Thursday, February 11, 2010

The Theory of Combo Decks

I wrote this last season, so I'm posting it here. I don't stand behind everything in here, and I was probably just flat wrong on combo vs aggro, but there's still some interesting theory in it. It's rather outdated, just by how derisive I am of most a+b combo decks, when the current top deck in extended is one with two a+b combos (and it loses to aggro). Here's the article:

Articles on Magic theory (even the very good ones) almost always ignore combo decks, by making excuses and exceptions for them, while talking almost exclusively about aggro and control decks. For example, in talking about the value of a mana curve, to act like every deck either plays dudes on the first few turns of the game, or sets up with removal spells and card drawing for mass removal, completely ignoring the set of decks that do nothing until they untap and win. The way I see it, this is part of a broader problem: most Magic players, even very good ones, seem to view combo decks as freak accidents or completely unique concoctions that have nothing in common except their speed and style of winning. This is wrong. For all the talk about the strategy and theory inherent in aggro mirror matches, aggro-on-control, or control mirror matches, there are just as many guiding principles behind combo decks- they just happen to be different.

Until recently, there was almost no way to talk about the theory behind combo decks. They don't get card advantage, really; Dark Ritual is pretty poor by that metric, and there's absolutely no way to describe a deck like Dredge that actively wants to dump its hand. Card quality? All but a few cards in the deck don't really do much of anything (by themselves), so no. Tempo? How is playing a land and saying go for two or three turns tempo? However, Zac Hill has written a couple fantastic articles about what he calls "interaction advantage." Giving a definition is difficult, but if I could take a shot at it, I'd say that it's the ability to interfere with an opponent's gameplan while advancing your own. Generally speaking, a pure aggro deck wins via tempo: its cards come out the gates faster and at higher quality than the opponent's. Control wins via card advantage, that's obvious. Combo decks win via interaction advantage: it plays a set of cards that win the game immediately that the opponent can do nothing about. A combo deck (or at least, a good combo deck) is not one which can simply win the game quickly. A hypothetical card that costs zero mana and reads "you win the game next turn unless your opponent plays a land or spell" probably isn't the basis for the next killer combo deck, as literally everything can stop it. All combo decks want to win as quickly as possible, of course, but they also must be difficult to disrupt to be any good.

Just like other broad archetypes have their subcategories, "pure" combo decks generally break into two camps: decks which feature some kind of engine or a single card that wins the game when played, or decks that require two (or more, but usually two) specific pieces in order to win. For lack of better terminology, I will call them engine and a+b decks.

As a straightforward example of an engine combo deck, here is LSV's list for TEPS, quickly on its way to becoming one of the most copy-pasted decklists of all time (behind Sligh, Trix and the original incarnation of The Deck).

4 Desperate Ritual
2 Electrolyze
4 Lotus Bloom
4 Manamorphose
4 Mind's Desire
4 Peer Through Depths
4 Ponder
4 Remand
4 Rite of Flame
4 Seething Song
2 Sleight of Hand
2 Tendrils of Agony

4 Cascade Bluffs
4 Dreadship Reef
2 Island
4 Shivan Reef
4 Steam Vents

The deck's engine is, of course, Mind's Desire. The deck works as well as it does not because if you have Mind's Desire and a specific card in your hand, you automatically win the game, but because the card works so well with every one of the 38 other spells, you will probably win when you cast it. This is an important point to note about engine decks: very rarely does a deck come along that, in goldfishing, will win a full 100% of the time upon casting its main card. A win percentage of between 50% and 99% (depending on the storm count and amount of mana floating) is judged to be good enough. A key mistake that people make in looking for the next big combo deck is looking for cards that will always win the game. These decks are (aside from not being engine decks) generally less reliable and more easily disruptable. This is mainly due to the fact that most of these sorts of combos are just so damn obvious that Wizards would never ship them out the door without making sure the deck was bad first.

This brings us to the next type of deck: the a+b combo deck. Here is a list of one of the few tournament-caliber decks of this type: Flash.

1 Badlands
4 Flooded Strand
4 Polluted Delta
1 Tropical Island
4 Underground Sea
1 Volcanic Island
1 Body Snatcher
1 Carrion Feeder
1 Karmic Guide
1 Kiki-Jiki, Mirror Breaker
4 Protean Hulk
1 Sylvan Safekeeper
4 Brainstorm
4 Cabal Therapy
4 Daze
4 Duress
1 Echoing Truth
4 Flash
4 Force of Will
4 Lim-Dul's Vault
1 Massacre
2 Mystical Tutor
4 Peek

In contrast to the above deck, Flash is one that has no engine that will win in combination with the other cards in the deck; it simply has Flash and Protean Hulk that, for two mana at instant speed, will automatically win the game if the opponent doesn't do anything (and usually even if they do). It plays a multitude of cards that, without the rest of the combo, are utterly useless on their own. The upside, of course, is the deck's breakneck speed, the fact that it doesn't need the "support" cards an engine deck does, and that it will win every single time in goldfishing, rather that it just being overwhelmingly likely.

The overwhelming majority of bad combo decks are of this latter type, because the cards to play are so much more obvious. This is also the type of deck that gives the popular impression of most combo decks: their cards are useless until they get two or three specific ones in play, in which case, they win. Most of the time, these instant-win interactions are better suited to being a part of an aggro or control deck. Project X, for example, was a GWB midrange deck that could play three specific cards to gain infinite life. Many people assumed, wrongly, that this was the deck's only strategy, and played specifically to combat that. If they opposing player was any good, they just attacked with dudes and won. Swans in Standard is similar: it's a control deck that can combo off when it feels comfortable about its board position. On a more personal level, one of my favorite decks of all time is a deck I made for Regionals about two and a half years ago that I called The End, and wrote an article about for StarCity. It played sixteen green mana accelerants, big fat green creatures, including Protean Hulk, which could be sacrificed to Greater Good or Miren, the Moaning Well to lock the opponent down infinitely. Usually, though, I just played big guys and attacked. Don't assume that any "instant win" combo needs to go in a combo deck.

So what do these decks have in common, aside from speed? The cards in both decks can be broken down into a few general categories. Obviously, it needs to play land. Because TEPS has to cast a six-mana spell whereas Flash only needs two mana, the landcount is higher in the former (though still lower than that of a "fair" deck).

Neither deck plays any of what would generally be thought of as card drawing or card advantage. Generally, since this will cost more mana and not necessarily get the card you want, card selection or tutoring is generally what's used instead. TEPS plays sets of Ponder and Peer Through Depths for this, whereas Flash uses Lim-Dul's vault and a couple Mystical Tutor. What are generally preferred here are not cards that tutor up just one specific type of card in your deck, but card selection mechanisms that can find whatever you need at the moment. The package TEPS is running can be used to find nearly anything in the deck, whereas Flash's Mystical Tutors, while they can't be used to find Hulk, can get any of the disruption or the namesake card. It also played Peek, but I'm not sure what the hell category that can fall under. (It's worth noting, though, that Flash was banned before Future Sight became legal, and would have run Summoner's Pact, which contradicts the idea of the card selection being able to find anything; however, it would have changed the skeleton of the deck to be even more consistent.)

Next up is the disruption package the deck uses. Combo decks use disruption in a different way from other types of decks: aggro might use hand disruption to take away removal and bigger guys, whereas control uses it as a fundamental part of their strategy to stop whatever the opponent is doing. Combo, on the other hand, only cares about taking the cards that could possibly interact with it. It might seem bizarre that it's a great first-turn play to use your Duress to take someone else's Duress, trading one-for-one for the exact same card while dropping a mana behind in the bargain. Often, though, that Duress you took is the other player's only way of interacting with you, so it's a fantastic deal for you. TEPS has the playset of Remand that, in addition to being tremendously useful on the combo itself (Remanding the Desire after the additional copies resolve), can stop anything the opponent throws at the deck's strategy. Flash plays a full eight hand disruption, plus Peek to make Cabal Therapy more useful, plus a one-of Echoing Truth to bounce anything annoying like Leyline of the Void, plus a full set of Force of Will. This is what really made the deck as good as it was: while it could have been faster if it had used those 17 card slots on more tutoring or card selection, instead it used them to completely prevent anything from interfering with the strategy. I don't mean to go all Mike Flores gushing about how great the deck was, but it's worth pointing this out to explain how it wasn't just all combo cards and ways to find them: it was going to screw up whatever your plan was for beating it, period.

After this comes the mana acceleration. Flash was unusual in this respect because, as it relied on a two-mana card, it really didn't need any (though some people added some anyway just so they could show off by winning on turn one or zero). TEPS, on the other hand, plays sixteen, which should tell you how important Ritual effects are to almost every engine deck. It's fairly easy to keep combo out of the top tier: don't print those Rituals. Guess what, there haven't been any, and Standard is a large mash of creatures smashing into each other and making vaguely sexual grunting noises.

Finally is the win condition itself. For almost every combo deck, the way that the deck gets the opponent from 20 to 0 (or 60 to 0) is almost irrelevant, because the way the deck wins isn't really what drives it. There was at least one other Flash kill, involving Disciple of the Vault, that saw play at Columbus, and Grapeshot plus Pyromancer's Swath is an alternative to the Tendrils kill- plus they almost all run Brain Freeze in the sideboard. Occasionally, people will get excited when they see a card that's the kill for some theoretical combo deck, or an improved kill to an existing one. Unless it speeds the deck up or drastically changes a matchup, chill out, it doesn't matter.

While it's great fun to explain why good decks are good, it's more useful to focus on something else: what separates the format-defining, DCI-enraging monstrosities that go on to take seven out of eight slots at the top of a Pro Tour from the bad combo deck a kid might bring to your FNM? They're the same general concepts. Here, then, is what makes a good combo deck:

First, and most obviously, it has to be faster than the best aggro deck in the format. This one is straightforward: if you get 50/50 or worse against the aggressive decks, why aren't you just playing those? However, as controversial as it may be to say, the speed of a combo deck, if it can beat aggro, is almost irrelevant, as in every other matchup, you'll be waiting as long as possible to combo off so that you can disrupt as much as possible.

Second, it has to not completely scoop at the first sign of an Island. TEPS, like Dragonstorm back when it was in Standard, had favorable matchups against blue decks partially because of Gigadrowse and the fact that its main card is virtually uncounterable by what's normally played in the maindeck. Essentially, this one comes down to having a solid enough deck that it can wait and set up its turn, and playing the right disruption elements to stop them. It doesn't necessarily have to be a good matchup- plenty of good combo decks, like Elves, don't have this- but if it's dreadful, there's a fundamental weakness in the deck.

The third part is the most complex, and ties into the last. The angle that you are attacking from (that is, the combination of cards that wins you the game) can't be the same angle of attack that other decks are already trying to defeat. What this means, usually, is that if you make a new combo deck that relies on you untapping with a certain creature in play, and that creature has no built-in way to defend itself, it's almost certainly a terrible combo deck. I can't tell you how many times I've heard people going insane over the new "broken" creature (or sometimes artifact or enchantment) on a spoiler that almost always costs four or more and doesn't win the game when you cast it. They'll frantically try to build a deck around it, win on turn four 10% of the time and get excited before realizing that their broken combo deck loses to the creature removal already played in every single deck. This point may seem to be contradicted by how good the Elves deck is in Extended, but it isn't. That deck doesn't depend on a specific one of its creatures surviving; its engine is Glimpse of Nature, not a single fragile dude. Plus, they all cost one or two, it doesn't necessarily depend on untapping with them, it plays four of a card that can bounce any of its creatures for free, and if you kill all their Heritage Druids, Nettle Sentinels and Wirewood Symbiotes, it'll just beat you down with the ten power they have remaining. The point is, it's incredibly resilient, and this is an aspect that every single successful combo deck needs to have, and it's much, much more important than what turn it kills, if that turn is fast enough for the format.

Fourth, the card selection available has to be good enough to find your key parts- and you can't just lose if you don't find the part you're looking for. If every piece of TEPS was reprinted in the next set except the card selection, and there weren't any good alternatives available, the deck would fall apart to its own inconsistency. Alternatively, as a deck with no card selection, if Elves doesn't find its Glimpse, it's perfectly capable of just making a board packed with dudes who will happily attack for 20 over a couple turns. TEPS doesn't have that option, but anyone who's played it will tell you how many games they've won without Desire by casting Tendrils, Remanding it, and casting it again.

Finally, if your new combo deck was also independently discovered by someone who posted it on MTG Salvation under a "decks for critique" forum, abandon it immediately, it's garbage.

The basic point is this: theory of combo decks is just as valid and just as important as theory about any other deck, and these decks shouldn't be ignored for purposes of discussing theory just because they win in some weird way.

On a related topic, an interesting thing to me has been the discrepancy of results among people playing Elves between the latest GP and all of the PTQs that have been going on. Simply put, I think that only a very small minority of people that play the deck are the slightest bit good at it; it really is impossibly difficult to play correctly. In a recent game when I was dicking around with a bad deck on Magic Workstation, I played against someone using the deck. They went off, and probably made upwards of twenty or thirty misplays on that one turn... and I lost anyway. I think that a lot of people play a few matches with it, think they have the hang of it since they're doing fairly well, and miss top 8 at a PTQ because they were never that good with it. It's not just the small, obvious misplays that they might make a thousand of over eight rounds, like forgetting to untap Nettle Sentinel or make a few additional mana, but big mistakes in strategy. Way too many people try to go for the combo and end up losing rather than just crapping five or ten power onto the board and making your opponent play more conservatively than the Heritage Foundation. And for god's sake, learn to play it fast! After my match was done at the last PTQ, one of the only matches left was a guy going off with Elves, and god damn did he just take forever. You may find yourself with a beautiful board position, and a beautiful hand. And you may ask yourself, how do I work this? And you may ask yourself, where is that Nettle Sentinel? And you may ask yourself, which creature should I play first? At this point, you should probably just tell yourself that this is not your winning deck. Or... maybe you should just goldfish it three hundred times. Either way.


Matt said...

Summoner's Pact would primarily have been used so that Mystical could indeed get a Hulk. (Or an ESG)

Post a Comment