Why You Should Keep More Hands (And Other Bad Advice)

How many times have you heard, "You should’ve mulliganed," and, "You should’ve played around it?" Two-time Grand Prix winner Jeremy Neeman tells you why you should stop listening to this bad advice in order improve your tournament results.

Sometimes you lose games of Magic.

Yeah, it sucks. I know. We’ve all been there. We’ve all walked away from the table wondering what we could’ve done better. Maybe if we’d taken that other card off the Forbidden Alchemy, maybe if we’d chumped that turn with the Spirit token, maybe if we’d thought about ripping that Galvanic Blast

But there are two absolute classic reasons for why you lost that game. I’m sure you’ve heard both of these many times before from your friends, from your opponents, from writers, from pros. You know what? Your friends, opponents, writers, and pros were all wrong. I say this because being controversial is a good way to increase hits.

But in all seriousness, these two reasons get way more airtime than they should. Sometimes they’re why you lost the game. Often they’re not, but because they’re so immediate and visible, you might be fooled into thinking they are. Well, for the next time someone tells you:

1. "You should’ve mulliganed,"

print off this article and give him a copy. (Or a wedgie. Maybe both.)

A fascinating statistic that I just made up but sounds plausible: over ten million words have been written on the subject of mulliganing. Upwards of 9.5 million of those words have been on why people don’t mulligan enough. It’s a topic writers love to harp on.

I have to say, ten years ago the writers were right. The average PTQ grinder used to keep all kinds of rubbish. I remember reading an article about mulligans on this very site back in 2003. The format was Onslaught Block Draft, and Ken Krouner was discussing with his friend Aaron if he should mulligan the following hand on the draw:

The eventual conclusion they came to was that yes, this was a mulligan—but not before Aaron kept it and promptly lost. Now, I have to ask: how many of you reading this article would’ve kept that one? If it’s more than one in five, I’d be surprised. These days, people know an awful hand when they see one. The information density in Magic is increasing, thanks to wonderful websites like the one you’re reading right now, the advent of Magic Online, and the sheer volume of high-level tournaments that get played (think SCG Open Series.)

If you want some historical perspective, go back and read Ken’s article. It’s a very good article, and all indications are that Ken was way ahead of his time. But some of the examples he uses make me sit back in my chair and blink once or twice. "Some might call me crazy," he admits, "but this hand has to go back." Yeah Ken, I agree it’s a rubbish hand, but crazy? Really? It’s a one-lander on the play, and that Birchlore Rangers ain’t a Llanowar Elves."

"When in doubt, throw it out." It’s a fine principle. But like any principle, relentless application of it can quickly turn into over-application. Most people know by now that if a hand’s close, throwing it back is the "pro" play. That turns into a problem when the desire to make the "pro" play overwhelms the desire to make the correct play.

Going a card down has real costs associated with it. You’re more likely to miss land drops, run out of gas, not have the two-drop when you need it; all that sort of stuff. Most "good" six-card hands are still only as good as a decent seven.

Ok, application time. The format’s Cube Draft, and you’ve drafted a U/B Fish deck. Would you mulligan this hand on the play?

For me, it’s a keep, and it’s not very close. Yes, you have five lands, and yes, it does hurt a little if they have the Inquisition / Thoughtseize / Lightning Bolt. But you have perfect mana, a Dark Confidant arriving on time, a bit of disruption, and utility lands to compensate for having a couple too many.

It’s better than the average six. It’s not a great hand, but not every opener will have a perfect distribution of lands and spells and plays on turns 2, 3, and 4. A lot of people overestimate the importance of the opener and underestimate the capacity to draw out of it.

For example, one-land hands on the draw. It’s common wisdom by now that these deserve to be mulliganed. The reasoning goes something like this: even though you might be 70% or 75% or 80% or whatever to draw that land by turn 2, that’s not your chance of winning the game. That’s the likelihood that you even get to play a game of Magic.

You have a 25% shot of missing your second land drop and consequently losing on the spot. And even if you do get that second land, you probably have to hit a third and a fourth to be able to play all your spells. What are the chances of that?!

Well, very good actually. Pretty close to 100% if the game goes long enough. Neeman’s first principle of mulliganing is this: every hand will eventually draw the cards it needs. You have lands in your deck, right? Chances are some of them are in the vicinity of the top of your library.

The two questions you need to ask about any sketchy hand are:

  • How good is this hand if I immediately draw out of it?
  • How good is this hand if I don’t immediately draw out of it?

One-landers get a big tick next to the first question. If you do draw a land or two quickly, your hand is the nuts simply because it has so much gas. Remember the oft-repeated mantra that screw beats flood: you can win a lot of games by just drawing less lands than your opponent (while still being able to cast your spells of course.)

In regards the second question, that’s where most one-landers fall down. But, and this is the crucial point, not all of them. Take, for example, the hand I had in game 1 of the quarterfinals of GP Melbourne two weeks ago on the draw:

This is not a great hand, but I kept and would keep again. Thinking about it in terms of the second question—how good is this hand if I don’t immediately draw out of it?—we can see that in fact the hand is serviceable even if that second land doesn’t come on time. Sensory Deprivation slows down the early rush, and once we do draw the second land, it’s not vital that we hit the third straight away because of Deranged Assistant. (Rule of thumb: most keepable one-landers will have a two-drop mana accelerant.)

Here’s another way to think about it: we’re 25% to draw two spells in a row, 25% to draw two lands in a row, and 50% to draw one land and one spell. So our hand is 25% to be really good, 50% to be still quite reasonable, and 25% to be awkward. But even in that bottom quadrant, we aren’t necessarily losing the game—we’re Depriving his two-drop of Senses and still feeling happy if we hit a land the following turn. If the hand was missing either Sensory Deprivation or Deranged Assistant, it’d be a mulligan.

Would you mulligan this on the draw?

Hopefully not, because this hand is quite solid. This was from a draft back in triple Scars of Mirrodin. I ended up missing my second land drop, having to discard Moriok Replica on turn 2, and still going on to win the game.

Despite the one land, this hand has a lot going for it:

  • A two-drop accelerant, making the later drops that much easier to hit
  • Necropede and Shatter to slow down the early rush if you do fall behind
  • Insane mid-to-late game in Scrapmelter and Skinrender, the two format defining four-drops.

Worst case, if you miss land drops and have to discard as I did, your hand is still good enough to win the game. Skinrender and Scrapmelter make up for lost time, and Necropede is very difficult to attack into. Of course, if you do draw lands it’s one of the best hands you could hope for. Flametongue turn 3 into Flametongue turn 4 isn’t fair at all.

Final example. This is from Pro Tour Paris last year, playing RUG against Vampires. Your hand for game 2 on the play:

Yes, you have no play until turn 4, but this hand is absolutely a keep. Your mana’s perfect, both your spells are great against them, and you have a ton of cards you can rip: Explore, Pyroclasm, Lotus Cobra, Preordain, Lightning Bolt. To top it off, they have four Inquisition of Kozilek in their deck, and keeping this hand makes that card a brick.

But let’s explore the other perspective. The logic to mulligan this hand goes something like this: "He’s an aggressive deck. If he starts off with Pulse Tracker into Kalastria Highborn, I’m already on the back foot. Then all he has to have is a Gatekeeper of Malakir or a Go for the Throat to handle my Obstinate Baloth, and I might die before I ever get Wurmcoil down. This hand is too slow. I have to mulligan to find a Pyroclasm or something similar."

Ok, so what we’ve done is elucidated a line of play from him that could lead to you losing the game. It assumes you draw a lot of lands or Jaces or similar and assumes he has a pretty good draw, but it could happen. The question is, though, how likely is it to occur? Relative to the chance of him having Inquisition or Kozilek on turn 1, or a slower draw, or no way to immediately deal with a 4/4?

Let’s say we do mulligan and find a Pyroclasm. "Great!" we think. "Now we won’t lose to the fast beats + Go for the Throat draw!" Then he plays turn 1 Inquisition, turn 2 Bloodghast, turn 3 Vampire Nighthawk, and we die uneventfully after missing our fourth land drop for a couple turns. By making our plan a little better against the cards we thought of, we severely compromised it against a larger set of other cards.

This segues quite nicely into the next classic counsel:

2. "You should’ve played around it."

Well, maybe you should’ve, maybe you shouldn’t have. It depends on four factors:

  • How bad it is if you play into it and he has it
  • How good it is if you play into it and he doesn’t have it
  • How much you lose if you play around it
  • How likely he is to have it

Here’s a simplistic example. Say it’s game 1, and you’re playing U/B Control against Delver. You have eight lands in play and Grave Titan in hand, and you’re on ten life. Your opponent has no graveyard, Geist of Saint Traft in play, five lands, and one card in hand. Do you play Grave Titan?

Hopefully everyone said yes, of course. You lose if he has Mana Leak, but by holding it you still lose if he has Mana Leak and you also lose if he doesn’t have Mana Leak. Geist hits you down to four next turn, you (might) draw a land and be able to resolve Grave Titan through Mana Leak, but you’ll die to the Angel anyway.

Now take that same scenario, but instead of ten you’re on eleven. Do you still run the Grave Titan out there and cross your fingers?

The correct play is to tap six mana and cast Grave Titan, and it’s not close. Seriously, not at all. Let’s do a cost/benefit analysis to see why that is.

Benefit #1: If he has precisely Mana Leak in hand and the top card of our library is a land that comes into play untapped, that play lets us still be in the game.

Benefit #2:

Now let’s compare that to the detriments of holding Grave Titan.

Detriment #1: If he has Vapor Snag or Gut Shot in hand, the play loses us the game.

Detriment #2: Also if he has Snapcaster Mage.

Detriment #3: Also if he has nothing but draws Vapor Snag or Gut Shot over the next couple turns.

Detriment #4: Or if he rips Invisible Stalker next turn.

Detriment #5: Or Sword of War and Peace.

Detriment #6: Or a Delver that subsequently flips.

Detriment #7: Or Ponder into any of the above.

Detriment #8: It’s like 75% to not matter even if he does have Mana Leak. We need to draw a land that comes into play untapped. That’s not very likely, and if he draws one of the many cards that will kill us over the next couple turns it’s irrelevant anyway.

Even if you were on fifteen life it would still likely be correct to play Grave Titan. You’re giving up too much and gaining far too little if you hold it in hand. Six damage a turn? If Grave Titan resolves now, you’re 90+% to win the game. If you draw a land in two turns and Grave Titan resolves then, you’re dead because Geist has already hit you twice.

Take this board state from Australian Nationals 2010.

Your board: Stone Golem, Awakener Druid, 4/5 Forest, Sylvan Ranger with Volcanic Strength (all tapped), six Forests and Mountains

Your hand: Two Pyroclasms, Berserkers of Blood Ridge

Your life: ten

His board: Garruk’s Companion with Volcanic Strength (tapped), six Forests and Mountains

His hand: two cards

His life: six

Your choices are: Pyroclasm twice, clearing the board, or play Berserkers of Blood Ridge and hope he doesn’t kill you with Lava Axe / Fireball. What do you do and why?

I once gave this situation as a poll to my readers, and I was shocked—shocked!—at the results. More than 60% wanted to get that Garruk’s Companion out of there at massive cost to their own board and chances of winning the game. Let’s cost/benefit break this down one more time.

Benefits to Pyroclasming twice:

Benefit #1: If he has Lava Axe in hand, you’re still in the game.

Benefit #2: If he has Fireball in hand, you’re still in the game, although if he draws creatures you’re not looking great.


Detriment #1: If he has Greater Basilisk in hand, you’re suddenly far behind.

Detriment #2: If he has Grizzly Bears and Giant Growth in hand, same story.

Detriment #3: If he has more creatures on the top of his library than you, you’re not looking great.

Detriment #4: Or a bit of removal. Anything other than land really.

Detriment #5: If you don’t sweep boards, you’re practically guaranteed to win against anything other than Lava Axe / Fireball. What’s more likely: Lava Axe or some generic combination of dudes and removal?

Again, it’s not close. Take the small chance of just losing on the spot over the enormous chance of getting ground out over several turns—every time.

So the next time you lose a game, don’t blame mulligans or Mana Leaks. Blame Brian Kibler.

Until next time,


@JeremyNeeman on Twitter