"How many lands do you play in Constructed?"

"Oh, that depends. It ranges from 18 to 27 depending on a ton of stuffâ€”format, curve, whether your deck is aggro or combo or control, utility land count. There are even some weird and wonderful decks that play as few as one or as many as 37. There’s no way to give a definite answer to that question."

"Fair enough. So I suppose the same is true for Limited? It depends on your curve, color requirements, format, all that sort of stuff?"

"Oh no. In Limited you always play seventeen."

"That seems odd. What if you have heavy color requirements or an unusually high curve?"

"Doesn’t matter. Play seventeen."

"But that’s the same as if you had only one five-drop and no excess mana sink."

"Seventeen."

"But—"

"SEVENTEEN!"

Sound familiar?

This, my friends, is the distinctive, stultifying sound of inflexibility.

Why is seventeen lands correct? Because it’s the rule. Why is it the rule? Because it’s correct.

Most of you reading this article know the "rules." You know that Hill Giant is better than Grey Ogre and both are substantially better than Eager Cadet. You know not to splash double colored cards and that you always want at least one more colored source than cards of the color you’re splashing. You know to take bombs over removal over creatures over tricks. You know that auras are bad, because they open you up to getting two-for-oned. And you know that Limited decks always play forty cards, seventeen lands, and twenty-three spells.

Butâ€”why seventeen? How has this unassuming number been painstakingly determined to be precisely right? Surely there must be some very powerful mathematics behind it.

In fact, there is. Seventeen lands in a forty-card deck is considered the right number for a very good reason: sophisticated computer models have run for thousands of hours. We’ve generated random hand after random hand and estimated the percentage chance of winning the game in every single different Limited format. It turns out, after this exhaustive analysis, that having precisely seventeen lands gives you a 56.4% chance of winning the game on the play and a 54.3% chance on the draw. Every land you add or take away reduces this by roughly 5%.

…

I’m kidding, of course. In other news, exactly 42.7% of statistics are made up on the spot.

A lot of articles talking about land count and deck construction have large, intimidating tables with headings like "fifteen lands, on the play" and "eighteen lands, on the draw." They have percentages to three decimal places. They have long paragraphs pontificating on the hypergeometric distribution, independent variables, and standard deviations. Their authors did actuarial studies with a double major in statistics and, frankly, aren’t about to let you forget it.

Speaking as a former math tutor who did university-level statistics, these articles are a lot of fun. There’s nothing like a good, incomprehensible table to give you that warm, rigorous feeling inside. If only they were as useful as they are interesting.

Sadly, real life Magic is neither warm nor rigorous. It’s messy; it’s complex; it’s confusing. You can naively calculate the probability that at least four lands will be in the top eleven cards, but how do you account for mulligans? What about color screw? What about the Ponders and Impulses and Llanowar Elves that alter the probabilities ever so slightly? Or the Civilized Scholars and Mad Prophets that push it ever so slightly in the opposite direction?

My point is that these decisions aren’t reached on the basis of mindless theory and number crunching. They’re reached on the fine balance of what exactly your deck needs. Seventeen is a useful guideline, and that’s all it is. Treating it as a hard and fast rule will only cost you the edge.

From 16 to 18

With all of that said, it’s a rare Limited deck that wants less than sixteen or more than eighteen lands.

In my entire Magic career, I can recall only twice that I’ve played nineteen lands. (Well, and once that I played 40 lands, but that’s another story.) One was in a reject rare draft, and the other was a Zendikar deck with a high curve and double Dread Statuary. Fifteen lands is perhaps a little more common. It happened with some regularity in sixteen-land formats like triple-Scars, Ravnica, and the original Mirrodin. Occasionally you would end up with four Myrs, two Spellbombs, and only one six-drop.

In Avacyn Restored, you want to be in the sixteen to eighteen range juuust about 99.7% of the time. (This making up numbers thing is fun. I could get used to it.)

But which direction do you want to go? When building your deck, you should always ask yourself two questions.

1. What happens if I get screwed?

Can you still win if you miss your fifth land drop for a couple turns? What about your fourth or your third?

Decks with a lot of two- and three-drops can err on the side of playing fewer lands. If your plan is to curve Kruin Striker into Riot Ringleader into Hanweir Lancer into Fervent Cathar, it doesn’t matter no fourth land is forthcoming. Particularly if you have less than three five- or six-drops, consider sixteen land. The 17th is needed mostly in decks that can’t function well on four mana for more than a couple turns.

Look at your spells as well. Are two of them Guise of Fire, another Righteous Blow, and another Pillar of Flame? Cards like this are great for establishing early tempo, great when you overwhelm them with efficient creatures while skipping land drop #5, and fall flat when you topdeck them with eight lands on the table. The more cheap, tempo-oriented spells you have in your deck, the less that fourth or fifth land really matters.

Another factor that lets you get away with a lower land count is cantrips. Alchemist’s Apprentice and Fleeting Distraction will help you dig to that fourth land drop when Mist Raven is stuck in your hand. Generally you can afford to be a land light if you have a few Apprentices in your deck. Keep in mind cantrips aren’t very good when you’re flooded, as generally every land past the seventh is a blank, so they’re 40% to draw you something useless.

I’ve heard the metric proposed that you should cut one land for every three cantrips that cost less than three and similarly for every two mana accelerators. Like the principle that you should always play seventeen land: it’s crude but gets the rough idea across.

A final important point is that it’s not only the number in the upper left corner that tells you how much mana a card is going to cost you. Burden of Guilt costs one mana, but if I’m using it for a few turns with less than six lands in play I’m not very happy. Similarly with Evernight Shade or Angelic Armamentsâ€”this is only ever a three-drop if someone is trying to Spell Burst it. Even though you might be able to function (barely) on fewer lands with cards like these, don’t be fooled. Consider Armaments and the like on the level of six-drops when doing your land count.

2. What happens if I get flooded?

If you draw six spells and eight lands, how well can your deck cope? Are those spells things like Thraben Valiant and Ghoulflesh, or are they things like Stonewright and Forbidden Alchemy?

Amass the Components is a prime example of a spell that’s well suited to an eighteen-land deck. When you have access to lots of mana, all expensive card draw is good. Three Gryff Vanguards would lean me towards including that extra mana source. But Amass the Components is the plumber of Avacyn Restored blue commons, because boy does it solve flooding problems. With three of these, I’d start eighteen lands almost regardless of the rest of my deck.

Mad Prophet is another perfect example. Any deck with a couple of these and a lot of clunk doesn’t need to be shy about land number eighteen. If you can turn lands into cards, you get to play extra! Of course, red decks in this format tend to be aggressive with a low curve, so the double Mad Prophet decks would probably play sixteen if it weren’t for their land-looting four-drops.

Curve is, again, a factor. If you’re playing a Griselbrand without the 18th land, you are a braver man than I. Even three six-drops and an Archangel might make eighteen the right call depending on the rest of your cardsâ€”Borderland Ranger and Alchemist’s Apprentice or Amass the Components and Stonewright. Rule of thumb: are you going to feel screwed if you miss your sixth land drop for a couple turns? If yes, get that 18th in there.

Mana sinks make flood more palatable. These are cards that, unlike Griselbrand, don’t need you to have eight mana available but become better when you do. The Trigons, Forbidden Alchemy, and Contagion Clasp are all good examples. Equipment is possibly the best, particularly clunky equipment like Moonsilver Spear and Angelic Armaments which gain a ton of value when you have mana up. From Innistrad, any of the expensive flashback spellsâ€”Forbidden Alchemy, Sever the Bloodline, Grasp of Phantasmsâ€”made eight-land draws very competitive, and I had many an eighteen land U/B control deck.

There are a few other circumstances when you would consider playing eighteen lands:

One is if your deck is extremely good and one of the few ways you can lose is if you’re unable to cast your spells. Martin Juza won Grand Prix Bochum with an insane infect deck featuring double Hand of the Praetors and double Contagion Clasp; he played eighteen lands just to be sure and didn’t drop a game.

Another is if your deck is awful. Its three colors with no fixing and half the time you’re not going to be able to cast your spells. In this case, play more land! You’ll at least be more likely to have the right colors at the right times. All things being equal, it’s better to draw a land too many than to effectively mulligan to four with three uncastable black spells in hand.

A third case is if your deck is fine but you’re simply one playable short. The 18th land is a fine inclusion in these scenarios; you’d rather it be a filler spell, but sometimes there isn’t one available. After all, an 18th land will be good at least some of the time.

Examples!

To finish off with, here are some examples. I’ve listed 24 possible cards for several draft decks below. Work out for yourself how many lands you’d play, what cards you’d cutâ€”and more importantly, why:

Enjoy! I’ll post my own answers next week.

Until then,

Jeremy

@JeremyNeeman on Twitter

From 16 to 18

Examples!

Deck 1

Deck 2

Deck 3

Deck 4