider dump comes from the insiders' own transfers. Missing: runs after 12 h, and live fills.

Scripts: /root/a1/a115/fetch.py, px.py, ins.py, insell.py, sim.py, summ.py, done.py (review container). Latest: http://159.223.88.168/hx-c9ois/sw/a1-15-kit-early.md