Plans and Playtesting

When playtesting, it’s easy to fall into the “Playing, Not Testing” trap. How can we maximize our testing to ensure an improvement in our game? How does our sideboarding strategy relate to our overall gameplan? In this excellent and enlightening article, Chad offers some sage advice on planning for success.

Let me tell you about a game I lost to a deck running Dromad Purebred.

I was playing the sort of Blue deck that has an insane late game but can occasionally get pushed off the table by an aggressive start. I had a slow draw and my opponent, on the play, came out with a fast Boros rush. He led with Thundersong Trumpeter, Skyknight Legionnaire, and Indentured Oaf. With good cards like those, the fact that he brought out a 1/5 on turn 5 didn’t stop him from getting me down to three life before I established control of the board.

Three life isn’t a lot, but I was actually in pretty good shape. His hand was empty, while mine had good cards — in fact, I had Disembowel in hand for almost any creature he might draw and my board position was just fantastic. I had active Lurking Informant to reduce the chances of him topdecking a win. I also had Train of Thought and Izzet Chronarch in hand, so I was never going to run out of gas.

My plan was straightforward. I would let him draw pretty much anything other than burn while I continued to build up my board and counter-attack. Burn spells would go to the yard.

Then I Lurked and saw Firemane Angel.

My opponent had six mana and would never hit ten before I killed him. The Angel fits nicely into the “bomb” category and even though I knew I was going off plan I decided to mill it. His next draw was Pyromatics and even people who run Purebreds know how to aim three at the dome.

In the kitchen at every party...

My plan gave me an extremely high shot at victory. He would need two direct damage spells in a row in order to draw one, and he probably only had four or five turns to live. By going off plan I changed “two in a row” from “just one, thankyouverymuch.” The odds are still good, but nowhere near as good.

Meanwhile, was the Angel really a bomb? Of course not — it was just a spell that I could handle. In the context of the game it was as much a threat to be milled away as a Signet would be, given an opponent with eight lands in play.

In essence, I diverged from my plan for that specific situation and went with the ultra-basic Informant plan — mill good cards. That plan is correct given a neutral board position and/or topdecking battle, but awful in the situation I was actually in.

Good beats…but what does this have to do with Team Constructed?

Let’s say you’re running Wild Vore and you’re on the play. Your opening hand is Island, Mountain, Sleight of Hand, Eye of Nowhere, Stone Rain, Compulsive Research, Magnivore. You keep, lead with Sleight of Hand and snag Steam Vents. You’re all ready to Eye your opponent’s first land (assuming he has no play) and to follow with Stone Rain followed either by Compulsive Research or possibly a 3/3 Magnivore. At this point your plan is the generic one for your deck — suppress mana, burn through sorceries, and beat face with Magnivore.

Then your opponent leads with Stomping Grounds, Kird Ape. Suddenly you have multiple potential plans:

1) Stick with the Eye of Nowhere your land, Stone Rain your land plan.
2) Eye of Nowhere the first land but then play Compulsive Research on turn 3 rather than Stone Rain, aiming to maximize the turn four Magnivore by discarding two Sorceries.
3) Eye of Nowhere the Kird Ape, followed by Compulsive Research, aiming to maximize the turn 4 Magnivore.
4) Play Steam Vents tapped, followed by Compulsive Research, intending to use Eye of Nowhere for maximum power (e.g. on a creature with Moldervine Cloak)

Plan 1 has the advantage that, assuming you draw a fourth land, your opponent will only have two mana up when your Magnivore hits. If they are playing a conventional Gruul build they may not be able to burn it out, in which case you can untap, play Compulsive Research and hopefully have a 5/5. (Alternately, of course, you could risk the delay of a turn to play Compulsive Research on turn 4 and drop a bigger Magnivore on turn 5.)

Plan 2 doesn’t try to prevent your opponent from playing Char but has a much better chance of not caring. Sleight, Eye, Compulsive, plus two discarded sorceries, puts your Magnivore at 5/5, which puts it out of range of a single burn spell.

Plan 3 does the most to preserve your life total; if he’s playing Gruul he’s unlikely to be able to muster more than three points of power on turn 2, and may well do nothing more dramatic than lay a Sophisticate or replay the Ape. That may slow his offense down more than bouncing his land would have.

Plan 4 doesn’t put a second sorcery in the yard, and doesn’t disrupt his development. It does, however, spare you two life from the Vents and gives you an out to Moldervine Cloak.

Test of underwear absorbtion

Which of these plans is better? You won’t really know until you test.

All too often when I see people testing they just play out the matches. They see who wins how often and draw a conclusion about how favorable or unfavorable the matchup is. They choose the deck they wish to play and/or modify their sideboard accordingly. That’s all well and good, but the principal purpose of testing is to determine what is the correct plan (or set of plans) in a matchup.

Let’s revisit the example above, with Bob playing Wild Vore. Let’s assume, hypothetically, that Bob’s instinct is plan 1, but that plan 3 is actually the optimal plan. The likely result is that Bob will playtest a bunch of games with his plan and fare badly. He’ll go to 18 and bounce a land; his opponent will attack him down to 16 and play a Frenzied Goblin. Bob will destroy the land; his opponent will attack him down to 13. Bob puts out a Magnivore; his opponent hits him down to 10 (paying R to prevent blocks) and plays a Rusalka and then burns Bob out after Magnivore hits.

In the end, Bob concludes that the matchup is horrible — after all, even with a perfect start of Sleight, Eye, Stone Rain, Magnivore, he can lose.

Meanwhile, a more flexible approach would see Bob trying out other plans. Sometimes he may come upon a new plan almost accidentally, simply because he’s exploring his options in a much more open-minded way than he could if the game result mattered. At some point, desperate for land, he dumps two perfectly good sorceries to Compulsive Research and as a result plays out a Magnivore that Gruul can’t kill. He tries that again when he doesn’t have to, and suddenly he’s stabilizing at a high enough life total that his opponent isn’t constantly finishing him off with burn.

Finding the right plan through testing may not turn a bad matchup into a good one, but it will certainly improve it — and at the PTQ level a single match win is the difference between making the elimination rounds and failing to do so.

My last article discussed the mono-Blue deck I took with Your Move Games Japan to the Neutral Ground PTQ. I called it Stake Through the Heartbeat because that matchup was pretty much an auto-win, but that isn’t the case if you adopt a bad plan. All you really need to do to give Heartbeat a chance (albeit a small one) is use your precious Hinders to fight over irrelevant spells like Early Harvest. Even with an active Jushi Apprentice you don’t have nearly as many hard counters as an old-school Draw-Go deck; if you waste them on non-business spells you may find them resolving Wild Harvest and generating more new threats than you can actually handle.

One of the reasons I chose Stake is that I’ve had enough experience with that kind of deck that I felt confident I could come up with good plans even against matchups I hadn’t tested, whereas few of my opponents would be expecting Jushi Blue and thus might not have a good strategy prepared for it.

Play leads to Plans leads to Sideboards

Earlier, I discussed the tendency people have simply to play out their matches during playtesting, as though the goal was to figure out the win percentages rather than to figure out the matchup strategy. If that’s all you’re doing it becomes very difficult to maximize your sideboard, because you can only sideboard “good cards” rather than cards that address exactly how the matchup works.

For an ironic example, let’s look at the anti-Heartbeat strategy Mike Flores proposed for his WWr deck for Team Constructed….

My swap has been:
-4 Manriki-Gusari
-2 Paladin en-Vec
-4 Savannah Lions
+4 Threaten
+3 Bathe in Light
+3 Scour

Q10) That seems terrible! What have you been smoking?
My theory was that Savannah Lions is terrible against a sideboarded Heartbeat deck… I just never want to see it against Carven Caryatid and Sakura-Tribe Elder. Likewise, Paladin en-Vec is a bit slow.

On the other hand, I want my most extreme cards in after boards. I think that despite a very consistent Game 1, my deck is behind against a Heartbeat player who transforms. However, while the opponent is ramping up to Keiga mana, I can get damage in, and then steal a million damage with a Threaten.

At the same time, Scour is really strong if the opponent doesn’t transform. You only have to leave open two mana for Scour to work. The Heartbeat player’s own signature enchantment allows R/W to power out this White Cranial Extraction.

Mike isn’t just boarding in random good cards versus Heartbeat. He’s looking at how sideboarded games play out — both with and without transformation — and picking cards that can solve the relevant strategic problems.

Someone who simply played WWr versus Heartbeat might well come up with Bathe in Light (since most of his losses are likely due to Savage Twister) and with some ingenuity he might come up with Scour — recognizing that the opponent has to put the target into play to go off, and that the target reduces the effective cost on Scour to 1W*. But how likely is he to come up with Threaten? Mike’s options are broader than most of ours because he is developing sideboard plans based on how the matchup works.

Once you’ve come up with a plan for post-sideboard games, you naturally want to test them as well. Plans that look good in theory may not be good in practice, or they may be so good that you don’t need to devote quite as many resources to them, freeing up sideboard slots for other matchups. How did Mike’s innovative plan play out? Despite winning the great majority of unsideboarded games, Mike was still losing most of his Heartbeat matches, suggesting a serious problem. (Hence the irony of the example.) From Mike, same article:

Q11) That makes some amount of logic (I think). So what’s the problem?
Using my strategy, the R/W is never efficient in Game 2. Scour and Threaten — both in the deck in Game 2 — are never good in the same game. You can lose games by holding Threaten in games where the opponent has no non-Caryatid creatures, or Scour when he plays Meloku. Bathe in Light is actually good all the time, because the opponent’s most dangerous Game 2 card — transformed or not — is Savage Twister.

Even though I’ve been behind against Heartbeat in the queues, the matchup is far from hopeless, even with my strategy (sub-optimal as it seems to be). The majority of losses also involve some kind of Garrison-/Shinka-/Tendo-screw, and the opponent rarely has more than three or so life at the end of a lost game.

This analysis of how the games play out might suggest modifications to the original plan. It might be wise, for example, to replace Threaten with Pacifism. You don’t get the flashy, “Might I be so bold as to borrow your Keiga for the turn” wins, but you’ve got a much better answer to Carven Caryatid, which is likely to show up whether the opponent transforms or not. (As Michael pointed out to me personally, Magic isn’t about Cool Things; Pacifying a Caryatid and swinging for four or five may be flat-out better than waiting to win with your opponent’s Keiga.) If Scour isn’t winning as many games as hoped, perhaps it should be replaced with additional burn to get in those last three points of life — or perhaps the Paladins shouldn’t be boarded out. Yes, they’re slow, but they also have protection from Savage Twister.

Even when you’ve got the right plan, testing will show you how to refine it. Consider Stake’s over-the-top sideboard against Gruul: three Threads of Disloyalty (going up to four total), three Slay, two Ribbons of Night and one Night of Souls’ Betrayal. How might one evaluate those numbers?

On general principles, Slay has to be better than Ribbons of Night. Three mana is a lot less than five, and Ribbons can’t even kill Rumbling Slum if my opponent happens to play one. Slay also fits the deck’s basic strategy better; since it’s an instant, I can hold open mana for permission or bounce (in case my opponent tries to Cloak up a Kird Ape) and then Slay at end of turn when he plays a Burning-Tree Shaman.

So why not four Slay and just one Ribbons? Apart from the fact that Ribbons can kill non-Green creatures, I’ve found in testing (some with this specific deck and some with decks like it, i.e. by experience) that I had enough “stuff” to deal with early creatures but that I was vulnerable to getting burned out as the game went long. Flames of the Blood Hand obviously affects this calculus somewhat, but they won’t always have that card or the mana to cast it when you play Ribbons, and you may have Shoal for it in any case.

More testing is undoubtedly the best way to find out what the right mix is. If testing showed that life total wasn’t critical but that Gruul decks were able to fight a long attrition war, it might even be that Ribbons should be replaced with Twisted Justice! (It’s not the case, but hypothetically it could be if games were going long and Stake was dying to the “last fattie.”)

Testing was also behind my wish to force more hard counters into the deck, even if said counters cost four mana. Against some decks, you have to play a long game in which mana can become so plentiful that Remand and Mana Leak don’t do enough, and Hinder and Shoal aren’t enough hard counters. Testing was also behind my rather controversial idea of sideboarding Plagiarize. Both Izzet Control and Wildfire games tend to go long, and eventually they will try to draw a bunch of cards in their main phase.

There is no shortage of examples, but the principle is what’s important. Testing should be used to explore and evaluate game plans and to tune your deck and sideboard accordingly, rather than simply figuring out who is ahead in the matchup. Taking the latter approach will almost guarantee that you get outmatched by opponents who understand the correct plan (and thus outplay you) and have tuned their decks accordingly (and have thus outbuilt you).

Hugs ‘til next time,

* Since Mike is running Karoo lands, this isn’t strictly true — two actual lands are required to cast Scour, as Boros Garrison will only produce three mana with a Heartbeat in play. Whatever.