fbpx

Magical Hack: Honolulu By The Numbers

Read Sean McKeown... every Friday at
StarCityGames.com!
My usual gig is to do a lot of math and come up a few weeks later with an enormous statistical work about the Pro Tour, but it seems my schtick has been outsourced to some very clever computer programmers who can make a metal box filled with trained monkeys do all the Excel manipulations for me instead. Well, not to be completely outdone by this article, I’ve run a few numbers myself to attempt to figure out what the most successful deck was arbitrarily, rather than by sheer win percentage.

For the first time in many years, a Pro Tour has used the Standard format outside of the World Championships, and the diversity of the Standard metagame following the release of Guildpact is astounding. Constructed Pro Tours often seem to be overshadowed by some inherent lack of variety, be it the Affinity/Anti-Affinity metagame of Mirrodin or the many different flavors of Tinker decks seen at Pro Tour New Orleans. This time around, however, there were somewhere in the neighborhood of twenty serious archetypes represented in Hawaii, and seven different decks in the Top 8… or five different strategies, depending on how finely you wish to split hairs about it.

It was a control-oriented format, with more Black/White control decks (with or without Green for Loxodon Hierarch) than any other deck in the format, which explains very nicely how it happened to be that two aggressive Zoo decks (one with White, one without), two aggressive Black/White “Rat” decks, and two anti-control “Owling Mine” decks made the elimination rounds. Blue/Red Tron (a trademarked Mike Flores Deck), and Heartbeat of Spring were the other two decks to make an appearance on Sunday. At the end of two days of dodging Zoo, the Owls were on the bench, and some amount of vicious stabbing later it was Kird Ape who reigned supreme. No wonder he used to be banned in Extended!

My usual gig is to do a lot of math and come up a few weeks later with an enormous statistical work about the Pro Tour, but it seems my schtick has been outsourced to some very clever computer programmers who can make a metal box filled with trained monkeys do all the Excel manipulations for me instead. Well, not to be completely outdone by this article, I’ve run a few numbers myself to attempt to figure out what the most successful deck was arbitrarily, rather than by sheer win percentage. For that effort, I entered all 410 decks played into Excel, correlated to their final standings, and poked and prodded to determine the statistical average finishes for each, and the standard deviation of those finishes to see how consistently the deck finished somewhere in that area.

The key concepts here are accuracy and repeatability. When you’re going into a tournament, you want a deck that can consistently do well, rather than one that will do well under certain conditions (such as “if no one plays Zoo.”) Weighing the decks and getting their average finish will do one thing quite nicely: it will take every single person who played the deck and average out their results, so that the second-place finish by Craig Jones is matched (or even negated) by the 398th place finish of one Oliver Robertson, who in his effort to keep up with the Joneses didn’t win a single match with his Zoo deck. The decks that do consistently better will have an average better result, fairly obviously. But the decks that do consistently better should also have a tighter spread on their numbers, averaging a larger number of high results than other archetypes, with very few poor results dragging down the average. To counteract the “problem” of balancing the Craig Joneses against the Oliver Robertsons, having a look at the standard deviation of the finishes while we are taking their average will give us an idea of how tight of a range they fell in overall, because we want to make choices that put us in high finishes, rather than flipping a coin to see of we’re a Craig this tournament. “Heads, I’m a Craig; tails, I’m an Oliver” is not the key to success at life, Magic, or women, and so the decks with the lowest standard deviation will have the most consistent chance of finishing at or near their average record.

Then, of course, there will be decks with low standard deviations (very solid chance of finishing near that number) but also a low average place finish, like 367th. These decks will clearly be termed as “bad,” because they very consistently did poorly. That is the opposite of being “a winner,” and should be avoided like the plague.

Archetype Average Finish: Variance: Points: Variance:
R/G Zoo 66 83.44 29 11.31
U/R Control 99 77.94 23 12.06
Ghost Dad 102 62.91 23 8.96
R/G/U Ninja 118 110.53 23 12.49
G/W no-Glare 130 93.30 20 10.85
B/W Aggro 137 108.83 21 11.37
Owling Mine 139 136.14 18 13.20
GhaziGlare 148 137.18 21 12.79
U/R Magnivore 167 147.70 19 13.39
Greater Gifts 171 109.83 17 10.51
U/R Tron 184 111.76 17 10.99
Heartbeat 186 127.21 16 11.87
B/W Control 186 115.05 16 10.55
B/W Rats 191 125.56 16 11.29
B/G/W GhaziGlare 198 143.31 16 14.47
U/R/W Control 210 188.35 15 16.65
Greater Good (No Gifts) 213 113.81 12 8.11
Zoo 217 116.96 14 10.44
B/G/W Control 227 105.50 12 8.74
R/W Burn 230 226.27 16 17.68
U/R/W Wildfire 230 147.08 14 10.61
U/W Control 223 27.58 11 2.12
Enduring Ideal 250 79.80 10 6.22
R/G/B Aggro 260 202.94 12 16.97
B/G/W Rock 260 123.84 58 9.93
R/G 265 80.37 9 4.81
Eminent Domain 268 136.57 10 10.89
Hana Kami Gifts 268 123.71 118.09 8
Red Deck Wins 279 118.09 8 6.36
U/B Tempo 283 99.02 7 4.58
Zoo (with Bob!) 284 50.2 8 2.12
B/W/R Enchantment Control 296 84.72 7 4.58
U/B Tron 296 143.79 7 10.21
R/G/W 343 53.53 5 2.87
Wildfire Gifts 345 12.73 5 2.12

For a few hints on what not to play, look at the nice and low standard deviation on Wildfire Gifts (2.12 match points, or within 12.73 places of each other) coupled with the bottom-of-the-barrel average finish of 345th place. Sorting the archetypes by best average finish, the real winner is the Red/Green Zoo deck, followed by Blue/Red Control and Ghost Dad; Blue/Red is a real surprise there, but it’s no surprise that after the fact we can see “the deck that won the Pro Tour” and “the best innovative twist of an established concept” as the two big winners at the event.

Naming conventions are a big thing here, especially since in a lot of cases it’s nebulous and ambiguous. (Never mind the fact that when I was compiling these results I noticed that over sixty players did not have decks attributed to them, as the Wizards page for this Pro Tour had a severe error in the L-S page that omitted a lot of people and repeated some decks over and over and over again.) The big difference between Red/Green and Red/Green Zoo is that every deck that played a turn 1 Elf or Bird was automatically Red/Green, even if that was the only difference in the deck’s philosophies. Zoo decks play one drops to attack with, but that doesn’t stop Red/Green decks from playing Kird Ape. It’s just that sitting next to the Ape you can have a Bird or Elf, or you can have Frenzied Goblin, the Little Man That Could at this tournament. Giant Solifuge versus Rumbling Slum, words words words Michael J. Flores! I don’t care what your four-drop of choice is, or even whether you play one or not; if you had an Elf, you were labeled R/G.

Black/White Aggro and Black/White Rats were often very similar, but the distinguishing characteristic is that the Aggro decks played Isamaru, Savannah Lions, or both. It’s a similar distinction to Zoo versus Red/Green, in that it’s a fundamental difference between the plans each deck has for the game and how quickly it should develop. Most decks with Lions didn’t have Rats, and so couldn’t be a Rat deck anyway; Shrieking Gargoyle may be a flying rat, but a flying rat is a Pigeon (trust me, I’m from New York City where we know these things at birth) and so Shrieking Gargoyle in your deck does not a Rat deck make.

Everything else should have a very clear distinction, because the difference between Greater Gifts, Hana Kami Gifts (no Greater Good to tutor for, no Yosei lock, but endless Splice recursion instead), and Greater Good (No Gifts) should be reasonably obvious. Greater Gifts is considered the standard plan, while Hana Kami Gifts diverges from that plan by excluding key elements, while Greater Good (No Gifts) uses some key elements of that plan… but not the overpowering, back-breaking Gifts Ungiven that can seal things up very nicely in a nice, tight, inescapable package by playing Gifts for Reclaim, Recollect, Yosei, and Greater Good.

Anything below B/G/W Control on this list should basically be considered the noise of Standard, with poor results that did not stand up to the rigorous testing. For “rigorous testing,” read “throwing decks violently at each other, with thousands of dollars in prize money for the one coming out of the pit-fights wearing the most blood (not theirs).”

This means that playing Savannah Lions and Lava Spikes was proven unviable, as was modern Enduring Ideal decks, fattie-based Red/Green decks, Adrian Sullivan Eminent Domain, stock Blue/White control, Hana Kami Gifts, four-color Zoo stretching past the breaking point for Dark Confidant, fattie-based Red/Green decks splashing for Hierophants, and a few other good ideas that just didn’t pass muster. Some of these are not necessarily dead, because only a very few people played them, but those people didn’t do well if their “average” record is “well out of contention for Day 2.”

This includes the deck Flores chose to talk about as of midnight the night before the Pro Tour, but then… only two people played anything quite like it, and it’s a Flores deck so I’m sure the hype will pick up more as he continues to talk about it.

The puzzle is… how to run the numbers smoothly to reward you for a low average finish (the smaller the number, the closer to the top you finished) and for low statistical variance in your finish (the smaller the number, the more likely you were to finish near that number)? Math buffs will realize that the operation we are looking for here is called “straight multiplication,” where you combine two numbers and see what happens, whether there’s a second date, the hair and eye colors of their kids, et cetera. Much simpler than other statistical and mathemagical operations presented herein, and so possessed of a certain… elegance?

Here’s what happens when the numbers fight to the death some more:

Archetype Average Finish: Variance: Points: Variance:
R/G Zoo 66 83.44 29 11.31
Ghost Dad 102 62.91 23 8.96
U/R Control 99 77.94 23 12.06
G/W no-Glare 130 93.30 20 10.85
R/G/U Ninja 118 110.53 23 12.49
B/W Aggro 137 108.83 21 11.37
Greater Gifts 171 109.83 17 10.51
Owling Mine 139 136.14 18 13.20
GhaziGlare 148 137.18 21 12.79
U/R Tron 184 111.76 17 10.99
B/W Control 186 115.05 16 10.55
Heartbeat 186 127.21 16 11.87
B/G/W Control 227 105.50 12 8.74
B/W Rats 191 125.56 16 11.29
Greater Good (No Gifts) 213 113.81 12 8.11
U/R Magnivore 167 147.70 19 13.39
Zoo 217 116.96 14 10.44
B/G/W GhaziGlare 198 143.31 16 14.47
U/R/W Control 210 188.35 15 16.65

Here, we’ve killed everything with an average finish below B/G/W Control, and the only reason that bottom-feeder was left in is because everyone seemingly thought it was a good deck for the Pro Tour, and of those “everyone,” it turns out six of them were right. (Possibly more: as I said, there are sixty or so decks unaccounted for in this analysis, which makes me wonder how the article that went up Monday got so lucky in having everyone’s decklist attributed properly. It must be good to have Teddy Cardgame contacts working on the inside, leaking accurate data while the casual at-home reader gets the incorrect data.) This is all sorted in descending order on the arbitrary category “AF x V”, which shows us the low-number (“good”) finishes and the low-number (“consistent”) variance very neatly.

Again, it seems that the big winners are Pierre Canali/Guillaume Wafo-Tapa/Erwan Maisonneuve’s Blue/Red Control deck, Red/Green Zoo, and Ghost Dad. It’s possible they won out because fewer people played them, and so there was less dead weight at the bottom of the bracket because they got lucky… but there are plenty of decks played by as few people that really stunk by this sorting system, so we’re attributing their being on top to “being a good deck” instead of “general lucksackery”. (And here you thought only John Rizzo and Mike Flores invented new words. Agnostically.)

A lot of things happened this past weekend. Some tricky decks playing Ebony Owl Netsuke and Sudden Impact migrated from Magic Online like a computer virus gone bad, trying to take over the operating system of the Pro Tour and flood you with .wav files of the incessant cheering of every twelve year old Magic player watching from home, and some Kird Apes told them to shove off and die plstks. Said Owls were told so earlier in the tournament, as they faced Kird Apes earlier in the tournament, but the two that made the finals were not voted off the island by Kird Ape and his friends from the Zoo until they got to the elimination rounds. Some people played Heartbeat decks, and things went… okay for them. The lowest-ranked Heartbeat player still won at least one match, which is more than can be said for some other decks. A lot of people played Godless Shrines, and I mean a lot of people, and some of them did well… most importantly, the ones who started cutting expensive cards and started putting in one-drops, or possibly playing removal spells that cost less than one mana.

Standard was proved to be a format where the critical play is attacking for two, and the better you are at it, or the more tricks you have to either a) attack for more than two or b) kill things dead that are trying to attack you for two, the better your deck seems to be. The premise coming into the tournament was that it was a control player’s tournament to win, with decks like Greater Gifts and GhaziGlare expected to continue beating aggro decks all day long, and their new friend Black/White Arena Control to start slapping around one-drops like a buzzing swarm of flies.

What actually happened is that the good guys won, by attacking for two, or in a certain Kird Ape’s case, attacking for eight. Creatures won the day, but the question seems to be: why?

The easiest answer is that attacking is an acceptable strategy at the moment, but in a world with anti-aggression cards like Loxodon Hierarch and Carven Caryatid we know that can’t be the whole answer. Wrath of God was everywhere, and it had close relatives Pyroclasm and Wildfire running around rampant to boot. It’s possible that fifteen decks changed the metagame so drastically that the aggressive decks had to win, because they disrupted the natural ecology of the format. If Owling Mine crushes control and does so with a bunch of people making Day 2 – and even losing to Zoo off a quadruple mulligan – does it surprise anyone that with what, nine Owl decks running around in a field of 120 or so people that the control decks were going to take some heavy losses throughout the day?

Put nine sharks in a school of a hundred fish that get eaten by sharks and some fish that sharks just won’t eat, and there should be no surprises that the next day you’ll see some sharks and some things the sharks didn’t eat, and everything else is a fine red mist diffusing through the water.

Owling Mine is already being likened to Turbo Stasis, as an excellent metagame deck that preyed upon known aspects of a metagame that could be reasonably expected to show up in the usual proportions. Some decks could never beat it. Some decks could mulligan to three, and beat them after they sideboarded in fifteen cards. Way more of the first type appeared than the second type, but the very hunger for preying upon control decks that saw them thrive in the end proved to be their undoing, as the natural predators for their predators, Kird Apes and Savannah Lions and Hounds of Konda, began to flourish in the empty space left in their wake.

Standard is a format about creature combat, and yet still a very reasonable home to a dozen or more viable choices. It isn’t even a game about Jitte, for that matter, but instead a format defined by the interaction of creatures and the positioning for advantage that ensues. (Some call it “racing.” Some topdeck Lightning Helix on their last turn to kill the opponent while at one life.)

Unless I magically obtain a more accurate list of which players played which decks, to flesh out the statistics to include those sixty or so excluded players, I can’t say much more than that the key innovations that made for the best decks in the tournament were Ghost Dad and Red/Green Zoo. Both took a key presumption about the decks that were going to appear as major archetypes and stood them on their ear, and both were very clever alterations of what we expected to see that worked beautifully in the metagame.



Next week, we will take a more in-depth look at Honolulu than merely running the numbers… because we will be discussing the Pro Tour in terms of its applicability to the next Standard format, Team Unified Standard.

Sean McKeown
[email protected]

[The information Sean used to compile his data can be downloaded here as an Excel spreadsheet. – Craig]