Best-of-One vs. Best-of-Three

Magic Arena uses a best-of-one model for its ranked games, and there’s been a heated debate between the people who believe that this is the natural course for Magic to take and the people who believe that best-of-one is too big a distortion, to the point where Magic starts to look like a different game.

Today, I’m going to talk about the pros and cons of the best-of-one model, whether I believe it’s the future of Magic or just an Arena quirk, and what you as a player can do differently if you’re playing best-of-one.

Here are the factors we should examine:

The Convenience

Simply put, playing best-of-one is much more convenient than playing best-of-three. It takes less time—potentially a lot less time. If I have 20 minutes, I can fire up Magic Arena and play a best-of-one, but I can’t do that in a best-of-three because there’s a high chance it’ll take more than 20 minutes. This is a clear win for best-of-one.

The Variance

The biggest gripe about best-of-one as opposed to best-of-three is the variance. In best-of-three, the better player will win substantially more than in a best-of-one, regardless of what else changes. We can observe this in team tournaments, which turn the game into a “best-of-three best-of-threes” and catapult the best players’ win-rates by over 10%.

That said, I don’t think this is a very big issue, because you can just play more games. Obviously it’s not a precise equivalency, but if you imagine that you can play 8 best-of-threes in one day or 24 best-of-ones, then it doesn’t look so random anymore. So if you’re comparing 1 best-of-one with 1 best-of-three, then yeah, it’s much worse for the better player, but that’s not the comparison—you should compare it with 2.5 best-of-ones, which is much closer (even though it’s not identical).

In some ways, this actually diminishes variance because it lowers the impact of a matchup. Let’s extrapolate this to, say, a best-of-seven. Imagine that instead of seven best-of-threes, you play three best-of-sevens. What if you hit a horrible matchup? Then, even if you’re much better than the competition, you’ll still very likely lose. And since you’re only playing three matches, a loss hurts way, way more. So, all things equal, I would rather play against twenty different opponents than three different opponents, because then I’m much more likely to play against a varied representation of decks.

The main issue with this approach, of course, is that each “change of opponent” in a tournament takes time. Playing three games against the same opponent takes less time than playing one game against three different opponents (and that’s not even considering the fact that game 1s in control matchups take much longer than games 2 and 3 as a general rule). Once you have to make everyone wait for everybody, like in a tournament setting, then there’s an even bigger problem. You wouldn’t be able to simply replace a GP Day 1 of 8 best-of-threes with 24 best-of-ones—that would take way longer. I don’t know what the actual number of best-of-ones that you can play is, and I don’t know whether it’s enough to offset the variance increase for best-of-ones. If 8 best-of-threes become 15 best-of-ones, is that enough? I don’t know.

For ladder, though, I think it’s good enough. Once you don’t have to wait for everybody, and delays aren’t that costly, then I don’t think it’s a problem to simply replace the 5 best-of-threes you’d play with, say, 12 best-of-ones.


The loss of a sideboard in best-of-one games is the biggest difference, and impacts the game in two main ways:

First, what if someone plays a deck like Dredge? I don’t want to have to main deck Leyline of the Void. There are many strategies in Modern that are only kept in check because of strong sideboard cards (Stony Silence, Rest in Peace, etc.), and what happens to these decks once we turn to best-of-one? There are no decks in Standard right now that are exactly like this, but can we guarantee that it won’t?

A while ago, we had a U/W Approach of the Second Sun deck that won almost all game 1s and was kept in check because people could board in Duress and Negate. Are we going to have to main deck this type of card in a best-of-one?

As a general rule, sideboards are equalizers—I improve my bad matchups, and my good matchups improve against me. There are some matchups that are almost impossible to win in game 1 that become even or favorable post-board, but in a scenario where sideboards don’t exist, then you’re much more at the mercy of the pairings lottery, which is not a good thing.

The second, not so obvious way, is that sideboards are an avenue for removing bad cards, and in a best-of-one this avenue is gone. Imagine a situation where I do have to main deck Duress or Negate—that doesn’t seem so bad after all, as there have even been decks that main decked these cards before.

But now imagine the same situation where I cannot side it out ever, and that seems a lot worse. Can you afford to main deck Duress if there are decks in the format it’ll do nothing against and you don’t have the option of siding it out? Can you even afford to main deck a card like Lava Coil or Seal Away?

Assume that you do main deck Duress. Then, that’s going to be extremely swingy for your win rate. Get paired versus a deck Duress is great against? You’re way more likely to win. Get paired versus a deck Duress is bad against? You’re way more likely to lose. And there’s not much you can do about that, other than make the decision of whether or not to put Duress in your deck. If you’re the opponent of said person, then you didn’t even make that decision yourself.

In this regard, control decks are ahead if the landscape of decks remains similar to what we have today. Control decks have a lot of potentially dead cards game 1, but they are always against decks that also have dead cards against them. So, if I’m control and my Seal Aways and Deafening Clarions are dead, then that very likely means that my opponent is also playing Seal Away and Deafening Clarion (because, if not, what are they playing? They don’t have any creatures!).

If I’m playing an aggro deck with some removal, however, then I do not get this symmetry—my removal is dead, but my opponent’s removal won’t be. This creates a scenario where control decks are always ahead or even in this exchange. Plus, control decks tend to draw extra cards and want a prolonged game, so any scenario where both players have cards that do nothing will favor them versus, say, a midrange deck. As far as aggressive decks go, the more proactive the better—think White Weenie as opposed to Big Red—because then you have very little or no dead cards in any matchups. Basically, in a best-of-one, putting something like Lava Coil in your non-control deck might be a disaster.

This is without introducing the variable of combo decks, which are the real winners. A combo deck can get to a spot where it has no dead cards against anyone, and everyone has dead cards against it. The game 1 win-rate of these decks is usually through the roof. Right now we don’t have a deck like this in Standard, but if such a deck comes to exist, then it will warp the format like nothing we’ve ever seen.

There are some solutions for this, particularly if we see flexible cards in the future. WotC has already confirmed that they designed the split cards with best-of-one in mind, and it’s possible other cards followed the same process—flexible answers like Knight of Autumn, Bedevil, Mortify, and Ravager Wurm. These cards are good in a best-of-one environment because they allow people to interact with unconventional card types without the risk of having to put a useless card in their decks. If you can play Bedevil instead of Doom Blade, then you get to destroy your opponent’s Teferi or Treasure Map, rather than have a card that does literally nothing.

If best-of-one is going to be big, then that is a trend that will need to continue. In fact, I like this trend of flexible answers whether we’re in a best-of-one or best-of-three scenario—having the ability to play a card like Knight of Autumn is just good for Magic (up to a point where it gets excessive, of course—the game isn’t fun if everyone is playing ten Vindicates).

More than just having flexible answers, however, they have to be careful not to create too many lopsided matchups. Obviously it’s unavoidable to have some very lopsided matchups—someone is always going to be playing a Turbo Fog deck—but you have to aim for a scenario where good matchups are overall 60/40 and bad matchups 40/60. If too many matchups are 80/20 and 20/80, and there is no sideboarding to fix it, then the pairings will decide everything. Basically, without sideboarding, there’s less room for error on R&D’s part.

The Knowledge Disparity

In best-of-one, you never know what you’re playing against. This is a difference that I haven’t seen mentioned anywhere so far, but that I believe to be quite substantial. In a best-of-three, you always know what your opponent is playing for at least half the games that you play, but in best-of-one you never know. This means that you’re going to mulligan blindly 100% of the time, rather than, say, 40% of the time. This is huge, because some decks are much better at mulliganing blindly than others.

Take, for example, White Weenie. For the most part, whether you mulligan or not has to do exclusively with the composition of your hand—do you have enough lands and enough of a curve? It doesn’t matter what the opponent is playing. Now, take a deck like Jeskai Control. Imagine an opening hand like this on the draw:

This is a hand you’d be happy to keep against a control opponent, but that you probably can’t afford to keep against an aggro opponent. In a best-of-three, you know whether this hand is good or not over half the time. In best-of-one, you simply never know.

The same is true for cards you’re scrying after you mulligan. If you have a proactive deck, then it’s easy to know if you want the top card or not. If your deck is reactive, it’s often hard to know if you don’t know what you’re playing against.

For example, at a GP a while ago, I was playing a U/B Midrange deck. I mulliganed and scryed into my one copy of Hostage Taker. I wasn’t sure whether it was going to be a good card or not, so I sent it to the bottom. It turned out that my opponent was playing a God-Pharaoh’s Gift deck, and if I had kept the Hostage Taker on top, I would easily have won. So, all other things equal, I would have won that game if it had been a game 2, but not a game 1. Control decks often have cards like this (Deafening Clarion, for example), whereas aggro decks generally do not.

The Multiple Decks

In a best-of-three, you bring one deck and you always play it. In a tournament setting where games are best-of-one, you could, in theory, bring more than one deck to a match, which might make things more interesting. I believe that this is basically impossible to do in paper for anything that is not a very small, specialized tournament, as telling someone to bring three different decks to a tournament is a very tough ask, but it can be done online and it is, in fact, the model that most online card games adopt (and the model the Player of the Year Playoff adopted).

This can be a good thing (if you run into a bad matchup then you will just get to play more matchups), but it can also be weird because I don’t believe we’ve found the best format for matches like this yet. If you do Conquest, for example (which is the format where each deck is retired once it wins, so every deck in a lineup has to win at least one game), you start running into concerns with exploitable strategies that aren’t fun for anyone. For example, someone could bring all decks that cannot ever lose to one specific deck and can’t beat anything else, and then if the opponent has that specific deck in the lineup, then they can’t win. If you do Last Hero Standing (which is the format where each deck stays until it loses), then matchup order plays a huge part in deciding the winner (because I could have a deck that’s very good against all but one of your decks, and then randomly get paired versus that one, in which case my excellent deck wouldn’t win a game). Both systems exist and can work, but they both have some flaws, so it’d be good to find something better.

It also gets awkward that, in Magic, there are no deckbuilding restrictions, so it’s hard to tell people to bring “four different decks.” What’s a different deck? In Hearthstone, it’s quite clear—a Druid deck is different from a Paladin deck. In Magic, you can’t have a “red deck” and a “blue deck” since colors blend. Standard is too narrow for big limitations (when I played the World Magic Cup, nearly every team was forced to bring the same decks since there was only so much mix-matching you could do without overlap, which I think made the tournament pretty unexciting as every round had the same lineup of three decks in a random order), but if your limitations are too small, then you run the risk of people bringing in the same deck with a small change (for example, you could bring mono-white, mono-white splash red, mono-white splash green, and so on). We have to find a good method of defining “different decks” in Magic and we haven’t done that yet.


Limited reacts a little differently than Constructed to best-of-one. The first major difference is that sideboard cards are now unplayable. In my first MTG Arena Drafts, I was still taking sideboard cards early, until I realized how stupid that was. It’s better to take a card that is 5% to make your main deck over a very strong sideboard hoser.

If we’re drafting with bots, then best-of-one is a little worse than best-of-three because I quite like sideboarding in Limited. But if we’re drafting in a pod, it might be better because instead of playing against three opponents, you can play a best-of-one against all seven. This should make things more interesting, since you get to play against a wide range of things, and also more fair in the sense that now you can’t hope to dodge the best deck or spike getting paired versus the two worst decks, and so on.

In Sealed, however, I think losing sideboards is big. I often transformed my deck entirely in Sealed, and I thought it was one of the most challenging and rewarding parts of Sealed tournaments. Having the ability to board into multiple 3/1s against a deck with 5/3s, or board them all out against a deck with pings, or bringing in two Cancels and two Mind Rots against a deck with bombs—those were all changes that let better players beat better decks in Sealed, and in best-of-one they don’t exist. This is an entire part of the game that vanishes in best-of-one.

Avoid playing sideboard cards in your main deck just because it’s best-of-one. Whenever I stream, I always have people telling me that I should play Crashing Canopy or Cleansing Ray, specifically because it is best-of-one, which doesn’t make much sense to me. If I thought these cards improved my best-of-one win rate, I would main deck them in best-of-three. Playing best-of-one isn’t an excuse to add a narrow card to your deck.

So, what’s the conclusion from all of this?

Personally, I believe best-of-one is currently better than best-of-three from a ladder perspective. You have to commit less time and you get to play against more different decks, which are two huge pluses. For tournament Magic, at this point in time, I don’t think we’re ready for best-of-one yet. Sideboarding is too important a mechanic to balance the game as the cards were designed. But we could move in a direction in the near future where best-of-one is good enough for a tournament, especially if it’s an online tournament where the pause between rounds can be shorter (since you have to play a lot more rounds). This would require conscious decisions from R&D to balance the game for best-of-one, and while we might be seeing evidence of that in some of the latest designs, we still need to have a little more of it. If that is not the case, then you run the risk of having games decided exclusively by a matchup lottery (though playing more rounds will certainly help that).

It’s possible that, in the future, we end up in a world where ladder is best-of-one and online tournaments are best of, say, seven, with different decks (and the games are best-of-one). For this to be the case, though, we have to figure out a good format for the best-of-seven and good rules for defining what counts as a “different deck.” Right now I don’t think we have the former and I’m pretty sure we don’t have the latter.

I also believe that formats like Modern and Legacy will be impossible to balance in a best-of-one unless they change a lot of things, so it’s probably going to be a Standard-only format, but we could in the future have a “newer Eternal” format (Ixalan on, for example) that can be good for a best-of-one.

For Limited, the ability to play versus all the opponents in a pod is very interesting, and I wouldn’t be opposed to experimenting with best-of-one there. I am disappointed to lose sideboarding in Sealed, though.

If you’re currently playing best-of-one as opposed to a best-of-three, then there are upsides and downsides to both styles of decks. Aggro decks get better because they are proactive and don’t care what the opponent is playing, and control decks get better because they get to blank the opponent’s cards and they have no way of getting rid of them in game 2s and 3s. The best decks to play are probably a powerful combo deck or an aggro deck that doesn’t run potentially dead cards (like White Weenie).

It’s a good direction to play more cards that demand more specific, narrow answers. Take Rekindling Phoenix. The best answers to Rekindling Phoenix are Lava Coil and Seal Away. In a best-of-three, I can reasonably see myself maindecking Lava Coil. In a best-of-one, where I can never side it out, I’m likely to look for a more flexible alternative. My G/R decks might play Thrash // Threat over Lava Coil, and my big red deck might play Lightning Strike. If you can identify what type of card requires an answer that is less likely to be played, then you can profit from best-of-one. I don’t know exactly what those cards are right now, but I’m going to try to find out. 

1 thought on “Best-of-One vs. Best-of-Three”

  1. Pingback: Beating Aggro in Best-of-One

Comments are closed.

Scroll to Top