I’ve done a lot of “Keep or Mulligan” articles for CFB, but I haven’t written a general article on mulligan theory, so I decided to write this one! In this article, I’m going to give you three tips to mulligan smarter.

1) Know Your Deck

Mulliganing is always contextual. You never keep or mulligan a hand—you always keep or mulligan a hand playing a certain deck, against a certain deck, on the play, or on the draw. Sometimes you don’t know what they’re playing, but you always know what you’re playing, and hands a certain deck would keep are hands another deck will mulligan.

One of the most important factors in deciding to mulligan is to know whether your deck mulligans well or not. Certain decks have a much harder time winning if they’re on six or fewer cards, whereas other decks don’t mind nearly as much, and knowing which deck you’re playing is instrumental to be able to make a decision on the fly.

As a general rule, I like to classify decks in my mind as either “quantity decks” or “quality decks.”

Quantity decks rely on a critical mass of resources to work. A lot of their cards are equivalent to each other, and there often isn’t something that stands out. Those decks usually mulligan badly because they don’t have a single powerful card to make up for starting on 6 cards, and each resource they’re missing makes all of their other cards worse.

Certain combo decks are prime examples of this, like Scapeshift or Storm. Scapeshift just wants to get to seven lands, and Storm wants to get to a bunch of spells (though this is lessened in the newer Modern versions with Gifts Ungiven). In Scapeshift, other than the card Scapeshift itself, every card does more or less the same thing, so having one fewer card just means one less card to work with. You’re not going to mulligan into better cards very often because all of your cards do the same thing. You’re simply going to mulligan into fewer cards.

Another example would be synergy-based decks, such as Merfolk. Every Merfolk that you draw makes all your other Merfolk better, which means that removing a card from your opening hand makes it so that every other card you have becomes worse, and there’s no card that can make up for this.

When I’m playing a “quantity” deck like those, I’m much more conservative in my mulligan decisions because I know I don’t have a powerful card to bail me out. I’ll often keep sketchier hands because I know the alternative is very bad. If you’re playing a deck like this and mulliganing frequently, chances are you’re mulliganing too much and should be more liberal with your keeps.

The second type of deck are the “quality decks.” In my mind, quality decks have certain cards that are much better than the rest of your deck, and that can bail you out from a mulligan because they are so individually powerful.

An example of this is a deck like Faeries. Faeries had a card—Bitterblossom—that was so extraordinarily good that a hand with six cards and Bitterblossom was better than most hands with seven cards but no Bitterblossom. The same was true for Caw-Blade, where most hands with Stoneforge Mystic were better than hands without, even if you had one fewer card. The U/R Ensoul Artifact that Mike Sigrist took to 2nd place at the Pro Tour is also in this category. Hands with Ensoul Artifact were much better than hands without it. Most recently, we have Mono-Red Aggro as a great example of a “quality deck” because Hazoret the Fervent had the ability to bail you out from basically any number of cards (in fact, it gets better sometimes if you start with fewer cards). Because I had 4 copies of Hazoret in my deck, I knew I could mulligan any marginal hands when I played Mono-Red.

In practice, this means that if you’re playing one of those decks, you can mulligan much more liberally. If a hand is mediocre, you don’t have to be scared to throw it away because you have a very powerful card that can make your six-card hand much better. I’ll never keep a bad hand in the Mono-Red mirror, for example, because I know that any hand with Hazoret is likely to be excellent, regardless of what else I have. If you’re playing a deck like this and you almost never mulligan, then in all likelihood you’re settling for hands that aren’t good enough, and you should mulligan more aggressively.

I remember when we first played Caw-Blade, Ben Stark asked the group about a hand he had in the tournament. He was playing the mirror, he was on the play, and his hand was:

Mana Leak
Spell Pierce
Jace, the Mind Sculptor
4 Lands

Most people thought it was a keep since it’s a very functional hand, but Ben disagreed—his argument was that you had such a high number of good 2-drops in the mirror (Squadron Hawk and Stoneforge Mystic) that, if you were on the play, playing one of those on turn 2 was a huge advantage that was worth mulliganing for. The opening hand wasn’t necessarily bad, but being on the play was very important specifically because of those cards, and you were wasting that advantage if you didn’t have them. After playing with the deck a little more, I started agreeing with Ben.

It’s also important to know your deck, because you need to know the range of hands you can actually get. When you look at an opening hand, you have to identify whether that’s normal for your deck or an anomaly. If it’s a normal hand, and you’re not playing against a specific matchup that forces you to deviate from the norm, then you probably have to keep it even if it has something you don’t like—then perhaps change your deck the next chance you get.

Imagine that you’re playing a super aggressive deck with 18 lands. Then you open a hand and it’s a 1-lander with a lot of 1-drops. You can’t really mulligan that hand—if you’re going to mulligan 1-landers with 1-drops, then you need to have more than 18 lands in your deck. It’s similar to playing a deck with 17 removal spells in it, and then you draw a hand with three removal spells—you could think “this hand is very bad against any non-removal deck, so I’m going to mulligan,” but you have 17 in your deck! A lot of your hands—and in fact a lot of your game plans—will be horrible if removal spells are bad, and if you can’t live with that, you shouldn’t have so many removal spells in your deck to begin with.

You also have to keep in mind that your deck is going to change after sideboard. Imagine you have a Zoo deck with 16 1-drops. You might develop a heuristic for mulliganing anything that doesn’t have a 1-drop. You have so many, after all, because you need them.

Now, imagine that, during sideboarding, you take out 8 of your 16 1-drops. In this spot, it’s no longer reasonable to mulligan a hand that doesn’t have a 1-drop. If you were going to do that, you shouldn’t have taken them out.

2) Know Your Opponent’s Deck

You don’t always know what you’re playing against, but if you do, then you should adjust your mulligan decisions accordingly.

One specific scenario that often comes up in Modern is when you’re playing against a discard deck. If you know you’re playing against a discard deck (and by that I mean a deck like Jund, with Thoughtseize and Inquisition of Kozilek, not necessarily 8-Rack), then you have to be more liberal with your keeps. Discard spells turn every deck into a “quantity” deck. Mulliganing for specific cards is a bad strategy against them because they usually have six or more 1-mana discard spells that can get rid of that specific card, so ultimately the exact contents of your hand matter less than the amount of cards in your hand because the best one will get discarded anyway.

For this reason, when I am playing against a discard deck, I tend to mulligan much less. I know that I can’t count on a specific good card to bail me out, because it’ll likely get discarded, so I keep a lot of hands I otherwise wouldn’t. The same is true for most attrition decks (if they’re trading 1-for-1 then you want quantity, not quality), but it’s even more pronounced for hand disruption.

For example, look at this hand I kept against my Jund opponent at the PT when I was playing Scapeshift:

Most people said they’d mulligan this hand, but I maintain that it’s not only a keep but a good keep. If you’re trying to mulligan this hand looking for acceleration, you’re going to be disappointed. Even if you do find acceleration, it’s likely going to be discarded anyway. If you mulligan into two pieces of acceleration, then they’ll discard your big spell and you’re in trouble again. This hand is completely immune to Inquisition of Kozilek (in fact, that was what my opponent led with) and is virtually immune to Thoughtseize as well.

Sometimes, you don’t know what your opponents are playing, and you have to decide whether you keep or mulligan a hand that would be great versus an archetype but horrible against another. Imagine you’re playing Jeskai, and you have the following hand:

This hand is good against aggro and bad against control, but if you don’t know what they’re playing, should you keep it?

In my opinion, yes. If you’re playing against aggro, then the game is faster, and your opening hand represents a much lower percentage of the cards you’ll see in a game. If you keep a very slow hand against aggro, then you don’t have time to draw out of it—the game will be over within 4 turns—whereas the reverse is not true. If you’re playing against control and you keep this hand, the game will still go on for many turns—you might even draw your whole deck! Unless they have a specific hand and guess that your hand is bad, you will have time to draw your good cards.

Given that, I’d rather keep a hand that is bad versus control in the dark than a hand that is bad versus aggro because the aggro deck will punish you immediately.

3) Know Your Plan

Every time you see an opening hand, you should think about how the game is going to play out. What are you going to do in the early turns? How do you see yourself winning this game? What needs to happen?

Some hands are plans onto themselves, or need very little. If you’re playing Burn and your opening hand is two Mountains, a Lava Spike, and four Lightning Bolts, your plan is pretty clear—you’re going to Bolt them until they die. Since that’s a reasonable plan and you need very little for it to happen, you happily keep the hand.

Now, imagine you have a hand of five lands, Lava Spike, and Goblin Guide. What’s the plan here? You’re going to play Goblin Guide, hope they can’t answer it, and hope you draw a burn spell every turn of the game. That’s possible, but not very likely, and your chances are better with six cards.

What if you’re playing Burn versus Storm and your hand is Lightning Bolt, Eidolon of Great Revel, and five lands, though? What’s the plan with this hand? Well, the plan is to play Eidolon and hope that they either can’t kill it, or that they take enough damage doing so that whatever cards you draw after that can finish the job. I think this is a reasonable plan and would keep this hand against Storm.

Sometimes, a hand needs something very specific to be good, and then your job is to figure out how likely that is to happen, and whether those odds are better than trying a new hand with one fewer card. A lot of the time, a hand looks bad but only needs one piece to become good, and the chance of you drawing that important piece is higher than that of drawing everything you need in six cards (or five).

Take this hand that I had in the finals of PT Hawaii, against Kibler, long ago. We were playing the R/G Ramp mirror and I was on the draw. I mulliganed my opening hand, and my second hand was something like:

I kept this hand. I didn’t draw anything useful, and went land, land, land, land, Solemn Simulacrum, and then I died. A lot of people criticized me for keeping a hand that effectively did nothing, but I am pretty sure it was correct, because this hand needs very little to be great.

In the Ramp mirror, the most important thing is to play a turn-4 Titan. The only way to do this is to have two mana accelerants, and one of them has to cost 2 mana. This hand is only missing the 2-mana accelerant. If I draw a 2-mana accelerant, then this hand is perfect. If I don’t draw a 2-mana accelerant, it does nothing.

It’s a risky keep—I could mulligan. But what has to happen for me to have a five-card hand that plays a turn-4 Titan? I have to draw two accelerants (one of which has to cost 2 mana), 4 lands, and Titan itself. What are the odds of having that in a five-card hand plus a couple of draw steps? I don’t know exactly, but I assume they’re low—much lower than my chance of drawing just a 2-mana accelerant in the original hand.

Now, imagine that you’re playing U/W against Dredge. You’re on the draw, it’s post-sideboard, and you have three Rest in Peace in your deck. You mulligan your opening hand, and your six-card hand is:

What is your plan with this hand? You’re very unlikely to win a game unless you draw specifically Rest in Peace, and very early at that. Given that your plan with this hand is “draw Rest in Peace,” it’s better to mulligan this hand and hope for a Rest in Peace and the mana to play it in your five-card hand (or at least for something like a Serum Visions).

In the end, keeping a hand is always going to present a risk, and mulliganing a hand is going to present another—no play is “safe.” You have to decide which risk/reward ratio is better. As a general rule, the worse your position is, the more risks you want to take. In the R/G hand, the risk is in keeping; in the U/W hand, the risk is in mulliganing. In both cases, I believe the reward is worth it, because your position if you don’t take the risks is precarious, and the only way you can figure this out is by knowing your plan after you keep each hand.