# A1.15 mc5 vs the 2nd window v3 rule (w2_v3_1m): overlap and combined book (FW Analyst 1, 6 Oct 2026, 15:07 Bangkok)

**Verdict: mc5 mostly buys the same winners as the 2nd window rule, just earlier (at +5 min instead of at the $1M gate). On its own, the rest of mc5 loses money.**
- 42 of the 58 w2 trades are also mc5 trades (72%).
- **mc5 on those 42 shared coins: 4.99x mean** (w2 on the same coins: 2.17x).
- **mc5 on the 159 coins w2 never traded: 0.75x mean, minus $3,950.** H1 0.69x, H2 0.77x.
- So mc5's whole profit (+$12,826) comes from buying the future w2 coins early. At +5 min you cannot tell which coins those are, so you pay for about 4 losers per shared winner.

## Numbers
- mc5: 201 trades at $100, 22 days, +$12,826. w2 (A2.14 trades csv, $500): 58 trades, +$28,635.
- **Combined, both trading: +$41,462.** 14 of 22 days green (mc5 alone 12, w2 alone 14).
  - Worst day combined: minus $1,485. mc5 alone minus $767, w2 alone minus $718.
  - Day P&L correlation 0.41, because both books win on the same coins.
- **Exposure:** the same coin is held twice in 42 cases ($100 mc5 at +5 min, then $500 w2 at the $1M gate). When that coin dies, both books lose on it together.
- Of the 159 mc5-only coins:
  - 47 still hit $1M within 2 h but failed w2's stricter gate (top share, army, age, sellable). mc5 made **1.45x** on them.
  - 112 never hit $1M: **0.46x**.
- The 16 w2 coins mc5 missed:
  - 9 had a market cap under $100k at +5 min (or no v3 data in the first 2 h).
  - 3 were under $250k.
  - 4 were $1M or more already at +5 min.

## Shared coins (mc5 x / w2 x)
| token | day | mc5 x | w2 x |
|---|---|---|---|
| 0x88042d98421b4d4014910bd9ca40d3cf3a309733 | 22 Sep | 9.52 | 4.77 |
| 0x55683371036401f8cae676c260efcde9be36a7f4 | 22 Sep | 0.08 | 4.82 |
| 0x8d5c42c096344e3ad4d5fcac8fe7a4b4ffca0150 | 23 Sep | 4.70 | 4.78 |
| 0xabe99cf268cdc8bd70a71a5283fdacbcb36714fd | 23 Sep | 14.91 | 0.56 |
| 0xa2508b15ab16826c720631b9afa0a90c453ae8a4 | 23 Sep | 6.31 | 0.03 |
| 0xfa1824e750f9a5bc7d253342935471b8f2709d70 | 23 Sep | 8.79 | 4.81 |
| 0xc3ba8a6f8d011881ceca8ad6e2b524232a40aa26 | 23 Sep | 3.32 | 0.56 |
| 0x97a5d4306a99571ecce5c2448ebfad85d7a8321d | 24 Sep | 6.38 | 4.79 |
| 0x5cee66125e28aec539365c7a34fd25a2210c3966 | 24 Sep | 0.49 | 0.57 |
| 0x364d0606e55b546ea76d038d38265116dfa36626 | 25 Sep | 0.46 | 0.55 |
| 0x0f3e825cdafbacdc552899a1185e5998f2c2ec9c | 25 Sep | 6.54 | 4.77 |
| 0xead015725c6cc0a38cdaea04cb7933210198b830 | 25 Sep | 8.09 | 5.03 |
| 0xa5a04c83d8c9228e7540978c65049a3bc41e3924 | 25 Sep | 2.24 | 0.54 |
| 0x52457611836c74812446d1b77127237ccb917238 | 26 Sep | 0.52 | 0.48 |
| 0xe99c8437b0b29325a4080b8c837ff4e5a6f9fc69 | 26 Sep | 2.61 | 0.57 |
| 0xc261532bc770c2b4ea10f97f72328aa0fc51b0d0 | 27 Sep | 5.45 | 4.77 |
| 0xce565efdc8527d54a147195a9f46ac559b46b5c2 | 27 Sep | 0.48 | 0.57 |
| 0x10897a25ded3b3f7d8cb9736fccf22c0ef5a414e | 28 Sep | 0.00 | 0.56 |
| 0xed7bc12c9c6e6484e17f7204c16a57a4e2792668 | 28 Sep | 0.41 | 0.00 |
| 0x60eafc81068bb9077fd8e5733d8b11bb2523747a | 29 Sep | 0.21 | 0.12 |
| 0x59b85970630dd8b808aad4257b0cdf02a8b22f90 | 29 Sep | 2.72 | 4.81 |
| 0x89262fb9d6b05de797c99a927994f7c2411469d4 | 29 Sep | 0.35 | 0.53 |
| 0x07a45bafa3972e242472105f66bd24c7418b280d | 30 Sep | 0.51 | 0.58 |
| 0x257723230c9cccf6a2df1e7cf67cffcfcf01eeb8 | 30 Sep | 21.08 | 4.78 |
| 0x60042b65a0c0a90ba75d17879850b6f99d9422eb | 1 Oct | 3.79 | 4.78 |
| 0xf15aab29c3d22cf2b469d4a9b0c0a72f31ba0711 | 1 Oct | 0.44 | 0.54 |
| 0xe4aadf624b3a2c44d80dc220832c9c998d39c18f | 1 Oct | 3.37 | 0.56 |
| 0xf7e98ed773dbd2bb3f2caec31e08f81b74d787c2 | 2 Oct | 0.51 | 0.56 |
| 0x6c963ce1aff02c18b2d0ef81b96df77b2f53c4ee | 2 Oct | 0.42 | 4.77 |
| 0xd2fdd92d0387f2b70680bdfccb9d685af2d94381 | 2 Oct | 0.41 | 0.55 |
| 0x50de23755e3ef741d14688a977e8c7c0800444d7 | 2 Oct | 0.48 | 0.58 |
| 0xba1a2d9783ebe2b76493c5c12eb9813a0e062843 | 3 Oct | 20.27 | 4.83 |
| 0xcb142bca82c9c852ef33c5100a6f26577ad40f3c | 3 Oct | 2.83 | 1.40 |
| 0x8cf263adfd88d72597a251908378a782a4e97aaf | 4 Oct | 3.21 | 0.57 |
| 0xfe0f309cee96912d9926b8f0786fe3347364e678 | 4 Oct | 0.60 | 0.58 |
| 0xa58ae1c2b27229aade2c3017934341dc4ecc2375 | 4 Oct | 3.14 | 0.00 |
| 0x8e41fd25e85e624560fa50ff9f117a02cd30b8b3 | 4 Oct | 15.96 | 4.80 |
| 0xff4c3278e735fa52b1012b06748407d26e047f18 | 5 Oct | 2.33 | 1.36 |
| 0xb91b4020518cd0cf824dca8291e94cd62c7f0b91 | 5 Oct | 2.93 | 4.76 |
| 0xc38c332012a9116dcade0c7f43b5d3c71799a9f8 | 6 Oct | 35.60 | 4.91 |
| 0x224a794031d54e49b46919ade2c55ff255dfdd97 | 6 Oct | 5.77 | 0.55 |

Read with care:
- The w2 x is capped by its 5x target. mc5 holds to 2 h with no target, so it keeps the big runs (35.6x, 21.1x, 20.3x).
- 0x55683371036401f8cae676c260efcde9be36a7f4 and 0xc3ba8a6f8d011881ceca8ad6e2b524232a40aa26 are A1.11 farm coins.
- The token list here is by token address. The mc5 x values are per pool (the A1.13 tables named pools).

## What this means
- **mc5 is not a separate edge.** It is an earlier, noisier entry into the same coins the 2nd window rule buys.
- **The idea to test next:** a w2-like quality check at +5 min (top holder share, army share, sellable depth) to drop the 112 coins that never reach $1M.
  - If that check can pick out at +5 min the 42 shared winners plus the 47 "hit $1M, failed gate" coins, mc5 would beat w2 per trade.
  - That needs its own walk-forward test, with nothing taken from after +5 min.
- **If both books trade live:** cap the total on one coin, or skip the w2 buy when mc5 already holds the coin. Otherwise one dead coin hits both books.

Code: a1code/v3ovl.py (research-box /root/fwa1).