1. The first step in sideboarding is determining what to side out.

Playing against a friend of mine in a local event, I slammed down my sideboard for game two and said "I have thirteen cards for you".

His response: "When you board thirteen cards in, you board thirteen cards out".

As inane as this statement may seem, he had an important point. If you want a card in your deck post-board and have to swap it out, odds are you are taking out the wrong card or are boarding too many cards in.

The first step in constructing a sideboard should be going through your main deck and figuring out what cards are bad in each matchup. Write down the number of bad cards and mediocre cards in each matchup and then work on the sideboard. Starting with the most important matchup add cards to cover the minimum number of cards you want to take out (the number of bad ones). Once you are done, try to consolidate the list down to fifteen by increasing the number of overlapping cards. These cards don't have to be great in any specific matchup, but just good in multiple ones. For example, my Jund lists tend to have one Master of the Wild Hunt. While the card isn't exceptional anywhere, it is good against any creature deck and even some control decks. A specific card would be better for each matchup, but the Master makes it so I can board out all the bad cards in each matchup. After this, if you still have too more cards for a matchup than you have mediocre or bad cards to take out, figure out which ones are the worst and leave them out unless they are actually better than a card still in your deck.

You can also make slight changes to the main deck to accommodate issues with this framework as well. I played three Maelstrom Pulse main in Jund only because I couldn't make a board with less than 17 cards that worked otherwise.

If you can't get a sideboard to work with the above guidelines exactly how much of an issue it is often depends on what kind of deck you are playing. Control decks tend to have more card draw, allowing them to overcome a dead card in their opening seven a lot easier. Beatdown decks tend to have more issues, but due to their overall consistency can function perfectly fine down a card if absolutely necessary. Midrange decks tend to be based on gaining incremental card advantage by pairing their threats and answers up properly, making being effectively down a card extremely detrimental. Combo and other linear decks tend to not have dead cards in them to begin with (this will be covered later).

2. Post-board games don't have to be played the same as pre-board games.

In the Jund against Mythic matchup, game one is rough for Jund but games two and three Jund is a heavy favorite. Jund only boards four to six cards and none of them are specific hate for the deck. Why is it that the matchup swings so heavily?

The answer is that game one Jund has to be the beat down. Their limited supply of removal will run out quickly, leaving them dead to Mythic's monsters if they are still alive. At Grand Prix DC my game one wins against Mythic were based mainly on killing them on turn 5 with Garruk's Overrun.

Postboard Jund completely changes to be the control. You board out the cheap creatures other than Putrid Leech for Doom Blades and non-Cascadeable spells. With the added removal and the greatly increased odds of cascading into one Jund can simply kill each of Mythic's threats as it hits and win at its own pace.

While this example is extreme due to how Cascade works it is seen elsewhere. The most common is when one player boards in an efficient hate card that recentralizes the matchup.

This also happens when two similar decks play. One of the ways to break the symmetry of a mirror or pseudo mirror is board into a different game plan. This is even better if your deck loses the mirror fight game one. A good example of this is the Merfolk versus Zoo matchup in Legacy. Game one you lose the aggression fight. Your creatures and removal are both outclassed; hence the reason the matchup was traditionally viewed as poor. While Saito's plan of boarding in specific hate cards is very reasonable, another plan that works is morphing into the control deck. With the same number of board slots you can bring in 4 Submerge, 4 Hydroblast, 2 Engineered Explosives, and 2 Vedalken Shackles and fight Zoo on a different level.

3. Know when to concede a matchup.

In the above case, I showed a Merfolk deck devoting twelve cards for one matchup. While in that case it might be warranted due to Zoo's popularity, the versatility of most of the cards listed, and the fact Merfolk tends to be favored against non-aggressive decks, sometimes it just isn't worth it. For example, how many cards do you have to devote out of Zoo to beat Storm combo? It is a similar number but Storm is a very small portion of the metagame, the decks those cards are also good against are similarly scarce, and Zoo has a lot of other decks that it has to combat. If you go short and only board a couple cards you will probably still lose the matchup anyways.

This is not to say you shouldn't board incidental cards you already have if they are good. Most Zoo lists have Red Elemental Blasts in the board, which can be used to slow down a Storm deck and steal a win. The issue is that unless you somehow have extra sideboard slots, a couple specific hate cards to bring a matchup from 10% to 25% just aren't worth it.

One last comment on the subject: If conceding the matchup isn't a viable option for the event and fighting it causes you to lose viability against the field, you may want to consider switching decks. That said, losing to even the most popular deck is reasonable depending on your strength against the rest of the metagame and how good that deck is against the field.

4. Aggressive decks should rarely have situational answers against control decks.

You are playing a Zoo deck in Legacy and want a good answer to Counterbalance. The first card that typically comes to mind is Krosan Grip. In Lorwyn Block and the subsequent standard format, Kithkin players often went to Wispmare as an answer to Faeries.

Consider what the card actually does in that matchup. Your goal as the beatdown deck is to win the game before your opponent sets up a stable board state. Almost every control deck should be able to accomplish this without the use of whatever specific card your answer lines up with, meaning that in a fair percentage of games you have a blank in your deck. When the main way for the aggressive deck to win is have more threats than the control deck has answers, blanks are very bad.

The exceptions to this are when the aggressive deck is completely bricked by the card it is trying to answer and unable to outrace it hitting the board (ie. Energy Flux against Affinity) or the answer is not situational. A good example of the latter is the fact Zoo decks can get away with running Path to Exile as there are so many creatures that it is only a blank in very specific matchups.

If the answer is also something that actually can kill your opponent, like a Qasali Pridemage or Volcanic Fallout, it suddenly becomes acceptable as a mediocre threat is still a threat.

5. Linear decks should sideboard very sparingly.

This mainly applies to combo decks but some other decks such as Affinity have fallen under this category. Tribal decks tend to not count as linear decks as their cards often support a range of effects that get better or worse depending on matchup. This is also not considering a full transformational sideboard as those are rarely viable.

First off, when sideboarding with linear decks you worry about two things instead of matchups: faster linear decks and specific hate cards. Linear decks also tend to be so full that they rarely have room for more than a few cards to protect their core.

The reason linear decks are so powerful is that all of their cards work together towards making an overpowered effect. While often the effects that linear decks need to answer are often extremely hateful and prevent them from functioning at all, boarding in extra cards can lead to the same issue. Linear decks need to assemble a certain number of pieces to reach their end game and by reducing the density of a given piece you start to remove your deck's ability to win. If you slow your deck down or lower the consistency you can do as much damage as a single hate card. The best example of this is Charbelcher combo. The deck has no way to find extra cards and has to invest all its resources into one threat. By cutting replacing threats and Rituals with answers the deck loses the ability to even present a threat. An opening seven with a Xantid Swarm is very similar to a six card hand game one.

The ways to sideboard with a linear deck are to either upgrade an existing portion of the core to one that helps fight the hate or trim the excess copies of specific pieces in order to fit a couple counter-hate cards. At Grand Prix Columbus I would typically sideboard out 2 Preordain and 1 Lotus Petal as those were the worst cards in the most redundant slots. Against decks with creatures I would want to hit but no counters two Duresses would become Thoughtseizes. Dredge decks in Legacy board out one of a bunch of different cards in order to maintain their ability to perform any of their functions. They also switch around reanimation targets to things like Blazing Archon or Ancestor's Chosen for aggressive matchups where Bridge from Below is inactive or Terastodon and Woodfall Primus against cards like Moat or Elephant Grass that otherwise lock them out.

This also puts linear decks into a position where they can pack answers for a large variety of problems. The main issue is just finding them when they are needed.

6. Most specific hate cards need backup.

Unless an answer completely shuts down the opponent's ability to play their cards, it still needs a game plan to back it up. Your opponent will know about strong answers to their strategy and will come prepared with outs to them. Note that this mainly applies to older formats as Wizards has moved away from harsh hate cards and strategies that require them in recent years.

The easiest form of backup is a solid clock. The reason Gaddock Teeg and similar cards are so effective out of Zoo is that while the other deck is trying to solve the issue they pose, the Zoo deck just kills them. Without a good clock, even immense amounts of hate can fall.

Another way to backup hate cards is with secondary disruption. Most commonly this is counterspells or Duress effects used to stop their countermeasures. If a control deck backs Leyline of Sanctity with counters odds are most combo decks won't be able to win. One less direct example is Humility in addition to Moat out of Landstill decks against Goblins. Moat puts the Goblins deck on one out of Siege-Gang Commander. While it doesn't take long for the Goblins deck to get to lethal with a Siege-Gang after a Moat hits, the control deck can set up the other half of the lock faster.

Hate cards can also be backed up by other hate so long as the combination requires multiple answers. This is a rare scenario as it often requires a large amount of sideboard space to reliably set up.

Some cards dodge this issue by answering the hate against them. The most common example of this is Chalice of the Void stopping the bounce or artifact removal directed towards it. Manabarbs also does this to some extent in Standard by punishing the answer to it (though the card still needs a clock to function).

7. Unexpected answers are easy ways to steal games.

If your opponent comes prepared for a given solution to their strategy it will take an additional line of action to finish them. If they aren't prepared your one answer might be enough to immediately end the game.

The best application of this is against Dredge. People often follow a very strict guideline for what cards they want to board in against them. In Extended it used to be Affinity always boarded in Tormod's Crypt, a deck with four sideboard slots boarded in Leyline of the Void, Domain Zoo boarded in Yixilid Jailer, and so on. As a result, Dredge had perfect information while sideboarding and could bring in narrow answers for each hate card. If a deck brought in something different, such as Affinity bringing in Leyline, it would usually auto-win game two. The Dredge deck could bring in a generic answer for game two, but the generic answer was much worse against most of the hate than the best option. While Dredge is an extreme example this can be applied in many other situations. I lost a match at Nationals because my opponent who was Mono-Green game one cast a Manabarbs off Lotus Cobra mana game two and I had no answer to it.

If you play the same trick for long enough people will start to figure it out, but it should work long enough to give you an edge for one event. If after that even your list is highly publicized and people know about what you did, odds are you did well enough already.

People will also be able to adjust between games if you win game two with the card or plan and force a game three. Sometimes it is correct to swap out the plan if you notice they appear to be boarding to deal with it, like how Blue-White early in the season would board Baneslayer Angels in and out to either have a threat against their opponent's lack of answers or leave their opponents with a handful of dead removal.