The most common decision I see Magic’s top players differing on is when to mulligan. You have top professionals like Martin Juza who look for any reason to mulligan a hand, top professionals like LSV who are more on the side of looking for reasons to keep a hand, and even professionals such as Gerard Fabiano who is rumored to have kept both a 7 land hand and a 6 land and Praetor’s Counsel hand in GP Paris this year, though I didn’t see it with my own eyes so I refuse to believe it.
Usually when I am discussing an in game play, after I explain the reasoning behind my lines and what I was trying to accomplish, whoever I am discussing it with will be explain how I gave up too much against other possibilities or how my play was good and they didn’t realize what I was trying to beat so they questioned the play. However with mulligan decisions it is almost unresolvable. This is because with the whole game yet to unfold the math behind the play is near impossible to do and or to argue convincingly.
Since we can’t convincingly do this math and find the right answer, I am going to discuss my general thought process on whether or not to mulligan a hand. The first thing I want to think about is whether or not my deck mulligans well as a whole. While obviously you would always rather start with 7 cards than 6, 6 than 5, etc etc, some decks’ win percentage drops much less with each mulligan than other decks. Generally decks that can recover card advantage through cheap to cast spells, win conditions, and combos mulligan very well. Some great examples of these types of decks would be Faeries, Thepths, and current Caw-go. Generally decks that rely on a lot of 1 for 1’s and win the game through some difficult to cast but powerful spell do not mulligan well. Some great examples of these types of decks would be 5cc or type 2 Valakut.
The next thing I think for in close hands is whether or not there are cards that are key to the matchup or bad in the matchup. Having a [card]Day of Judgment[/card] in your opening hand in the Caw-go mirror is already close to a mulligan, so if my opening hand was not spectacular but had a Day in it I would be considerably much more likely to mulligan it than if you switched the Day to a card that was useful in the matchup though not a key card ([card]Stoneforge Mystic[/card], [card]Squadron Hawk[/card], [card]Preordain[/card] to find Mystic and Hawk).
For example on the play in a game 1 in the Cawgo mirror I would mulligan:
Day of judgment
but I would not mulligan the same hand with Gideon instead of Day of Judgment.
I consider both to be close hands and I am not particularly excited about the second, but the cards I am looking for to keep a hand are [card]Stoneforge Mystic[/card], [card]Squadron Hawk[/card] and [card]Preordain[/card] in that order. However the first hand is still bad enough to mulligan even though you have Hawk because you also have day, which is virtually a dead card. The second hand is just good enough to keep because you know all 3 of your spells will be useful.
With that knowledge in mind I am going to try and figure out whether keeping or muliganing will result in a higher win percentage. For example one interesting mulligan decision that came up for me when testing for Top 8, where I was about to play the Cawgo mirror, was a hand like:
Jace, the mind sculptor
This hand might seem fine but that’s all it is. The fact of the matter is the player on the play in the straight UW Caw-go mirror is a big favorite because if you just hold serve you win. If you play a turn 2 [card]Stoneforge Mystic[/card], then they have to answer with one, or they almost always lose. If they answer with one, then if you play a Hawk they again have to answer with one or they almost always lose and so on. This hand is completely reactive which means they don’t have to answer anything. On the other hand you do have the ability to counter their 2 drop and force a long game. While you need to be on the play or you wouldn’t be able to counter their 2 drop there are inherent problems with the strategies behind this hand.
For example if the game is going to go late, which it almost assuredly is, the player on the draw has an advantage instead of the player on the play. With no form of pressing tempo or racing the player on the draw has the advantage of having an extra card in his hand on his turns to help him make land drops. This will cut into your win percentage with this hand. In Pro Tour Paris I played the UW cawgo mirror 3 times in the swiss and once in the Top 8 for a total of 13 games. The player who played first went 12-1, with the only game being won by a player on the draw being game 2 when I played LSV and he took a mulligan then got extremely flooded, with his only business being a [card]Stoneforge Mystic[/card] which I had a [card]Divine Offering[/card] for. Obviously the player on the play is not 92 percent but I do believe he has a large advantage, somewhere between 66 and 75 percent. I feel like that particular hand is right around 50/50 to win the game. You have no real advantage that you are pressing. You can stop most combinations of their early game, however in the late game you have no Colonnades or Tectonic Edges, but will probably draw into some by the time you get there, meaning you are probably at a slight disadvantage to start. Of course you can still draw into a Stoneforge Mystic over the next couple of turns and there are other things that could happen to result in you winning the game, so I think you would win the game between 45-50 percent of the time. If you mulligan I believe a random 6 on the play in a game 1 of the UW Caw-go mirror will win the game between 55-60 percent of the time, down from the total percentage a keepable 7 would win.
Anyways the math behind this is far too difficult to compute accurately and if you disagree with my assessment of the percentages, that’s of no importance. The idea is to understand the thought process of deciding whether or not to keep or mulligan a close hand. Since the goal is to give yourself the highest chance to win the game, this is the thought process you should be following.
This assumes a lot of information; of course a lot of the time you don’t know what your opponent is playing, or it’s Limited and you don’t know anything about their deck or you played a short game 1 where you saw 6 of their presumed 22-23 spells. I can’t really give you rules as what to mulligan and what not to because it’s always going to depend on your deck. Even in Limited you should try and think about whether your deck needs to come out well or whether or not it has hard to cast spells that can win you the game. For example Martin Juza, who should be considered the all-around best limited player in the world(not close), completely dominated in Zendikar and Scars Limited. Both of these formats are formats where aggressive mulliganing is correct because it is extremely important that your deck comes out well or your hands have synergy. This is due to the aggressive nature of Zendikar and the Comboish nature of Scars decks.
However in formats like Rise of Eldrazi, where the games are slow and sloppy and you do not need to come out well, it was correct to keep virtually any hand that would make land drops and just trust your deck to cough up win conditions. I would happily take Martin on my team in any Limited format but I don’t think it’s coincidence that the pro I know who mulligans the most aggressively dominated in formats that reward that and didn’t do quite as well in a format that doesn’t. As an over generalization I will mulligan any hand in limited that has 0,1,6, or 7 lands and keep any hand that has 3 or 4 lands, with the 2 landers and 5 landers being the close ones. However, like I said, that’s just an overgeneralization. In my opinion, one of the things that most prevents good Magic players from being great is when something is right 90 percent of the time they do it a hundred percent of the time. They will keep any 3 or 4 land hand even when they have the knowledge that they are playing against a much better deck and can’t really win unless they draw a specific card or two that the opponent can’t deal with. Often in spots like this you will want to mulligan more aggressively even if you had, say, a 3 land 4 spell hand. I’m not saying you should mulligan the perfect 3 land hand with all good spells (but not your best spell), but that you should consider mulliganing the 3 land hand with a 4 drop, 5 drop, and 6 drop (and still not your best spell).
When deciding whether or not to mulligan a close hand, there is no one piece of information that overrides all the rest. It’s about weighing these ideas and making the choice that you believe will give you the highest overall win percentage. First think about whether or not your deck mulligans well, remember you will still choose to keep some close ones with Cawgo and mulligan some close ones with Valakut, but you want to this to be on your mind and weigh into your decision making process. Next think about if there are key cards in your hand for the matchup and or if there are particularly bad cards in your hand for the matchup. Then use this information to guess your estimated win percentage if you keep this hand, based on the different ways you can envision the game playing out with it and compare it to your estimated expected win percentage with a random 6 cards in the matchup. Of course use these estimates to make your decision because the goal is to win as often as possible.