If you are like me, one of your biggest June disappointments was the lack of GP Singapore video coverage. After all the awesome GP Charlotte and GP Copenhagen footage and commentary (minus the ridiculous Amulet Bloom questions out of Charlotte), I really wanted to finish the month with the Affinity and Jund showdowns in the Singapore semifinals. We may have missed out on the coverage but we definitely got a lot of action out of Singapore. Between Affinity’s win, the big question marks surrounding Grishoalbrand combo, and Jund’s commanding performance, GP Singapore is a great way to finish off the month. But with all the data and decks to process, it’s easy to miss the big takeaways from the event. This article aims to shed some light on those datapoints so we can understand what happened at GP Singapore.
In the spirit of my GP Charlotte and GP Copenhagen reflection articles, I want to unpack the GP Singapore metagame to help us see what decks did well and what decks did not. Many players fixate on the Top 8 or even Top 16 of an event to determine metagame health, deck viability, format composition, etc. But these results are often incredibly misleading. To account for that, this article examines the Day 2 metagame alongside the Top 32 results, pulling out some quantitative and qualitative lessons we can learn from GP Singapore. As a preview of what’s to come, let’s just say that it’s no coincidence Olivia Voldaren is presiding over our article today…
Comparing the Day 2 and the Top 32 Metagames
If there’s anything I hate about GP coverage, it’s the lack of Day 1 metagame numbers. Yes, I understand Wizards probably doesn’t want the metagame to get “solved” or anything like that (something we know Wizards is worried about). But in this information age of Magic, players are basically doing this anyway, so those Day 1 numbers would at least give some interesting context around the GP. Because we don’t have anything on GP Singapore’s Day 1, it’s impossible to know how archetypes did relative to their initial Day 1 showing. Maybe 30% of the field showed up at the GP with UR Twin sleeved up, even if many of them didn’t make it to Day 2. In the absence of this datapoint, we need to look for other points of comparison to figure out how decks performed and what the metagame looked like over the weekend.
Thankfully, Wizards threw us a bone here in releasing the Top 32 decklists from GP Singapore in addition to the Top 8. Normally, I’m happy to just get the Top 8 and a metagame breakdown. Top 32 is like Christmas in June. Wizards made up for this by posting the most confusing Day 2 metagame breakdown in recent memory, but at least we have data to work with. To get a sense of deck performance, I am going to compare the Day 2 prevalence for decks with the Top 32 prevalence for decks, looking for areas where decks either over or under-performed, and checking what percentage of any given deck actually made it into the Top 32 bracket. This helps us understand how decks performed and what decks succeeded and failed at the GP. For instance, taking our UR Twin example from above, if 10% of Day 2 was on UR Twin, we would expect to see 10% of the Top 32 also on the deck. But if 30% of the Top 32 was UR Twin, then the archetype overperformed its expected share by 20% points. And if UR Twin only made up 2% of that Top 32, it would have underperformed by 8% points. These are the kinds of differences we want to look for in our analysis.
The table below shows all decks with above-average representation (4+ appearances) on Day 2 at GP Singapore. It excludes the 34 decks with one, two, or three showings (everything from Scapeshift and Grixis Twin to Skred Red, Stompy, and Mardu Midrange). For each deck, the table gives the Day 2 count and metagame share along with the Top 32 share. The second-to-last column expresses the difference between the Day 2 share and the Top 32 share. Higher positive shares (e.g. Jund’s +9.6%) indicate overrepresentation in the Top 32 relative to the Day 2 breakdown. Lower negative showers (e.g. Burn’s -6.8%) indicate underrepresentation in the Top 32 relative to those Day 2 numbers. Finally, the rightmost column shows what percentage of the Day 2 players converted into the Top 32.
|Deck||Day 2 N||Day 2 %||Top 32 %||Day 2 %|
vs Top 32 %
The table is sorted on the “Day 2 % vs. Top 32 %” column, but feel free to resort on other metrics (“Conversion” . Here’s a quick example of reading the table. Affinity had 13 players during Day 2 making up 9.9% of that metagame. But it had a 15.6% share in the Top 32, which was an overrepresentation of +5.7% points. That said, only 38.5% of the Affinity Day 2 pilots made that conversion to the Top 32.
Looking over the table, particularly when sorted on that percent-comparison column, we immediately see decks falling into three brackets. In the first bracket, we have these heavy hitters with significant overrepresentation in the Top 32: Jund, Affinity, and UR Twin are the big ones here. These decks exceeded their expected Top 32 prevalence by a good margin, suggesting that there’s something about the decks and/or the metagame context which made these successful. In the next bracket, we have a chunk of decks without any significant difference between their Day 2 and Top 32 shares. This is everything from Temur Twin/Grishoalbrand down to Abzan Company. Although these decks had some overrepresentation and underrepresentation, the magnitude of those differences isn’t very big. It’s also not statistically significant relative to the other differences we see in the population. Because of that, it’s hard to say anything definitive about these decks: the difference in their Day 2 and Top 32 shares falls within expected variance. Finally, we have the third bracket, i.e. the loser’s bracket. These are decks like Grixis Control and Burn which sent a fair number of players to Day 2 but very few into the Top 32. They have significant underrepresentation values, which suggests that something went wrong with their deck or the metagame for them to do so poorly. The sections below break out those decks, giving some context on their performance and some higher-level takeaways regarding each deck in Modern.
The Winners: Affinity, Jund, and Twin
Overrepresentation does not always mean overperformance. But in the case of these three decks, there is strong reason to believe that it does, both from a quantitative standpoint and a qualitative one. Let’s start with the tournament’s big winner: Affinity. Affinity absolutely crushed it at GP Singapore by almost every metric. It sent two pilots to the Top 8, won the entire event, overrepresented its Day 2 share by 5.7% points in the Top 32, and was tied for most-played deck on Day 2 overall. Although we don’t know how many people showed up with Affinity on Day 1, this suggests that other players also thought Affinity was well-positioned and trusted in their robots to carry them to victory. This didn’t work out for everyone: only 38% of the Day 2 pilots actually converted into the Top 32. While that’s not as bad as Grixis Control (0%) or Burn (8%), it’s also not quite the Jund conversion rate of 50%. But that’s a much bigger deal for a deck like Affinity that either succeeds spectacularly or fails miserably. For Affinity to put up all these numbers, even accounting for its 38% conversion rate, you would need a metagame that was at least somewhat hospitable to it. GP Singapore was apparently such a metagame, with opponents either forgetting to pack meaningful sideboard hate (no Abzan means no Stony Silence), or opponents moving away from decks that are historically good against Affinity (GP Singapore had a lot less Twin than we saw at other major events). All of this underscores the point that all Affinity players have known for years: if you are good at Affinity, keep playing it. Sometimes the metagame is hostile to you, but if it isn’t, you will always be rewarded.
Affinity may have won the entire event, but Jund was perhaps even more a winner from a data perspective. Jund had the highest overrepresentation in the Top 32 (+9.6%) and the second-highest conversion rate of any of the most-played decks (50%, behind only UR Twin at 57%). Jund saw two pilots in the Top 8, 12 throughout Day 2, and most importantly, a huge competitive advantage over its rival, Abzan. Abzan had only four players make it to Day 2, only one of whom made it into the Top 32 (in fairness, Ray Wee did make Top 8 with his deck, but it’s extremely metagamed and not yet indicative of broader archetype viability). The metagame has spoken and Jund again emerges on top. This is huge news for Jund mages, and for the format generally, because there were some question marks surrounding the deck after the events at the beginning of June. With so many players on big mana decks like Tron and Amulet, or on bolt-proof decks like Grixis Control or UR Twin, Jund’s red-based edge wasn’t looking as hot in practice as it had in the theorysphere. But GP Singapore, along with all the other supplementary datapoints over the month, have further solidified Jund’s status as the tier 1 BGx deck of choice. Remember that this deck had a huge target sign on its head all month. Everyone knew it would probably be the BGx deck of the GP, and everyone knew the hype was strong with it. And yet, the deck still prevailed. A big reason for this was cards like Olivia Voldaren and Chandra, Pyromaster, which helped Jund play a more durable game against decks that would otherwise go over the top as the game went late. We still need to finish crunching the June numbers, but Jund is looking like a huge winner from the month, and GP Singapore is very much emblematic of that.
Then there’s UR Twin. Betting against UR Twin in Modern is like betting against the Patriots in the NFL. They both create polarizing opinions, they both stir up strong feelings, and they are both really, really good no matter what else you think about them. UR Twin was a sleeper winner at GP Singapore; although it didn’t have big Top 8 finishes, its overrepresentation relative to Day 2 and the sheer number of players it sent to the Top 32 are impossible to ignore. It’s significant both in that UR Twin remains the best Twin option despite Temur and Grixis innovations, and in that UR Twin is capable of succeeding despite players knowing this deck was going to be big. In that respect, Twin was similar to Jund. Jund may have had more hype going into GP Singapore (indeed, the only decks in the entire format with as much June hype as Jund would probably be Amulet Bloom and Grishoalbrand), but UR Twin was the known enemy everyone expected. Despite that expectation, UR Twin put up respectable numbers and continued to show its tier 1 status. From a format health perspective, this suggests that Twin is both a good deck but also not a too good deck. It’s okay for nonrotating formats like Modern to have established tier 1 players, and Twin has shown it is capable of filling that role while not also winning every single event. That’s a healthy place for a format to be, and a healthy station for Twin to occupy.
The Losers: Burn, Naya Company, and Grixis Control
I’m hesitant to really call these decks “losers” from the metagame perspective. It can be very misleading to extrapolate a deck’s viability from a single event, especially when you have multiple other events from the very same month which suggest alternate stories. But if we focus just on GP Singapore as one datapoint in the entire month, this was not a good event for Burn, Collected Company, and Grixis Control. Let’s start with Burn because, quite frankly, there’s not a whole lot to say about it. Burn is always going to be the tier 1 aggro deck of Modern that won’t often win major events. You’ll always see it at every stage of the tournament, you’ll always see it in smaller events, but you can’t expect it to succeed on big stages. There are just too many people who are prepared for it, and that preparation is often going to be independent of hype around the deck. This is in contrast to something like Amulet Bloom, whose fortunes will rise or fall depending on the number of Blood Moons in the room. But Burn? You will always have cheap removal (Bolt), efficient countermagic (Dispel), and lifegain (Feed the Clan/Kitchen Finks) to keep it down. Indeed, Modern players have gotten a lot better at boarding against Burn and beating the deck than they were in late 2014. Players have moved away from Destructive Revelry-lite answers (Leyline, Claw) and into spell-based ones that Burn struggles to play around. So long as this continues, which it likely will for the foreseeable future, Burn will see a lot more GP Singapores to come.
Collected Company is a more interesting case. On the one hand, GP Singapore was no GP Charlotte, where we saw lots of Company decks in both the Top 8 and Top 16, as well as plenty of Company success on Day 2. On the other hand, we do see Company in GP Singapore’s Top 8 (in an awesome but bizarre Kiki Pod variant by Tay Jun Hao) and in the Top 16 (in a more conventional but still awesome GW Hatebears shell by Tao Bozhi). So the failure wasn’t necessarily in the card itself, which remains a strong engine and continues to enable green decks we never would have seen without it. Instead, the failure was in the builds around that card. Naya Company was huge at GP Charlotte, in no small part to Paul Rietzl’s 10th place finish on the deck and plenty of on-screen coverage of his games. Modern players always have a special soft spot for Naya decks (see any article by Brian Kibler ever), and it was exciting to see the deck’s renewed viability off Company. Unfortunately, GP Singapore did not do the deck any favors: 6.1% of Day 2 piloted the deck compared with an embarrassing 0% for the Top 32. Although I don’t like extrapolating metagame-wide success rates from one event, those aren’t the stats you want to see on a deck that is supposed to be good. To me, this just means players need more practice with both building and selecting Company-based decks in Modern. Between Elves, Abzan Company, Naya Company, GW Hatebears, and all the other decks people are still experimenting with (Slivers! Werewolves! Humans! Oh my!), there’s a lot of room for the archetype to grow. It also suggests some degree of metagame preparedness for these strategies (see Jund’s reliance on cards like Olivia and Chandra). GP Singapore just underscored these points.
Then there’s Grixis Control, which to me is the oddest failure from GP Singapore. Tabling the issue of whether or not Grixis Control is really a “control deck” (spoiler alert: it is), this deck, whatever you want to call it, seemed well-positioned for the GP. Its had lots of MTGO and paper success all month, including Top 8 and/or Top 16 finishes at both of the previous GPs. Our collective understanding of the deck has also improved dramatically in the past few weeks, with players understanding what counterspells to use (Dispel is insane), the different ways to build the deck (our very own Trevor Holmes wrote a great article on this the other week), and how they need to pilot their builds to victory. It just didn’t pan out at GP Singapore, with zero of the deck’s six pilots making it from Day 2 into the Top 32. Of all the failures at the tournament, this is the one I’m most willing to write off as a temporary anomaly. The deck has had such consistent and commanding success at every other June event until this point, and we can’t throw that out with just one subpar event. After all, it’s not like zero Grixis Control players made Day 2 in the first place, so the deck was at least viable enough to make it that far. Of course, that all said, there is one metagame context which might have caused trouble for Grixis Control. The Griselbrand and Nourishing Shoal combo is surprisingly resilient to control elements, chiefly on its instant-speed timing and the strength of the splice-into-arcane mechanic. At absolute best, this is an even matchup for Grixis Control. At worst, an experienced Grishoalbrand pilot will turn this into a 40-60 uphill battle. Just a few of these matchups could have sunk Grixis Control’s fortunes at GP Singapore.
The Oddballs: Grishoalbrand and “Weird Decks”
Before wrapping up, I want to quickly discuss the ten ton Demon in the room and all the related “weird decks” in the same category as Grishoalbrand. One of the biggest problems with Modern is all the ban talk surrounding the format. Part of this is a function of Wizards’ heavyhanded management of the banlist, and lack of transparency around some decisions (even if I largely agree with their bans/unbans so far). But another part of this is player unwillingness to consider potentially broken decks, mostly combo decks, as anything other than ban bait. Instead of thinking about ways to metagame against a deck, whether with sideboard bullets, deck choices, or clever maindeck inclusions (Dispel, I choose you!), many players immediately turn to the “Wizards ban please” approach. Grishoalbrand falls squarely in this category, with all the Amulet Ban ban maniacs quickly jumping ship to Griselbrand/Shoal/Goryo’s Vengeance after less than two weeks of results and very little in the way of concrete evidence.
GP Singapore showed us that Modern is able to contain these kinds of fast, resilient, and potentially broken combo decks through the usual policing methods. We saw this at GP Charlotte with the omnipresence of Blood Moon, and we saw it again at Singapore with all the countermagic, discard, and Scavenging Ooze strewn across the format. This suggests both that Modern is capable of self-regulation and also that Grishoalbrand isn’t as strong as many people thought it would be. Just looking at the table earlier in this article, Grishoalbrand had slight Top 32 overrepresentation relative to its Day 2 performance, but it was neither statistically significant nor significant in its magnitude (it’s hard to draw conclusions from a population of four cases). This reflects our general metagame findings on the deck, which point to Grishoalbrand being a solid tier 2 combo deck but nothing crazier than Amulet Bloom or Infect. That is also to say, this returns to the concept of “weird decks” in Modern and how that diversity affects our format. I will admit that decks like this highlight the need for versatile police spells in Modern, so we aren’t in a situation where you need Silence/Moon/Rest In Peace in your sideboard or you just lose. But I also think many Modern players overstate this state of affairs. Between cards like Ooze, Dispel, Spell Snare, Thoughtseize, and other Modern policemen, we really do have a lot of maindeckable bullets with widepsread utility in the format. So until we get more data to suggest a real metagame problem, all the theoretical Grishoalbrand banning discussion needs to be off the table. GP Singapore is yet another datapoint that suggests the deck is a welcome addition to Modern, and no amount of theoretical hyperbole will change that.
Modern After GP Singapore
As a final note before we wrap up, remember that we can’t extrapolate too much from a single event. Just because a deck succeeds at one GP, that does not mean it’s a successful deck in Modern. The same goes for decks that fail, or decks that just don’t do much at all. Of course, if you find decks that succeed at a single GP that are also succeeding across the format (e.g. Jund, UR Twin), then that’s a great case for a real success story. But if you have isolated incidents that don’t make sense in the broader metagame picture (poor Grixis Control), then we will need more data and results to make those top-level conclusions.
With the month winding down, we are almost at the point where we can tally up the numbers and give some high-level conclusions on the state of Modern after June. I for one am very excited to go through the stats and see what happened, checking for broader trends that can inform our deck and card decisions as we go into the PPTQ portion of the summer. After all, Modern season is far from over, and there are still plenty of tournaments left to take down and metagame data to compile. Just looking at GP Singapore, I am heartened by the results of the event and the relative diversity in both its Day 2 and Top 32 metagames. It’s a great time to play Modern and I’m excited for what the rest of the summer brings.