# A1.24 Trade count check (Analyst 1, 7 Oct 2026, 09:40 Bangkok)

## Read this first
- **The engine matches the replay: it misses 0 of 7 gates since it started (6 Oct 13:04 Bangkok).** Same 7 coins, same receiver fails, same 3 passes.
- **The live book should have 2 trades, not 1.** Since live went on (6 Oct 19:25), 2 coins passed: CHONK (6 Oct 23:52) and CrawlScan (7 Oct 05:56). The CHONK live buy failed on a code bug (lowercase `to` address). The Coder already fixed it and logged it (sw-coder-status, "missed live fill"). It is not a filter problem.
- **Why 6 Oct looks like 1 trade:** the engine only started at 12:58 Bangkok. Three replay passes came before that (HOODCHAN 06:47, STONKROBBERS 12:42, SS 06:45 on the receiver edge). CLAUS (13:51) passed, but the w2_v3_wide_s03 book did not exist yet; only the 1m and hold books traded it. With those, 6 Oct had 4 to 5 passes, a normal day.
- **7 Oct so far is just quiet:** 2 gates in about 9 h (MISO failed receiver 20%, CrawlScan passed). At about 9 gates a day you expect 3 to 4 by now, so 2 happens about 1 time in 3.
- **The market did not change and the filter did not change:** gates per day are flat at 5 to 12 (about 9). Receiver fails are about 1 in 3 gates, the same as before. Age (3 h) and duplicate-token never fail.

| Since engine start (6 Oct 13:04 Bangkok) | Replay | Engine |
|---|---|---|
| CLAUS `0x4d8b2411255580b19a0b9fc90ed36ee15e7e1af5` 13:51 | pass (army not replayed) | PASS (wide book not live yet) |
| CLAUS `0x4fdcc31f5514644bf512b296057f025ddd4f9308` 16:48 | pass ex army | fail army 29.4% |
| CLAUS `0x206211d456a3a9dbe45e13507aac5131ace9c716` 19:07 | pass ex army | fail army 9.5% |
| CORES `0x2cdf271007c7ebb6013d55492e156031c535094f` 19:48 | fail receiver 25% | fail receiver 25.0% |
| CHONK `0x8da021f34c2f2a95696c4fd842b0cd0cc4e18468` 23:52 | pass | PASS, live buy error |
| MISO `0x587129b682c1ff9aa01c0f0faff28e5bbe50172c` 04:33 | fail receiver 20% | fail receiver 20.0% |
| CrawlScan `0xa6caf95634d4bcffdfb535eec889c9c928ff13d5` 05:56 | pass | PASS, live trade |

- The engine also flagged one gate the replay does not see: SI `0x6c0f032f19a121b1767fdc71ae35d4db0084e999` (6 Oct 11:25, before the engine started, never scored).

## Replay passes before the engine (5 Oct 00:00 to 6 Oct 12:58), engine never saw them as live gates
| Gate (Bangkok) | Pool | Age min | Receiver | Engine |
|---|---|---|---|---|
| 5 Oct 00:06 | `0x2a878208df14eecf7761daa73f4c1c070cece119` | 33 | 3.9% | not running |
| 5 Oct 04:13 | `0xba2ca7b9cf44cd44fc8dbd60e2e7ed88b8587d83` | 4 | 4.2% | not running |
| 5 Oct 06:25 | `0x0f6741982785a62e850b611cd012eb798b738037` | 37 | 1.3% | not running |
| 5 Oct 06:48 | `0x5560b76440e9b6660ab6ca8c7ae1941c92abffcf` | 43 | 0.9% | not running |
| 5 Oct 07:53 | `0x76a4c5b933c29c859fab6b3f38b540dbe7ac2aec` | 97 | 0.3% | not running |
| 5 Oct 22:48 | `0x5078092c196727b4798c5c4d583d69146bec3929` | 52 | 1.2% | not running |
| 6 Oct 06:45 SS | `0x3939ce55ec4e0ff8df9aa6469a795c793da06d55` | 1 | 8.3% (edge) | not running |
| 6 Oct 06:47 HOODCHAN | `0xb4b0dfeaa2e2a838f0f7a304f005ebe32390d4ef` | 24 | 1.8% | not running |
| 6 Oct 12:42 STONKROBBERS | `0xdcc026752d7cbf4fe2a14d53986bd3d22de72024` | 46 | 0.7% | not running (started 12:58) |

## $1M gates per day and why they failed (replay, Bangkok days)
| Day | Gates | Fail receiver | Fail age | Fail dup token | Pass before army | Engine army fails | A2.21 trades |
|---|---|---|---|---|---|---|---|
| 29 Sep | 11 | 3 | 0 | 0 | 8 | n/a | 6 |
| 30 Sep | 10 | 5 | 0 | 0 | 5 | n/a | 4 |
| 1 Oct | 10 | 4 | 0 | 0 | 6 | n/a | 4 |
| 2 Oct | 12 | 6 | 0 | 0 | 6 | n/a | 4 |
| 3 Oct | 5 | 0 | 0 | 0 | 5 | n/a | 2 |
| 4 Oct | 12 | 1 | 0 | 0 | 11 | n/a | 5 |
| 5 Oct | 6 | 0 | 0 | 0 | 6 | n/a | 5 |
| 6 Oct | 8 | 2 | 0 | 0 | 6 | 2 (of 4 scored) | 2 (replay data ends 06:47) |
| 7 Oct (to 09:17) | 2 | 1 | 0 | 0 | 1 | 0 | |

- Army is the other big filter: before army, about 6 to 7 coins a day pass; after army and the sell check, A2.21 kept about 4.5 a day. Army needs the engine's wallet memory, so I can only show engine army results (6 Oct on).
- CSV: http://159.223.88.168/hx-c9ois/sw/a1-24-per-day.csv . Engine gate dump: http://159.223.88.168/hx-c9ois/sw/a1-24-engine-gates.csv

## Method
- Every Uniswap v3 WETH pool created from 28 Sep 21:00 Bangkok to now (2,726 pools created from 29 Sep 00:00 Bangkok, plus older pools for context; factory `0x1f7d7550b1b028f7571e69a784071f0205fd2efa`), every Swap for 3 h after the first swap.
- Gate = first buy that brings the USD bought at a $1M+ cap to $20k (cap from the pool price, live supply, hourly ETH close; times from real block timestamps). Age = minutes from first swap, max 180. Receiver = top Swap recipient share by count of buys up to the gate (the engine's own definition). Dup = token already gated in another pool.
- Engine side: w2.db `pools` (gate seen) and `coins` (scored), plus w2.log. The engine bootstrapped at 12:58 and started its cursor at 13:04 Bangkok on 6 Oct; every gate it saw after that has a score row in `coins`, so restarts skipped none.
- Check on the replay: it finds 56 of the 58 earlier gate-set coins since 29 Sep.

## Limits
- No army and no sell check in the replay. Cap uses the hourly ETH close and the latest supply, so a coin right at $1M or at 8% receiver can flip (SS is the edge case).

## Asks for Supervisor
- No filter change. The CHONK live miss is the only real loss, and it is already fixed.
- Expect about 4 to 5 trades a day on average, with quiet days of 1 to 2 (6 Oct had a 1-in-16 low by count, but that was the start-up gap).

First principles score: 90%. Every gate comes from the pools' own Swaps, and the engine side is its own database and log. Missing: army and sell check in the replay.

Scripts: /root/a1/a124/rep.py (review container). Latest: http://159.223.88.168/hx-c9ois/sw/a1-24-trade-count.md