# A1.13 v3 WETH: "market cap $250k to $1M at +5 min" as a trade (FW Analyst 1, 6 Oct 2026, 14:56 Bangkok)

> **CORRECTION 15:03 Bangkok (A1.14):** the pull rule here missed most pulls (it measured WETH added, and most launches add 0 WETH). With the fix (Burn of 50%+ of the liquidity units): hold with no stop gets pulled in 68 of 159 trades. **The 0.5x stop is REQUIRED.** Rule numbers at $100 are now 1.78x (H1) and 1.58x (H2). See http://159.223.88.168/hx-c9ois/fw/a1-v3-mc5-fragility.md

**Verdict: PASSES as a right tail feeder, NOT as a take profit trade.** Buy at +5 min, hold to 2 h (no take profit, no stop) made money in both halves: mean 2.08x (first half) and 1.74x (second half) back at $100. It still beats 1.0x with the top 3 coins removed (1.39x / 1.28x). Median trade loses (0.70x / 0.21x). Every take profit version (1.5x, 2x, 3x) is about 0.9x to 1.1x, so it is flat or losing.

## Method (honest fills)
- Universe: 818 v3 WETH pools (13 Sep to 6 Oct) with a market cap of $100k+ at +5 min. Mint and Burn logs were pulled for all of them.
- Entry: the pool state at +5 min (the last swap before 300 s). Market cap uses only swaps before 300 s, so nothing from the future leaks in.
- Depth: in-range liquidity is read from real swaps (L = WETH moved / change in sqrt price), taking the median of the last 8 clean swaps. The swap is run on the virtual reserves.
- Sell value is capped at 90% of the WETH really in the pool, and is only marked on swaps of 0.01+ ETH.
- Costs: pool fee plus GMGN 1% each way, plus $0.05 gas per transaction. The fill lands 3 s after the trigger.
- Pulled = 0: any Burn taking 50%+ of the WETH added so far, before our exit.
- Halves: first half is 13 Sep to 24 Sep (10 trading days), second half is 25 Sep to 6 Oct (12 days). Picks were made on the first half only.
- Sizes: $100 and $500. At $500 results are about 0.1x lower. At half depth, $100 gives the same result and $500 gives 1.51x.

## Results, $100, market cap $250k to $1M at +5 min, no filter
| exit | H1 n (a day) | H1 mean / median | H1 green days | H2 n (a day) | H2 mean / median | H2 green days |
|---|---|---|---|---|---|---|
| hold to 2 h | 44 (3.4) | **2.08x** / 0.70x | 3/10 | 159 (13.2) | **1.74x** / 0.21x | 10/12 |
| hold to 1 h | 44 | 1.22x / 0.74x | 2/10 | 159 | 1.28x / 0.54x | 8/12 |
| stop 0.5x, time 2 h | 44 | 1.78x / 0.49x | 3/10 | 159 | 1.68x / 0.50x | 10/12 |
| stop 0.6x, time 2 h | 44 | 1.75x / 0.58x | 3/10 | 159 | 1.44x / 0.59x | 8/12 |
| take profit 3x, stop 0.5x, 2 h | 44 | 1.19x / 0.49x | 2/10 | 159 | 1.10x / 0.51x | 8/12 |
| take profit 2x, no stop, 2 h | 44 | 1.18x / 0.74x | 3/10 | 159 | 1.02x / 0.98x | 7/12 |
| take profit 1.5x, no stop, 2 h | 44 | 1.00x / 0.86x | 2/10 | 159 | 0.97x / 1.36x | 7/12 |
| take profit 1.5x, stop 0.6x, 1 h | 44 | 0.88x / 0.60x | 1/10 | 159 | 0.96x / 0.72x | 4/12 |

- First half pick (best mean on H1): $250k to $500k, hold 2 h. H1 2.43x (n26), H2 **1.67x** (n97, 8.1 a day, green 7/12).
- The 20+ fresh buyers and under 5% repeat filter does NOT help. With hold 2 h: H1 3.49x on only n8, H2 2.60x (n48). Drop the top 3 coins and it falls to 1.28x / 1.18x, and drop the top 5 and it is under 1.0x. The base rule is more robust.

## How fat is the tail (hold 2 h, $100)
- H1: drop the top 1 coin and the mean is 1.79x, drop the top 3 and it is 1.39x, drop the top 5 and it is 1.07x.
- H2: drop the top 1 coin and the mean is 1.52x, drop the top 3 and it is 1.28x, drop the top 5 and it is 1.13x.
- About 1 coin in 30 gives 10x or more, and those coins carry the result. They are real: 25 to 78 ETH sits in the pool at 2 h and market cap is $1.7M to $17M. For example:
  - 0xb4b0dfeaa2e2a838f0f7a304f005ebe32390d4ef (6 Oct): 35.6x
  - 0xeb1646dbfbeacf76483df1092c208388c029c534 (30 Sep): 21.1x
  - 0x6e8a6a1af8ec53f3f09c7303e5403558dee42d0a (2 Oct): 20.3x
  - 0x7608c38ea8cc2082b043e886185979942cdeb58e (4 Oct): 16.0x
- Dollars a day at $100 a trade:
  - H1: about +$370 a day on the mean, about +$130 a day without the top 3.
  - H2: about +$980 a day on the mean, about +$370 a day without the top 3.
  - Most single trades lose. Trade count jumped from 3.4 to 13.2 a day, so the v3 flow is growing.

## Limits
- The data stops at 2 h, so "hold 2 h" means selling at 2 h. Coins that keep running after 2 h are not counted (upside not captured).
- Our own buy is not fed back into later prices. At $100 this does not matter, because pools hold 10+ ETH.
- Pulls are only measured by the Burn 50% rule. 7 of 44 trades in H1 were pulled and 2 of 159 in H2.
- With a tail this fat, one week of live trading can easily be red. It needs 30+ trades before the sample tells you anything.

## Proposed feeder rule for the 2nd window
At +5 min after the first swap: if market cap is $250k to $1M, buy $100, then hold to 2 h with no take profit. A stop at 0.5x is optional: it costs almost nothing in mean (1.78x / 1.68x) and cuts dead coins early. Paper only until Ben says go.

Code: a1code/v3lp.py, v3sim.py, v3simrep.py, v3diag.py (research-box /root/fwa1).