How many games should you play to test an Azuki TCG deck change?

Change one thing at a time, and judge it on what you saw rather than the record. Ten games show whether a new card gets played and does its job, but a deck that wins half its games can go 8-2 or 2-8 over ten by luck alone. A 5-point difference takes hundreds of games to see.

One time I went 7-3 and thought I was a genius. Then I went 3-7 with the SAME deck. Luck is so sneaky.

How to tell

  • You change five cards at once, then cannot tell which one helped.
  • A 6-4 week convinces you a swap worked.

Example: How far luck moves a 50% deck

A deck that truly wins half its games still lands anywhere in these ranges about 95% of the time:

GamesWins (95% of the time)Win rate
102 to 820 to 80%
206 to 1430 to 70%
5018 to 3236 to 64%
10040 to 6040 to 60%

So test the things ten games can show: does the new card get played, does it come out on time, and does the loss you were fixing happen less.

How to fix it

  1. Change one card pair at a time: two copies out, two in.
  2. Play the new list against the same decks as the old one. In Practice you can pick the opponent’s deck.
  3. Watch the job, not the record: is the new card played, and does your most common loss happen less?
  4. Keep a change that clearly does its job. Revert one you never get to play.

In other card games

Questions

How many games does it take to know if a deck is good?

About 100 games pin a win rate down to roughly 10 points either way. Ten games only tell a very strong deck from a very weak one.

Should you change several cards at once?

Only when you rebuild on purpose. Change one pair at a time and you know which swap did what.

More deck-building guides

Beta