A couple of weeks ago, I was at Grand Prix Liverpool where 504 teams of three players each convened to battle in Modern. This article has nothing to do with Liverpool, and here’s why. At most tournaments we presume some connection, however tenuous, between deck choice and the result. The idea is that successful people run successful decks, and maybe we should follow their lead.

At team GPs, on the other hand, no one tracks individual players’ records, including players themselves, and for good reason. As soon as two members of a team win their round, the third match either stops or at least stops being relevant. This has the curious effect where a person can finish first at a tournament without winning a single match. A trio can go 18-9 in individual matches and be 9-0 in the standings. This last case actually happened once when I was doing coverage, whereas the former is hopefully an urban legend. Either way, we don’t want to look at these players’ decks for guidance.

So I spent a weekend in Liverpool gathering absolutely no meaningful data at all. At the exact same time though, 1,765 players spent their weekend in Portland, fought for their own personal record, and generated a lot of very interesting data. It always takes a little percolating before the Modern community adopts additions from a new set. Grand Prix Portland then was the first major event where Arclight Phoenix and Crackling Drake were out in full force, with a few copies of Risk Factor, Creeping Chill, Beast Whisperer, and Assassin’s Trophy in between.

Sample Metagame

Although I was covering a different event on a different continent at the time, I managed to take a look at 971 of the lists people submitted for Grand Prix Portland. What can I say—my spies are everywhere.

971 decks is the largest sample I have ever able to get my hands on, both in the absolute and on a percentage basis. It’s more than 55% of all decks that were played at GP Portland. We rarely know what the base is working with, and rarer still do we have such a good idea. After all, higher numbers mean higher confidence. That is, we can assume that the full Portland field didn’t deviate more than a few percentages from the following:

The next most played decks in line, included under “Others” above and listed alphabetically here, were: Belcher, Blue-Black Control, Cheerios, Enchantress (actually Eidolon of Blossoms), Goblins, Grishoalbrand, Martyr of Sands, Red-Green Eldrazi, Splinter Twin (actually Kiki-Jiki, Mirror Breaker), Treasure Hunt, Whir of Invention Prison, a couple more Izzet variants, and this green-white strategy with Knight of the Reliquary/Ramunap Excavator/Azusa, Lost But Seeking whose name I don’t know. What do you call that deck? Let me know in the comments!

For the purposes of rational analysis, I believe it’s legitimate to label all of these as “Others”—despite or even because of the undertone of judgment. Of course, one can’t help but root for the brave souls who try to do weird things with Protean Hulk, Soulflayer, or Madcap Experiment/Platinum Angel/Chance for Glory/Pact of Negation. My sample included one example of each. But we cheer their efforts because they put themselves in the position of the underdog, because theirs is a trick that’s so hard to pull off, and especially because the cool combos become boring when witnessed with regularity. In fact, just look at Protean Hulk, whose 15 minutes of fame were over in a, let’s say, flash.

So yeah, it is true that my sample contained no less than 90 different archetypes, 28 of them represented by a single player, from Allies to Slivers, and Grand Architect to Green Devotion. But that doesn’t make them a relevant metagame factor.

Much more interesting than the bottom end of the popularity spectrum is the top. That such a large part of the community agreed on Spirits is a surprise. We don’t have similar data from past events with which to compare Portland’s Day 1 field. But at Grand Prix Stockholm in the middle of September, Spirits wasn’t even the most played deck on Day 2. Back then, White-Blue Control still had almost twice as many copies in the running on Sunday. I suspect it was at least more popular to the same degree at the tournament’s onset. So Spirits claiming the throne is indeed the big headline here.

The other news and the more recent development concerns Arclight Phoenix. Together the various decks with Arclight Phoenix made up more than 7% of the Day 1 field. That’s not far from Tron numbers, and I imagine the firebird’s metagame share will grow further. Consider for a moment that it took Spirits almost five months from the release, or unleashing, of Supreme Phantom to lay definite claim to the pole position.

Experts agree that a wide format with an insane number of viable archetypes rewards players, more than anything, for being intimately familiar with their own deck and its matchups. This hands us a very good reason why the Modern metagame is known to move slower than others. I now wonder, though, if it moves even slower than commonly accepted. On the other hand, it is notable that it keeps moving this much. Just this year, we’ve seen the rise of White-Blue Control to the top of the metagame, followed by Hardened Scales’ climb to the upper echelons, a Dredge intermezzo, and now Spirits as number one. Maybe Arclight Phoenix will be the new king soon.

Adding up all the Rock-style midrange decks based in black and green yields a metagame share of 6.4% and grouping together White-Blue Control and its Jeskai cousin yields 6.2%. These archetypes still are a big part of Modern, although control in particular has fallen far from its former glory. The strategy is notoriously weak to graveyard recursion and usually relies on its sideboard or on a miracle—that is, Terminus, to combat. Maybe this should change. Maybe it’s time for main deck Rest in Peace. The portion of the field that makes more than incidental use of the graveyard as a resource, meaning fewer Tarmogoyf and Snapcaster Mage, and more Arclight Phoenix and Bedlam Reveler, is already closing in on 25% …

Success Rates

Having a good idea of the starting field’s composition enables us to look at metagame snapshots further down the line in the tournament’s progression and to infer how successful an archetype is relative to the others. Indeed, it is a prerequisite. Without it, a deck’s high representation on Day 2 allows multiple interpretations. It may have done well on Day 1, but it could just as well have been inordinately over-represented. Because the two aren’t unrelated—strong decks happen to be more popular—it’s typically a combination of the two. So a Day 2 metagame breakdown on its own provides some insight—it just doesn’t give the full story.

In the case of Grand Prix Portland, we don’t have a Day 2 metagame breakdown. But we can look at how many of the 971 players from our sample qualified for Sunday’s rounds…

 

The above is sorted by the percentage of people who made the cut. Overall, 22.97% of our 971-strong sample did. This is our baseline. Archetypes who placed more than this share of their pilots into Day 2 over-performed, while less than that means that they under-performed.

Please note that smaller absolute numbers leave more room for lucky hits and unlucky misses. For example, that 50% of all Grixis Control players continued playing on Sunday shouldn’t be taken as evidence that Grixis Control is one of the best decks in the format. It may be true, but there’s altogether too little data to say.

A number of data points here, on the other hand, do offer valuable information. For example, with 38 in our sample, Krark-Clan Ironworks had quite a lot of horses in the running initially, and that 17 of them were still in the running on Day 2 was remarkable. I don’t mean to suggest that KCI actually has a 45% chance to qualify for the second day of competition at a GP. This number obviously reflects the high end of variance. At the same time, it’s equally obvious that KCI outperformed almost everything else.

Similarly, all of the folks who picked Spirits and Tron seemed to have bet on the right horse. The two most played decks both also outperformed the field. Neither beat the average by much, but that’s exactly what we expected to find: A larger contingent comes closer to the overall average almost by its very nature.

The perfect counterexample poses Burn. The rate at which Burn made Day 2 doesn’t justify its huge following on Day 1. The deck was severely over-represented in our sample, not by any fault of my sampling but because it mostly just crashed and burned itself. Don’t play with fire, kids! Or if you do, consider adding some Faithless Looting, Fiery Temper, Manamorphose, Bedlam Reveler, and Arclight Phoenix.

Indeed, the two Arclight builds without blue put up very impressive numbers in Portland, regardless of whether the deck went for random discard and Hollow One or not. Granted, the sample is too tiny to provide conclusive evidence. But the 62.5% in the Hollow Phoenix row sure looks scary, doesn’t it? You recall how I wrote earlier that I expect Arclight Phoenix’s metagame share to grow? Well, I wrote that part before I ran the numbers for Day 2. That Izzet Phoenix did similarly well too comes almost as an afterthought, although this build had a line-up significant enough to yield credible results.

For comparison: About the same amount of players from our Day 1 pool brought Hardened Scales and Izzet Phoenix. Admittedly, to reiterate, I only work with a sample of 55% of the field here, so the real numbers may be off by a bit. But the available numbers have Izzet Phoenix outperforming Hardened Scales by so much that it’s hard to imagine that the scales weren’t tipped in its favor across the whole. The best part is that Hardened Scales itself outperformed the average already.

Meanwhile, both White-Blue Control and Jeskai Control fell short even of average results. Humans and Jund came closer, but that isn’t the most glowing recommendation either.

Finally, one of the biggest players in Day 1 was Death’s Shadow, and the deck even ended up in first place when all was said and done. But that’s variance for you. I don’t want to be overly confident in my sample. In fact, I wouldn’t even expect a full breakdown to reflect the various archetype’s strengths with perfect accuracy. But I trust even this imperfect aggregate more than I trust the single data point of Death’s Shadow winning the tournament. On average, Death’s Shadow had below average results.

TL;DR

Modern is in turmoil. Spirits has arrived at the very top of the metagame, and the Human tribe can’t compare with regards to numbers or win rate. Phoenixes are on the rise and outperform almost everything. Burn is way more common than it should be, and KCI is still running hot. Former favorite W/U/(x) Control is losing many of its followers as well as its matches, whereas Tron remains a popular and successful deck choice.