Ground Rules

The objective of this tournament is to find the “best” creature in Magic: The Gathering.

This will be achieved by running a weekly single-elimination tournament between all the (non-reprint) creatures in each set, to determine that set’s champion. Each champion will then play each other in a single-elimination finals to determine the best creature in Magic.

Rules of a Game

A game between two creatures will work as follows:

1. Each creature will be represented by a 60-card deck. The 60-card deck will contain only basic lands and the creature card in question (the maximum 4-of rule does not apply here, nor does the Legendary rule).
2. A deck can choose which cards – i.e. which combination of basic lands and creature cards – it draws in its opening hand. It can choose all of its subsequent draws as well.
3. Besides that, standard rules apply. The first deck to get the other deck down to 0 life (or decking, or some alternate win condition) wins the game.
4. For each game, the higher-seeded creature (i.e. the creature card with the higher price in Scryfall) will be on the play.
5. The winner of the game advances to the next round. The loser is eliminated.

The most important aspect of all this is that we are looking for a deterministic outcome; i.e. one that does not involve any randomness. The winner of each game should, in theory, be the creature that wins if both sides play “perfectly”.

Thus, we’ll need to impose the additional assumption that each deck has full knowledge of the opponent’s deck and optimal game plan, and will make their decisions accordingly.

Questions

What if there’s a stalemate and nobody can attack profitably?

As I mentioned, there are ways to win in Magic besides doing damage. Like decking the opponent.

And so, if a stalemate does occur, then the deck that’s on the play (the higher seed) will win by default, by virtue of drawing one less card. Absent any kind of card-draw shenanigans, of course.

But wait! Between that and being a turn behind in combat, isn’t this format unfair for the lower-seeded creatures?

Perhaps. Though I should note that, just like in regular Magic, drawing that extra card is also a meaningful advantage in itself. Probably doesn’t make up for the two disadvantages above, but we’ll have to see.

In any case, think of the higher-seeded creature as having a “home-field advantage”, just like in pro sports. And those higher seeds earned that advantage by being the more expensive card. (I’m assuming here that price is a rough indicator of the quality of the card, as judged by the entire Magic community.)

What about those games where the outcome depends on how well the respective decks play? (when to activate abilities, when to attack, etc.)

The beauty of this exercise is that it doesn’t matter how “well” the two decks play; because in theory they are playing optimally.

In most cases, the optimal action will be very obvious. Sometimes, though, it does get more complex. In those cases, I’ll put a bit more of an explanation for how I arrived at the outcome I did.

And, of course, there are those rare instances where one (or both) creatures have so many decision points that the game’s outcome is close to “unsolvable”. In those cases, I’ll try my best to hand-wave it to a somewhat logical conclusion, while freely admitting that it very well may not be the correct answer. I’ll put a special note for those games, and leave it to a future machine learning algorithm for finding a more rigorous result.

Why did Card A beat Card B? Card B is just so much better.

Yes, despite me promoting this as finding the “best” creature in Magic, I’m fully aware that real Magic does not work that way. In reality, the cards are played with other cards, which completely changes the paradigm.

Unfortunately, to come to a deterministic outcome, this is the simplification we have to make. Which means that a lot of top-tier tournament cards may see disappointingly early exits; because their totally broken abilities are near-useless when every other card in your deck is that exact same card. At the same time, some random creature that nobody’s ever played with seriously could very well rise to the top.

But that’s all part of the fun here.

What about cards that have random effects, like Hypnotic Specter or Crystalline Giant?

I will have to concede some randomness for cards like that; and let them play as is. Meaning that the results won’t be completely deterministic. But neither are the Scryfall seedings (which fluctuate day by day), so it is what it is.

Why are you doing this?

Short answer: because it’s fun.

More specifically, I have a strong affinity for three things in this world: (1) Magic: The Gathering; (2) competitive sports; and (3) math, economics, statistics, logic, and all that nerdy stuff.

So when the idea came to do some sort of March Madness (college basketball) style tournament between cards in my favorite game, running each match by essentially solving a math problem…I could not pass it up.

And, who knows, maybe this abstract set-up can be the first step to developing a fully functional Magic: The Gathering AI!