Let me pull the raw match data for that head-to-head right away.
Good, both archetypes are in Pauper. Now let me query the raw match data directly.
Your suspicion is correct. There IS double counting happening. Here is the full breakdown and the explanation:
The raw query returns 49 rows, but a significant number of them are duplicates of the same physical match, appearing twice with different timestamps on the same tournament. Here is every row, grouped by event:
--- DOUBLE-COUNTED MATCHES (same match, two rows) ---
These tournaments each appear with two different timestamps (UTC vs local), causing every match within them to be stored twice:
Pauper Challenge 32 - 2026-02-15 / 2026-02-16 (link: https://www.mtgo.com/decklist/pauper-challenge-32-2026-02-1512833052)
Mont Weekly Pauper - 2026-02-18 (two timestamps: 00:00 and 07:00, link: https://melee.gg/Decklist/View/1dbf1304-a9df-44cc-83b5-b3f401570468)
Pauper Challenge 32 - 2026-02-19 / 2026-02-20 (link: https://www.mtgo.com/decklist/pauper-challenge-32-2026-02-1912833518)
Pauper Challenge 32 - 2026-02-20 (two timestamps: 15:00 and 22:00, link: https://www.mtgo.com/decklist/pauper-challenge-32-2026-02-2012833528)
Pauper Challenge 32 - 2026-02-21 (two timestamps: 11:00 and 18:00, link: https://www.mtgo.com/decklist/pauper-challenge-32-2026-02-2112833543)
Pauper RC Qualifier - 2026-02-22 (two timestamps: 16:05 and 23:05, link: https://www.mtgo.com/decklist/pauper-rc-qualifier-2026-02-2212833504)
Pauper Challenge 32 - 2026-02-22 / 2026-02-23 (link: https://www.mtgo.com/decklist/pauper-challenge-32-2026-02-2212833562)
--- UNIQUE (NON-DUPLICATED) MATCHES ---
The remaining 25 rows are genuine unique matches:
Wait, that comes to 23 unique matches in the non-duplicated set. Let me recount the de-duplicated total: 23 unique matches from the unduped group plus the 13 unique matches that were double-counted = 36 unique matches total. Of those 36: Red Madness wins 14, losses 22. So the de-duplicated record would be closer to 14-22, not 22-25.
To summarize the data quality issue: 7 tournaments in this window have the same tournament stored under two different timestamps (most likely a UTC vs. local time ingestion bug), causing each match within those events to be recorded twice. The matchup tool is reading 49 rows and then deduplicating some internally to arrive at 47, but it is not catching all the duplicates because the pair_id deduplication likely does not account for the timestamp split. This is a legitimate double-counting bug worth filing.
Sources: Data from 49 raw match rows across 13 tournaments (MTGO Challenges, Melee.gg leagues, Cardsrealm events) from 2026-02-09 to 2026-03-10. Sample links: https://www.mtgo.com/decklist/pauper-challenge-32-2026-02-1512833052 | https://melee.gg/Decklist/View/1dbf1304-a9df-44cc-83b5-b3f401570468 | https://www.mtgo.com/decklist/pauper-challenge-32-2026-02-2112833543