Procedural Playtesting

There are a huge string of events coming up: Grand Prix, Pro Tours, StarCityGames.com Opens (including the Invitational in Indianapolis). How do you prepare for all of these formats? Ari gives you a step-by-step breakdown.

Recently, the majority of my Magic-related time has been spent preparing for the upcoming string of high-level events. Grand Prix Providence is the
lead off, followed by the StarCityGames.com $50,000 Invitational in Indianapolis, after which I drive straight home and hop on a flight to Pro Tour

In testing for these events, I’ve found myself really looking at how to make my efforts matter and how to actually make the best decks for these
events. Without a dedicated plan, most of the time, things haven’t really gotten anywhere. Given the limited time span to handle all of these, I sat
down and thought out what actually needed to occur to make progress based on how I approached previous events I did well at. While I’m sure a lot of
the things I came up with are standard advice, as far as I know, they’re scattered and disjointed in a process where sequencing matters a lot. I’m also
sure a lot of players are like myself and would feel a lot more comfortable knowing they had a solid plan before actually starting anything.

So, based on my experience, here’s what I am currently using to try and maximize the value of my playtesting.

Step Zero: Take your time.

This doesn’t mean investing a ton of hours into testing. Expect it to take a definite time commitment to produce relevant results, but more importantly
pace yourself. Putting forty hours of testing into a single week before the event is probably not as effective as pacing forty hours over a few weeks.
There are diminishing returns past the first few hours as people get distracted, as well as in the long term as ideas tend to stagnate in people’s
minds over a long testing session without breaks. There are also the non-testing hours between sessions, where random inspiration can and will strike.

The only moderate exception here is Pro Tours, where the gains in the last week are greatly boosted by people from everywhere physically testing
together at the site and ideas mixing. Even then, you still almost have to show up with something to be remotely prepared to play correctly in the

Step One: Pick a deck.

There are two main ways to do this. The first is to just pick a deck you think you would want to play in the dark. This really only works in
established formats, as without information or initial lists, making any reasonable decision here is near impossible.

The second is to just build some litmus test deck. In new formats, these are usually just linear aggro or aggro-control decks, but in already
established formats, they can be a strong, versatile control deck or a popular combo deck. Ideally, the deck will be something that punishes bad or
overly rough decks, forcing evolution and optimization in order to compete.

Step Two: Build more decks and battle.

If the format is a defined one and you are championing a deck, start by building a gauntlet. This doesn’t mean just building the most popular decks,
but instead a spectrum of strategies. Battling against four different beatdown decks doesn’t actually help you understand what your deck needs unless
they’re all presenting different problems to answer. Similarly, try to make sure each deck has a unique plan of attack rather than several overlapping
ones. For example, I generally prefer to test against Junk and Merfolk in Legacy than Team America, as it gives me a better idea of how strong each
individual disruptive card is against the deck and how the deck should react to each one. It’s far easier to piece together strategies than figure
everything out at once. Also try to avoid non-interactive matchups. If both players are completely goldfishing, no actual experience or information is
being gained. Lopsided matchups are still fine, but if the only thing you can say is “I can’t stop anything they do, and I die,” odds are
that was something you already knew. Ideally, you should have around four or five different strategies to play against. If you’re planning on going to
Grand Prix Providence, a good spectrum would be Merfolk (Force of Will aggro-control), Junk (discard-based attrition with powerful single threats), Zoo
(dedicated beatdown), Storm (resilient semi-interactive combo), and U/W Counterbalance (pure control).

If you’re just playing a litmus test deck, just build everything you want to try. Play a few games to get a feel for what is going on, then move on.

Most importantly, this step should not take you forever. The goal is not to get exact percentages, but instead just see what happens. Most importantly,
you want to gather information on what cards are working and which aren’t.

Step 3: Evaluation and Evolution

If testing was simply about seeing what current deck won the most, it would be much easier. Instead, things can change.

There are three basic outcomes here. The first is that you are championing a deck, and it beats every one of the basic strategies. Congratulations, you
did it. From here, you first want to identify which cards in your deck aren’t doing anything and cut them. Then proceed to step five.

The second is that a deck has been winning and losing. Stop and figure out why you lost the games you did. Were you drawing too many blanks? Cut some
situational cards or high drops. Did you have mana issues? Consider playing more lands, or rebalance your colors. Did your deck not have enough action?
Consider cutting some of the weaker one-for-ones or clunkier cards for ones that do something. Was one specific matchup that bad? Figure out why, and
consider what you could change based on that.

One easy trap is to just do this for the one deck you are playing the most or very specifically not change the default deck in an unknown format at
all. Simply put, every single deck has to evolve. Not only is a stagnant deck a possibility for missed potential, but odds are if you play against the
deck at the event, that person will have put in the work. There’s no point in beating last month’s list today.

Finally, a deck might just be losing. If the reason is obvious, like the aforementioned mana and curve issues, just fix it and try again. There will be
times where the deck is just fundamentally flawed. If it’s a linear deck that just isn’t coming together, take another look at all the options and
replace the poor cards. Sometimes alternatives just won’t exist to make the deck better or the base concept is fundamentally flawed, and if they don’t,
just bin it. If it’s not, gut the parts that were working and make a new shell.

Case Specific Aside: People really have no clue how hard it is to assemble a two-card combo when you only have four of each piece. If you have a solid
tutor that gets either part, then everything can usually piece itself together, and if you can play copies 5-8 of at least one part, you can get there,
but without these, you have to really dig and likely aren’t looking at winning until at least turn five on average. Sure, you might just randomly have
it 15% of the time, but the other 85% is a different story. If all the pieces are independently good, you can usually get away just grinding along, but
when one or more is a blank, then it starts to become an issue. Maybe the format is such that you don’t need to win until turn five (or later depending
on your mana requirements) or you have room for auxiliary cards that buy just enough time, but really think about this before you start playing
Splinter Twins or something similar.

This is probably the most important step in all of testing. It’s really easy to just miss this step and just keep stacking up new proxied up sixties
based on wild ideas and binning things that don’t work without thinking, but you are far better off just making a queue of new concepts to try and
processing them systematically.

Step Four: Go back to step two.

Even in a previously undefined format, you should now be close to making a gauntlet. Again, aim for a variety of threats, and try to hit every major
point. Brew new ideas based on what you learned from the first cycle. Proceed through step three and back to two again until you have a deck you are
actually impressed with enough to play.

Keep in mind that in an undefined format, at least some of your testing up until the very endgame should be devoted to running steps two and three with
new ideas. Completely locking yourself in early is usually not a very good plan for success, and this ties back in to step zero.

Step Five: Prepare for the worst.

So, you now have a deck that you want to play. It’s tested solidly against a decent field of decks, and you’ve worked out a lot of the initial kinks.
Now what?

Simply put, figure out how you think you can lose and see if you can beat it.

Test against all the different kinds of hate you can think of, pre- and post-board, and see if anything that isn’t just inbred garbage actually beats
you. If so, think about why. Typically, you have three options here.

One is to switch cards to shore up the matchup. In combo, this usually means adding counters to their disruption and finding the point where you
maintain stability while maximizing answers. In aggro, this either means adding some interaction like counterspells or removal or changing to a more
value-based setup to grind out control. In control, this usually just means switching your answer and threat suite a little. Once you find a mix here
that works, go back to a litmus test matchup and see if everything still works out to prevent your deck from getting too far ahead or inbred.

The other is to change the way you play the matchup. The best examples of this I can think of were Jund in Standard and Faeries in Extended playing
against Naya. I started both trying to assume the control role against what was labeled as a beatdown deck, but kept losing. Even when boarding big,
game-ending threats like Wurmcoil Engine out of Faeries, I was finding myself out of answers. Both times, I realized there was no way to actually beat
Naya going long because Vengevine generates unlimited card advantage for them. Instead, I decided to just start attacking and burning my spells to
pressure them. Turns out, this worked both times. Jund simply attacked with bigger creatures faster than Naya could clog the board and actually created
a dominant board presence faster, while Faeries actually had time to do what it usually does and Time Walk them a couple times while attacking for a
ton. Maybe you need to be taking a bunch of damage from their early drops, as you can’t beat their top-end card. Maybe you need to play your guys more
conservatively, as their endgame card isn’t actually that dominant. Try new things and see what happens. One special case of this is when you are the
aggressor, but they rapidly trump you. It might be right to just spin the dice and force them to have the answer to your initial rush. Sure, they have
the Wrath 60% of the time, but the other 40%, you get there, and you weren’t winning if they had the extra time anyway.

Finally, you can play a transformational sideboard. The two usual scenarios for this are when a combo deck is completely trumped (i.e. Storm vs.
Counterbalance) or when a midrange-style deck is out-controlled in the mirror (i.e. Old MBC against Tog or Wake). The default assumption here is you
just jam a bunch of creatures and hope for the best, barring some exceptions where a complete transformation is possible (i.e. Doomsday). This is
usually a last-ditch effort, and you should test it against them re-boarding for it.

Step Six: Fill in the rest.

Now is the endgame, where you start counting out the sideboard. Figure out the number of cards you don’t want from your maindeck in each major matchup
(even at a Pro Tour, by now top tier decks should be emerging), then shape a sideboard to make it work out correctly. Use your previous experience to
figure out what you want in the matchups. There are definitely articles out there to help with this process; I even wrote one a while ago. A good
recent place to start would be Mike Flores
article from last week

From here, things don’t stop. Keep playing, if only to keep the proper decisions to make in specific scenarios fresh in your mind. Try new things;
maybe something will surprise you.

A few last tips:

-Almost always, this is not a one-person process. At bare minimum, you need one set person to test against or a group of random people who aren’t
dedicated to testing but reasonably play the spectrum of decks. Testing against yourself is actually impossible to do correctly, as the loss off hidden
information is something you can’t 100% bias yourself to ignore. You also lock yourself into play patterns that other people might disagree with and
might just be wrong.

-When testing for an unknown format, a good idea is to have two smaller groups split up and basically build the format on their own, then recombine to
see where everyone is at. Looking at what decisions the groups made along the way to come to where they are when they reconvene is one of the best ways
to actually figure out a format, and the different decision trees help increase the variety of decks being tested.

-When it is possible to do so, playing in events is one of the most productive ways to test. It combines the testing against completely different
things aspect of groups coming together while actually applying pressure. It’s easier in testing to miss when a problem costs a couple games out of
twenty or fifty, but when it costs you a match, you remember it.

-To briefly talk about Limited, the two important goals are playing enough to recognize all the tricks and just trying new things. Specifically, look
for cards that aren’t purely bad but don’t have a home, whether due to what people perceive as the format or them just needing a full deck to function.

-This guide is not the end all of testing. It’s just a basic guideline to keep things on track. There are probably things I missed in here that I
either forgot, don’t personally like to do but that work for other people, or just plain don’t realize.

Bonus: The part where I talk about actual decks.

Standard is the format I’ve currently decided to pass up testing, as it will have the most information before the event where it matters and only makes
up half of that event. My first instinct is Caw-Blade is still the best, but you might not actually want Sword of War and Peace, as it doesn’t really
beat or match Gideon or Jace in the mirror, while the mana from Sword of Feast and Famine can give you a definite leg up in those fights. Splinter Twin
looks very slow on the surface, but again, if you accept not actually being able to reliably combo until turn five or six, I wouldn’t fault someone for
playing it. I could also see a less combo-centric build with other value outlets for Splinter Twin to make it less of a blank. As for other decks, the
only things I’m considering now are upgraded versions of the super-aggressive decks like Tempered Steel based on the Phyrexian mana creatures, and even
those are hard to justify against Gideon Jura and Day of Judgment.

Oh, and in case you haven’t heard, Spellskite is very good. It’s a Seal of Turn Aside, only it just straight up eats Lightning Bolts, blocks well, and
carries equipment.

As for Legacy, I have a default line for everyone who asks what I like. I obviously like Storm, but other than that, Merfolk is the best default
choice. It just punishes loose decks very well and isn’t shabby against good ones either. W/U Counterbalance or Landstill is very good if you feel you
can play it well and actually play fast and push your opponent’s play enough to end every round. The MUD Poison deck I
wrote about in my last article is actually very good right now and probably only gets better; a super aggressive Zoo deck with Goblin Guides, Steppe
Lynxes, and lots of Fireblasts is always a fine choice to just bash people. The Sneak Attack + Show and Tell deck gained a lot from Mental Misstep
being a better roadblock than Daze, and if you want to just play an all-around solid deck with good game against actual everything, you want to look to
G/W creatures.

The percentage of double takes the last line has caused so far is at least a majority.

What would make me suggest a deck full of fair cards and no Brainstorms? Simply put, the deck is just all good, and not in the old Rock sense. You have
a high density of powerful threats that also happen to double as answers and card advantage. You even have game against Storm, as you can afford to
play a bunch of hate bears with your main plan being so solid against fair decks.

Mike Antrim, the local player who made me appreciate the deck, has a slightly nontraditional build of this deck that I feel is worth sharing. He has
had a fair amount of success with the deck locally in a metagame that’s actually tougher than the StarCityGames.com Opens in my experience.

This list eschews the recent additions of Green Sun’s Zenith and Fauna Shaman for a more old-school Death and Taxes base. Here are his comments on some
of the choices:

Mother of Runes: This gal wins every attempt at combat in the creature mirrors, makes sure your Equipment lands a hit, makes your big guys punch
through various Tribal walls of blockers, and if you untap with her, locks out a lot of the ways people have to interact with you, namely Swords to
Plowshares but also bounce spells from combo. It takes a lot of work for High Tide or Storm to beat Mother of Runes into Ethersworn Canonist.

Green Sun’s Zenith: The card is not bad, but Aether Vial is much better and requires a high density of actual creatures to work. Mike has not tested
post-Mental Misstep, so things may change to make Vial worse and let this card actually take up slots.

The Equipment suite: Sword of Feast and Famine does nothing here, while Light and Shadow surpasses War and Peace against control, as the Raise Dead
ensures that if they take the first hit, they can’t just rip a piece of removal to get out of it. Control decks also tend to run low on cards against
this deck most games, as they have to heavily rely on trading removal for threats, making the damage less significant, and protection from red doesn’t
actually save anything, as the +2/+2 also protects from Lightning Bolt and Grim Lavamancer. Batterskull has the problem of being wanted against decks
that can just kill your Stoneforge Mystic if you get it, leaving you with a five-drop.

Fauna Shaman plus Vengevine: Mike has not tested this, but just playing a threat every turn is faster than trying to get this engine online. Vengevine
is also not good with Aether Vial.

Tarmogoyf: There were a lot of complaints about this card. It doesn’t actually do anything besides provide a big body, and you have to rely on them to
fuel it most of the time. That said, the card is just so good at fighting creatures and clocking people that it stays.

Serra Avenger: Flying is awesome for pushing through a Sword or just any damage in creature matchups, not to mention the card is good with Aether Vial.
It just provides a good clock on its own for two mana while also being evasive, and no other card really does that.

Condemn in the sideboard: If you play against Cephalid Breakfast, you’ll regret this not being Path to Exile. Otherwise, against creatures, it’s just
better, as it better suits your plans of having inevitability with Knight or Equipment, making life total not matter while also not messing up your
ability to randomly mise them out with Wasteland. I can personally vouch for this card being amazing every time I’ve watched him cast it.

The two cards I didn’t ask about that I wish I had were Natural Order and Mangara of Corondor. I’m almost certain Mangara is just impossible to support
with one Karakas and too clunky or fragile even if you played more, but Natural Order seems almost legitimate. It clashes with the plan of every card
just being good, as you have to run more situational ones and have to play a pure blank, but at the least might be a good sideboard card to force them
to fight a new angle. Regardless, if you don’t want to play Mental Misstep, this deck is a good place to start.

I’ll be back again in two weeks, hopefully with a winner’s report from Grand Prix Providence. If you have questions about G/W, Storm, MUD Poison, or
any other Legacy deck, ask on the forums or shoot me a message on Facebook or Twitter (where I can be found as armlx).