Join DiscordWhat wins games
Win rates and matchups from Gundam Card Game games logged with the match tracker. The tier list next door ranks archetypes by tournament results — placements can never produce a win rate, and these games can never tell you what took down an event. Read them together.
Archetypes
Measured over the last 7 days. The upper bar is share of the field — what tracked players ran into, which samples everyone playing the game. The lower bar is share of what tracked players themselves brought. Win rate is over every game the archetype was piloted in, with its 95% interval.
| Archetype | Field / piloted % | Win rate | Going first | Going second | Games | Length |
|---|
4 more archetypes appeared fewer than 15 times and are not listed.
How these numbers are made
Where the data comes from. Every figure is aggregated from games players logged with the tracker extension. Nothing on this page names a player, a deck or a match: a rate is only published once at least 5 different players contributed to it, so no number here can be read as one person’s record.
Two sides, two qualities. A tracked game tells us the reporting player’s deck exactly — the capture carries the list they queued. The opponent’s deck is inferred from the cards they revealed in play, because a capture never sees a hand or a library. In this window we placed a deck on the other side of 1% of games.
Thresholds. An archetype needs 15 appearances for a row and 30 decided games for a win rate; a matchup cell needs 12. Below those we show the game count and withhold the rate rather than print a percentage that a couple of games could flip. Every rate carries a 95% Wilson interval — a 60% over 30 games and a 60% over 3,000 are not the same claim.
Draws and mirrors. Draws count as games but are excluded from win rates, which are wins ÷ (wins + losses). A mirror is even by definition, so the diagonal reports games played and no rate.
Counted once. When both players in a game run the tracker we receive it twice, once from each seat. Those pairs are collapsed before anything is counted, so a game is never both a win and a loss in the same total. Matchup cells then fill in both directions from that single game, which is the same fact read from the other chair — that is why a row and its transpose always agree.
Known biases. Players who install a tracker are not a random sample of the ladder, so absolute win rates likely run above the general population; matchup rates, which compare two decks inside the same sample, are more robust. Opponent identification also correlates with game length — a game that ends before anyone commits a board reads as unidentified — so blowouts are under-represented in the matrix, though not in the archetype win rates, which count every game whether the opponent was read or not. Preview queues on unreleased cards are excluded entirely.