I'll need to look up the card IDs and archetype info first, then run the detailed comparisons. Let me fetch everything in parallel!
Got all the IDs. Now let me run all three variation queries plus their mirror matchup queries simultaneously. The date range is ~12 weeks back from 2026-09-22 = 2026-06-29.
Let me fix the queries â the issue is likely the comment syntax. Retrying all three overall winrate queries plus the three mirror queries simultaneously:
The entries are matching but the match join is returning zero W/L/D â likely an issue with how matches link to entries. Let me investigate the matches table structure to fix the join.
Good â the results use "WIN"/"LOSS" not "W"/"L". Let me rerun all three overall winrate queries and all three mirror queries with the corrected result strings simultaneously:
Good progress! Got overall WR for all 3 variants and mirror WR for variant 1. Now I need mirror WR for variants 2 and 3:
All data collected! Here's the full breakdown:
Build definitions:
| Build | Entries | W | L | D | WR |
|---|---|---|---|---|---|
| V1 | 40 | 135 | 102 | 3 | 57.0% |
| V2 | 226 | 635 | 453 | 31 | 58.3% |
| V3 | 7 | 35 | 18 | 0 | 66.0% |
| Build | Entries | W | L | D | Mirror WR |
|---|---|---|---|---|---|
| V1 | 18 | 22 | 10 | 1 | 68.8% |
| V2 | 107 | 87 | 86 | 6 | 50.3% |
| V3 | 2 | 1 | 4 | 0 | 20.0% |
â ď¸ V3's small sample (7 entries) makes its 66% WR unreliable â likely noise. V1 (40 entries) and V2 (226 entries) are the statistically meaningful data points.
Sources: Data from 207 tournaments (2026-06-29 to 2026-09-22), 71% MTGO / 29% Melee.