Flores Friday - Fulfilling Expectations: How To Build A Sideboard That Isn’t Embarrassing
Our Managing Editor Craig Stevenson asked me to describe "The Flores Approach to Playtesting," and I started to write such an article... but during that process I stumbled on something that seemed infinitely more interesting: How to build a sideboard that is not embarrassingly bad. Curious? I hope this is helpful to you because most people play sideboards that are literally embarrassingly bad and make me want to vomit when I look at them.
This article is about the process of assembling a sideboard, not necessarily what you would put in your sideboard or how sideboards work in general. For this information, you can refer to lots of great articles; here are two that I've written in the last year or so.
Before we begin, I'd like to get some stuff out of the way and make sure we are speaking a common tongue.
1. Under abstract circumstances, there is literally only one reasonable criteria for choosing a deck. No matter what Stephen Menendian will have you believe, Zvi, Jonny, and I are all on the same page on this one. The criterion is simply that you choose the deck that gives you the highest likelihood, mathematically, of winning the tournament. There are variations to this rule (a common one is that you only need a 2-0 for a ratings invitation to the Pro Tour), but in large part, the above is the only rule that you can reasonably follow. In discreet terms, that usually means choosing the deck that has the highest likelihood of winning any individual match, but remember that the metagame shifts as you proceed through the tournament. The single greatest innovation of the Japanese, and the reason why the rest of the world has not caught up with them, is that they build their main decks to win in the early part of the tournament and their sideboards to win in the late stages of the tournament, whereas (for example) overrated American designers cling to outmoded playtest ideas and often find themselves with no chance of winning the tournament ("nice Day One deck"). I am sometimes harsh in my criticism of decks; the reason is because I - and very specifically Jon Finkel - think that this rule is the absolute minimum threshold to clean play, and that for Constructed, the single most important decision you can make occurs before you shuffle up the first time. When Jon was the reigning World Champion he would commonly shake his head at whatever Standard tournament he was drafting near, puzzled as to why every player present was not, say, Tinker. Note that the presence of this rule doesn't actually say what that right deck is; what I will say is that when you have a competent team of playtesters who all opt for different decks, probably every one of them is wrong.
2. It's really only important to prepare for the decks that are likely to show up. One thing that many forum denizens get their panties tangled in a bind about is when writers and deck designers don't take their innovations under consideration when presenting their tournament preparation. I am actually flabbergasted on a regular basis when I, for example, make Top 8 with a new deck, and am directed to read in some other forum how I was playing a substandard version of an "archetype" that - at least as far as I kn[e]w - didn't actually exist in any significant way before I published my article. You can actually see the number of people who read your deck posts! I am not trying to deprecate anyone's efforts when I say this, keep in mind. I think that innovative deck success is closely aligned with keeping the numbers in the know as small as possible; but that said, not only can you not expect other thinkers to take your ideas into consideration if they are niche, actually doing so would decrease their ability to win! I'm sure that the combined brainpower of the StarCityGames.com forums has produced one or two very good ideas before the mainstream, the Pro community, or certainly YT... But when only a half a dozen people who are all contributing to a thread are aware of Innovation X, it seems silly to prepare for it for a tournament where you are unlikely to face any of those six or so people. There is a fine line, however, between preparing for the decks that are likely to show up and not preparing for the decks at the end of the tournament; just because only one copy of a version shows up doesn't mean that it didn't show up... Especially if that version is your End Boss to a Blue Envelope.
3. "Being right" in Magic deck selection is ultimately about putting yourself in the best possible position to win if you repeated the same tournament several times. The best players often talk about how many times they would make Top 8 if the same tournament were run, say, ten times, a hundred times. A great example would be Raphael Levy in Honolulu. Raph ended that tournament with only two match points... His Owling Mine didn't have the same smooth path as, say, Antoine Ruel's. Was Owling Mine a bad choice? In fact it was a very strong choice, outperforming almost every deck on ratio, and Levy just got a raw deal. Unless he spontaneously invented Heezy Street, it is not likely he could have made a better choice with the options he had available to him... which is probably small consolation, given that he is one of the most talented Constructed players on the planet and finished Honolulu with two match points.
4. We are all on the same page (I'd guess) in that we want good sideboards; the rest of this article will be about how to get there... But the question is, what makes for a good sideboard? The answer is simple: your sideboard is there to give you (or help you maintain) positive expectation in as many relevant matchups as possible, or at least to the point of reason. At some point generating positive expectation in a matchup might require so much sideboard space that it drowns out your ability to do anything else with your sideboard; you have to make a decision. Is that matchup important enough to you to use, say, 13/15? Sometimes it is! Sometimes you will settle for a coin flip. You can have your reasons... But make sure they are based on something.
Okay, actual process...
The single most important thing you need to understand when assembling your sideboard is how much room you actually need to win your expected matchups. Your sideboard is a precious thing, and each card in your sideboard is four times as important as any card in your main deck (4.0666... if you are Jon Becker). This will dictate the whole shape of your sideboard! Notice all the little things you need to know in order to make the correct decisions here... For example, you need to know what your expected matchups are, the important ones, and in particular the ones that will occur at the end of your tournament; you can't possibly fulfil your expectations if you can't last to the end of the day. You need to know what your Game 1 percentages are in the important matchups, as these will tell you all kinds of things like whether you are set on playing an unplayable deck.
R/G Beats — GP: Milwaukee Top 64
I love and hate this deck. I love it because it's awesome. I hate it because I was playing Burger King for Top 16 of Grand Prix: Milwaukee after sliding two or three matches and out of Top 8 contention; I stabilized on about twelve, hiding behind two Jungle Barriers, with a stack of Elephant tokens on my side, having killed every creature on his side but two worthless 0/1 Birds of Paradise with Flametongue Kavus, content in the knowledge that he didn't have nearly enough burn to kill me... And then he drew Reckless Charge and did me six, chumped with his other Birds... And then he drew another Reckless Charge and took 16th (we both had gigantic Constructed ratings, and whoever won was going to claim a PT slot not on the pass down but on ranking).
I showed the deck to Zvi, and we decided it was the best possible deck against B/U Psychatog, which was a dominant deck of the era. Look at the sideboard. What exactly would Kowal be siding in against, say, ZevAtog? Fires of Yavimaya, maybe (Zvi had already deposited the fourth Yavimaya Barbarian where he belonged)? In fact, Zvi took the opportunity to showcase not only his sideboarding skills but his gamesmanship; he would deliberately look at his sideboard, decided that he didn't want anything, and then go on to win Game 2 like he had just won Game 1.
Lesser designers (who probably don't have Zvi's higher education in economics) might devote sideboard space to Psychatog decks... for... just because there was such a thing as a Psychatog deck and they expected to maybe have to play against one. You see these errors in sideboards all the time, particularly with reactive cards. People add Disenchant or Sudden Shock or whatever reactive card just because they feel like they need to remove something even when it is not a good use of resources to do so and get paid off (or at least don't get stung too badly) because they inevitably can find some measure of interaction.
Why is it wrong, though? The answer is simple opportunity cost. If you use two slots for this, you don't have those two slots when that comes a calling. There may a matchup that requires a deep commitment from your sideboard to become winnable; when you start slashing off your resources because you feel like you need to be able to kill some kind of permanent that might come up in the abstract, you remove your ability to flexibly approach the most challenging opponents.
Most of the best decks - the best decks, not the cute ones, or the "innovative" ones, or the ones that seem interesting and you want to go out and try - are all fours and ones. The reason for this is simple: Magic is a game where the following things are the most important: 1) cost, 2) power, 3) speed, and 4) consistency. Cost is the single most important factor in making a card choice decision (there is a reason people say that "Magic is largely a game of two-drops"); the best cards in the game tend to find an intersection at points one and two, and the best decks combine the first three, if not the first four (look at Trix, the best deck ever: mostly cheap cards, largely aggressively envelope-pushing cards, wins on the fourth turn after demolishing the other guy's hand, drawing an extra 14, and gaining 20 life, sometimes sooner). Is Shivan Dragon a more impressive creature than Wild Mongrel? Once it is in play, sure. Make Shivan Dragon cost two mana, and almost everyone will play it over Wild Mongrel. However, Wild Mongrel costs two, and therefore gets to be the best two-drop ever... Until you get to Arcbound Ravager and now Tarmogoyf. Absolute cost is so important that B/U/R/W decks will splash for Tarmogoyf or some similar. Speed, both offensive and defensive, is very closely tied with cost. I recently wrote an article on the Mother Ship about the heretofore undiscovered topic of defensive deck speed. Have you ever lost a game to creatures with two Wraths in your hand? How about a game where you finished with a pair of permission spells in hand? Very likely either your resources didn't come out or you made a mistake on turn 2 that proved fatal; in either case you likely failed on speed. Most tournament players can grasp at least the basics of card economy but fail to see the importance of sequencing in the first three turns of the game. I had to walk away from Steve Sadin's match against Brian Lynch at the NEC last year. Sadin should have won the U/W-on-U/G/W matchup but he didn't see that he had to Spell Snare Lynch's Signet on turn 2. It was a subtle sequencing error that I knew was going to cost him four turns later. The problem was that both guys were going draw-go but Steve was eventually going to have to play Azorius Chancery, which would put him in discard mode. He would instinctively play Azorius Guildmage, losing mana on his own turn, and might accidentally find himself in a permission fight where he could not win, even if he "won." This is exactly what happened and with Steve's mana tapped, Lynch was able to pull off a Tidings from which Steve could never recover. The game was lost on turn 2. Lynch, if I recall, finished second after big ninja Lundquist.
This leaves us with consistency. Jon once told me that one of the great misconceptions about him was that he was the opposite of Dave Price. They saw Jon as Blue and Dave as Red, Jon as control and Dave as fire, and were not able to reconcile these opposites. In fact, Jon maintained that he was the same as Dave... Both of them favored consistency over all else, stuffing their decks with fours. Why would they do this?
One thing you have to understand is that the core of the strategic game of Magic is painfully simple. Go back to the very first point in the preamble to this article. You want to be playing the deck that gives you the best chance of winning the tournament. Very often that is "the best deck." Strategically the goal in Magic, and in any strategy game, is to make every game the same. When you have the best deck, you generally have the best strategy, the Elder Wand, the strategy to vanquish other strategies. You want that strategy to come out, game in and game out. You maximize the chance of this happening by making your deck as consistent as possible. At U.S. Nationals I was asked to critique a Standard Assault / Loam deck by a player who had lots of awesome cards but wasn't able to perform in the 8-Mans. "How many Loams?" I asked. "How many Assaults?" The answer was three and three. I asked the corollary question of how often he won when he got the two cards online together (not surprisingly "almost always")... Yet he was not winning almost always. The first step in fixing his deck was just to play the maximum number of his best cards.
Almost always when you find a deck that is not predominantly fours and ones, it is in some wise terrible. You can play three Keigas, sure, because that card is a Legend and a six drop (you might make certain assumptions about drawing one by the time you hit six mana) but what about, say, three Thawing Glaciers? Three Remands? My eyes are bleeding just thinking about some numbers you see. If stray twos and threes are encroaching on the best cards, pulling them to twos and threes, you can usually make the assumption that this deck does not have the best strategy*. Think back to Lucas Glavin's GP: Boston Cephalid Brunch deck. Lucas did not have the best strategy. He had a very good strategy, two in fact. He used one strategy to beat a branch of decks that were vulnerable to it, and the other strategy to race the decks that were not vulnerable to his primary strategy. Note that he lost in the finals to an Aluren deck that could easily laugh off his primary (Life) strategy... The addition of the second strategy was what allowed his deck to remain competitive.
In the alternative, the man behind the compromised best strategy deck probably just wants to give his opponents a handicap; I don't know. When Justin Polin won a PTQ with B/U Trix I asked him about his numbers, and was puzzled by his four - rather than the commonly accepted three - Donates. He said that MikeyP had told him that Trix is the best, and you really only get in trouble when you start diluting the Trix for more colors and fancy window dressings, and if you just play four copies of your core combo, and four copies of your broken mana and card drawing, you have the best chance of winning a PTQ. Sound advice. Did I mention he won? Sometimes a player will tell you he wants to be able to do something else and is willing to shave so he can... Either he doesn't have the best strategy, or the metagame has changed in such a way that he has to defend his strategy from a different angle, which might be an acceptable compromise, but is almost never right** in the abstract.
The other reason you want to play fours is very simple: You want to maximize the speed and consistency of drawing your most powerful, aggressively costed ("best"), cards as early in the game as possible. One thing that is neat to do is to tally your wins not just based on who goes first, but on whether or not you had a particular card in your opening hand (remember The Innovator's article on Mishra's Workshop?). Over many games, this can teach you better mulligan decisions. If you're not brave like me (I'm not), you can at least make your deck as close to right as possible. Some decks, like Kuroda-style Red, played like completely different - supercharged - decks based on whether or not they had Sensei's Divining Top or Culling Scales in the opener. Other decks come out like Harvey Feirstein in a hot pink unitard if they have Thawing Glaciers in the opener. You know how your deck "feels" by playing it, but by correlating your wins directly with the cards in your opening hand during the vital first few turns of the game, you can really open your eyes and start thinking - and building - economically rather than superstitiously.
This is a long introduction to a relatively short sideboarding skill: the swap.
Here is another Burger King deck, the one I used to win New York States 2006. Like most Angel decks, This Girl was good at beating up on aggressive decks. However, the first time I hit a Zoo, he won the flip, and went Watchwolf-into-Moldervine Cloak while I was tapped for a Signet! I had Demonfire (a medium good but unremarkable card against beatdown); the swap of Demonfire-for-Repeal was fairly simple. I got the next two and eventually advanced to Top 8.
Against Zoo, Demonfire is too expensive. The cost of killing a Watchwolf - even unenhanced - is the same as Wrath of God or Lightning Angel, both of which are better in-context than a straight one-for-one, at least for the purpose of answering a Watchwolf.
This Girl has Strategy Superiority. It probably had the best plan in the room, certainly in that one matchup. The cantrip Repeal helps get This Girl to a living end game where its strategy can take over; straight up, it is obviously a more mana efficient answer to most threats than Demonfire.
The Demonfire swap is not a simple one and you might have missed it. There are much more obvious swaps, like taking out main deck creature elimination versus Prison in favor of (more) artifact hate. Generally you can't be afraid of removing one-for-one attrition cards unless they are trained on the fulcrum of a matchup; the only thing to worry about in cases like these is that you may be opening yourself up for death by transformation.
Swaps of any size (but particularly four-ofs) are important because they give structure to your sideboard. While the goal of a well-wrought sideboard is to be able to achieve positive expectation in as many of your expected matchups as possible, the most common form of the best sideboards is 4/4/4/3. Why? Because no matter what path you take to tuning your sideboard - tweak, swaps, repositioning, transformation, a smattering of trashy unaligned one-for-one reactive spells - sideboards are fifteen cards, and they follow the same rules as building main decks for selecting the best cards (only four times as tightly), and you can't play four of the last card, too. We know from examples like Kowal's G/R fast deck that sometimes you don't need to devote any space to certain matchups where you are ahead... But when you're behind, how do you know how much room to devote to a matchup? The answer is EV equations.
You can often see how good a sideboard is by how devoted it is to particular matchups in ways that can only be arrived at by tireless playtesting feeding reliable EV equations. For example, say you have a deck that seems very good but is a dog against one popular deck. How do you approach any of the following:
1. A 10% matchup?
2. A 30% matchup?
3. A 40% matchup?
A more specific question is how good can you make Game 2?
If you can only get your matchup to 50% with sideboarding, that means that you are going to miss Game 3 half the time, regardless of whether you have a 10, 30, or 40 percent Game 1.
What if you get to 60%? Is that good enough? Don't you have one bad game and two good games? Aren't you favored to win? Yes and no. You are probably favored to win (3)... But not by much (you are about a 55/45 favorite). This may seem obvious, but people simply don't play like they know it: You need bigger and bigger post-sideboarded percentages to win the lower and lower matchups.
You might be able to skate by on a 60% sideboarded game in (3), but what about (2), just one game in ten different in Game 1?
With a 60% sideboarded matchup, you win the match outright 18% (.3, your likelihood of winning Game 1 * .6, your likelihood of winning Game 2) of the time after Game 2, and you lose outright 28% (.7, your likelihood of losing Game 1 * .4, your likelihood of losing Game 2) of the time.
The remaining 54% ((.7*.6) + (.3*.4), also conveniently (1-(.18 + .28))) of the time, you go to Game 3. Once in Game 3, you win 60% of the time.
Your chances of winning are .18 + (.54 * .6), or 50.4%
Your chances of losing are .28 + (.54 * .4), or 49.6%
Basically, you are left with a coin flip. Is 50.4% greater than 49.6%? Sure, but not by much. Subtly if you are a dog in game one, you don't want common coin flip matchups because you are 60% likely to be going second in the deciding game***. Depending on how much room in your sideboard it takes to get to a 60% matchup, it might not be worth going down this path at all. For certain a piddly 60% sideboarded matchup will not correct 10% Game 1 expectation; you are only a 41% favorite with two favorable games, and operationally probably less than that as you are 90% likely to be going second in the deciding game.
Playtesting Game 1 will tell you how bad of a dog you are, give you a flag as to what you should be targeting for your sideboarded matchup, and by corollary, how much you need your sideboard. For a 40% matchup you probably want to target 7/10 working in the preferred environment of real number ten game sets, because winning is winning and drawing is not winning instead, and the goal is to win the tournament. Anything less than 7/10 in real test games is, as we said, insufficient, but at a 7/10 in sideboarded games, you will win the matchup about 2/3 of the time. In late playtesting, I often go "into the tank" and just run hundreds of solitaire Apprentice games against myself, playing only sideboarded games against known configurations until I have come up with something that can achieve the necessary 7/10 or even 8/10 (or say hold in the 5/10 or 6/10 range for good matchups against a lot of known resistance), and only then I bring my ideas back to my playtest partners to try to either pass on the knowledge or validate the data.
I have included a quick and dirty cheat sheet for sideboarded expectation. The math technically works, but it is not 100% accurate because matchups change based on who wins the flip, and sideboarding percentages differ between Games 2 and 3 because you will not be going first in both of them***... But this is just a quick and dirty and should still be helpful to you. I firmly believe that you can't hit your goal if you don't have a target, and up until now, I'm guessing you never even thought about this. I marked the points where a bad matchup crosses into the 60+% range; you'll note cute stuff like no matter how bad your Game 1 is, you are always 100% to win if you have a 100% sideboarded matchup.
After you've figured out what percentage you are going after, the next question is how much space you have to devote to your sideboard in order to win whatever terrible matchup is bothering you. One thing that I can't stress enough is that you can get a 60% (and often better) matchup out of whatever deck you have in front of you against whatever deck you have to beat, so long as your deck is not unplayable; you just have to want it. If you can't figure out a way to beat the dangerous Matchup X, provided it is a big enough part of the metagame that you know you have to beat it in order to win the tournament (remember your goal) then your deck is unplayable. That's it. Unplayable. Imagine you are preparing for a Time Spiral Block Constructed PTQ and you have what you think is a sick deck, good against almost everything... Only you are a dog to B/U-based Teachings with Coalition Relic. Well, that isn't the worst thing in the world. You can probably take a loss. However if you think you can actually win an 8+ round PTQ without facing this deck you are kidding yourself. You will probably have to face it twice over 11 rounds, meaning that even if you beat everyone else, you still have to weather the bugbear at least once. Do you see where this is going? Being a dog isn't the worst thing in the world, but having no plan at all, shrugging and saying you are just going to lose if you play against B/U - which, keep in mind, is what an alarming number of tournament players say - is a violation of the first rule.
Now say you are only a slight dog - say 40% - if you can get to a 50% matchup sideboarded, you are almost within coin flip range against B/U. This is something worth knowing! If you have to play B/U twice, that means that it's not a pipe dream that you can beat it once. What if you're only 20% to win Game 2? Your target is 80% sideboarded, remember, and 60% doesn't even give you a fair coin flip. 80% SIDEBOARDED. Wow. Can you do it? If you can accomplish that, you now get to lose a different matchup. You're going to beat B/U both times.
Going from 20% to 80% is not easy. Very likely it is not going to come via a simple swap where you draw a class of halfway relevant card instead of a more-or-less dead card. You've got to work, but sometimes you come up with the engine for The Greatest Deck That Never Was (and sometimes it even works!). For this you have to go into the tank and think out of the box. Not everyone can do it. Now, at least, you know that is what you have to do.
The least costly way to accomplish a radical positive delta in matchup percentage is by figuring out what makes one kind of deck beat your deck and then blank that entire class of cards. Your goal is to do this as consistently, using the smallest number of sideboard slots, as possible. Blanking the source of the opponent's Strategy Superiority simultaneously generates virtual card advantage and lets you feel smarter than your opponent***. The only problem with this is that it doesn't necessarily work if the opponent sees it coming. The best example I can think of is Kuroda-style Red, where Josh was a dog to Blue but an overwhelming favorite after boards... Except when Neil decided to go fatty boombatties against his burn plan, which no other Blue deck was clever enough to do.
Repositioning is another tool. For example, say your combo/control hybrid deck is one of the best decks, but is not great against aggression with vigilance and untapping, and has no long ball. You use seven of your sideboard cards and bam! You're board control, which is a problem for the fire-free beatdown opponent. You still have eight cards to work with. Granted, six of them are probably more anti-beatdown to compliment your repositioning, but your deck was good enough main - and against everything else - that you can afford it. I think you get the picture.
To make what is probably now a 5,000 word article a little simpler, keep in mind your goals for your sideboard. Goals are important because we almost always fall short. We are human and imperfect. If we don't shoot for the best outcomes we can't be surprised when we fall short of even our compromised targets. It's not important to sideboard in matchups that you are going to win without a sideboard, or to sideboard against decks that are not going to show up. Sometimes it is fine to sideboard not at all when you are such a long dog that your opportunity cost would be every other matchup, provided you can still win the tournament (the opponent is rare, or rare enough that you won't ever play him or he can be your otherwise sturdy deck's lone loss). For the rest, use the cheat sheet. You will have some overlap. For example, I sided three Culling Suns in my B/W Bats deck in Charleston just for Raphael Levy's U/G deck on Day 1, but they came in handy against Jon Sonne's token deck on Day 2, too; I didn't need them against Jon, but it's always nice to get some positive delta.
In sum, structure your sideboard as best you can, targeting the percentages you need to hit, based on the popularity of opposing decks, and how much you need to beat them. For instance, I marked the points where a matchup goes to about 60% on the cheat sheet, but say your format is Mirrodin Block Constructed, late meta. You can't win a PTQ with a 60% matchup against Ravager. You're just going to go 4-2 and miss Top 8. 60% isn't good enough in a format that narrow. In the Tier 2 metagames, you have to play the percentages faster and looser. I started this article by invoking a conversation I had with Jonny Magic about what makes for a competent deck decision. He is the first person to acknowledge that the distance between best deck and lowest deck in the viable range is a lot closer than when Tinker was an "automatic" idiot test (but that doesn't mean that there isn't still a best choice). It just means that you have to settle, in some cases, for lower good percentages, and especially with the disappearances of Chokes, Perishes, and Meltdowns from the common arsenal, you can't necessarily rely on your sideboard topdecks so much as use them to craft your best available strategy.
* There are exceptions, of course... Hybrid decks, redundancies in bullet decks, redundancies over fours, stuff like that.
** Remember what "right" means.
*** es muy importante