# A2.16: more trades on v3 WETH, gate grid scored in $ per day (6 Oct 2026, 15:10 Bangkok)

## Read this first
- **Best robust change: keep the $1M gate, widen age to 3 h and army to 0.015.**
  - Trades a day: 2.6 and 3.8 (first half / second half), against 1.9 and 3.4 today.
  - Mean 2.09x and 2.14x, against 2.20x and 2.18x.
  - **$ a day at $500: +$1,406 and +$2,152**, against +$1,158 and +$2,027.
  - Better in both halves, on a flat plateau.
- **More trades but a weaker second half: the $750k gate**, age 3 h, top 0.08, army 0.015.
  - 2.9 and 5.6 trades a day.
  - 2.10x and 1.72x.
  - $1,616 and $2,004 a day.
  - Worst day −$1,878 against −$1,000.
- **Pure walk forward fails.** Ranking cells by first-half $ a day picks the $500k gate: $1,649 a day in the first half, but $807 a day and 1.18x in the second. Do not lower the gate to $500k.
- **The plateau is flat** across gate $750k to $1M, age 1.54 to 3 h, top 0.08 to 0.12 and army 0.009 to 0.015. Every cell gives $1,766 to $2,214 a day in the second half. No knife edge.
- Exits are the no-stop set (5x, in-range half-pool pull, 12 h).
  - **Median trade 0.24x in every cell**: most trades end in a pull near 0, and the mean comes from 5x hits.
  - The 0.6x stop is shown for comparison.

## Top cells (ranked by the weaker half's $ a day at $500, cells with 20+ first-half trades)

| Gate | Age | Top | Army | Trades/day A / B | Mean x A / B | Median x | $/day at $500 A / B | $/day at $2k A / B | Worst day at $500 A / B | With 0.6x stop A / B |
|---|---|---|---|---|---|---|---|---|---|---|
| $750k | 3 h | 0.08 | 0.015 | 2.9 / 5.6 | 2.10 / 1.72 | 0.24 | 1,616 / 2,004 | 5,396 / 6,514 | −387 / −1,878 | 2.08 / 1.50 |
| $750k | 3 h | 0.10 | 0.015 | 3.0 / 5.8 | 2.06 / 1.75 | 0.24 | 1,589 / 2,171 | 5,284 / 7,090 | −387 / −1,878 | 2.05 / 1.46 |
| $750k | 3 h | 0.10 | 0.009 | 2.9 / 5.7 | 1.99 / 1.78 | 0.24 | 1,455 / 2,214 | 4,812 / 7,261 | −387 / −1,498 | 2.08 / 1.47 |
| **$1M** | **3 h** | **0.08** | **0.015** | **2.6 / 3.8** | **2.09 / 2.14** | 0.24 | **1,406 / 2,152** | 4,800 / 7,418 | 0 / −1,000 | 2.06 / 1.90 |
| $1M (today's rule) | 1.54 h | 0.08 | 0.009 | 1.9 / 3.4 | 2.20 / 2.18 | 0.23 | 1,158 / 2,027 | 3,968 / 6,999 | 0 / −1,000 | 2.09 / 1.90 |
| $500k (walk-forward pick) | 1.54 h | 0.12 | 0.009 | 3.9 / 8.8 | 1.84 / 1.18 | 0.24 | 1,649 / 807 | 5,165 / 1,312 | −761 / −2,676 | 1.41 / 1.12 |

- **Days:**
  - first half = 14 Sep to 27 Sep (14 days; the lane was quiet before 21 Sep, so first-half $ a day is understated for every cell alike);
  - second half = 28 Sep to 6 Oct (9 days).
- **Full grid:** 96 cells, ranked, http://159.223.88.168/hx-c9ois/sw/sw-a2-16-cells.csv

## Second-half $ a day at $500, no stop (plateau check)

| Gate, age | top 0.08 | top 0.10 | top 0.12 |
|---|---|---|---|
| $750k, 1.54 h | 1,837 | 2,005 | 1,919 |
| $750k, 2 h | 2,046 | 2,214 | 2,128 |
| $750k, 3 h | 2,046 | 2,214 | 2,128 |
| $1M, 1.54 h | 2,027 | 1,895 | 1,766 |
| $1M, 2 h | 2,152 | 2,020 | 1,891 |
| $1M, 3 h | 2,152 | 2,020 | 1,891 |

(army 0.009; with army 0.015 every cell moves by $170 a day or less.)

## What I would put in the paper book
1. **Main rule widened:** gate $1M, age ≤ 3 h, top ≤ 0.08, army ≤ 0.015. It beats today's rule in both halves with the same edge per trade.
2. **Second paper book `w2_v3_750k`:** gate $750k, age ≤ 3 h, top ≤ 0.10, army ≤ 0.015. It nearly doubles trades. The edge is lower in the second half, so it needs its own 14-day test before live money.
3. **Never the $500k gate.**

## How
- Strict A2.8 gates at every level: 790 gates over 300 coins, army and hubs only from earlier coins.
- Gates where the wallet data ends before the gate are left out (all levels 99% covered except $1.5M at 47%, so $1.5M is not in the table).
- Exits use in-range liquidity (A2.14) with ticks from all 300 pools (51,281 Mint/Burn events). $500 and $2k fills from pool depth, 1% fee each way.
- **Check:** today's rule reproduces 58 trades: 2.09x and 1.90x with the stop (A2.14: 2.09x / 1.90x), and 2.20x and 2.18x without it (V5: 2.20x / 2.19x).
- Scripts: container /tmp/w2/an/s16.py, /tmp/w2/a214/truel_all.py, /tmp/w2/a216/grid.py, cells.py.

## First principles score
7/10.
- **Strong:** real pool exits, both halves, plateau shown.
- **Weak:**
  - the second half is only 9 days;
  - more trades also means more of the same day's coins, so the worst day grows;
  - the no-stop exits carry pull risk (median 0.24x).

Version 1, 6 Oct 2026 15:10 Bangkok. Latest: http://159.223.88.168/hx-c9ois/sw/sw-a2-16-gate-grid.md