MC4 Data Deep-Dive: Hogaak and Graveyard Hate
Mythic Championship IV 2019, otherwise known as MC4 or MC HammerHogaak Edition, is in the books. If you missed the event, you've probably heard horror stories of Hogaak, Arisen Necropolis dominating from Day 1 into Day 2 and beyond. If you didn't miss the event, there's no way you avoided the vocal pros and players decrying Hogaak's dominance, design, and influence. Although the MC4 Top 8 appears relatively diverse, and although our finals was an Ancient Stirrings showdown straight out of 2018, make no mistake: Hogaak is the talk of the weekend. This deck will set the tone going forward into August's upcoming Modern Grand Prix events.
Post-MC4 Modern content is, and will continue to be, focused on Hogaak. This will inevitably include strong claims about Modern's health or lack thereof, Hogaak's suitability vs. bannability, Modern's graveyard fixation, Faithless Looting, and a number of related issues. These are important topics and questions for writers and readers to engage, but they need to start with an honest, accurate, and thorough breakdown of the MC4 itself. Those claims also need data to back them up. It is very easy to reduce MC4 to "ban Hogaak" and leave it at that, but we must not fall into this trap. Instead, we must carefully unpack the data bonanza that came out of this weekend. We must also understand the limits of extrapolating from the MC4 generally.
In that spirit, I'm going to tackle three topics today. First, I'm going to discuss the significant limitations of Top 8-level data, both to frame our MC4 analysis and as a caution for future events. Second, I'm going to identify key dynamics underlying MCs more broadly. This will help us understand the limits and context of MCs relative to events like GPs or MTGO tournaments. Finally, I'm going to bring it all together and establish some data-driven conclusions specific to Hogaak and graveyard hate at MC4. Here's a table of contents to help you skip around if you need to:
Looking Beyond the Top 8
The Top 8 never tells the full story. If you don't take any other conclusions from this article, remember this: T8s do not give complete pictures of formats or even tournaments. This is particularly true at MCs due to the six rounds of Limited (plus a bunch of other limitations listed later), but it's also true at a Grand Prix or StarCityGames Open level. In general, T8 standings are highly visible and paint an accessible narrative of an event. They make great soundbites and Wizards frequently cites T8s in both articles about a format and even banlist updates. Modern players and content creators also cite them when making claims about the format. In those regards, T8s matter and you should pay attention to these high profile standings as loose indicators of format direction. Unfortunately, for the most critical Modern consumers, there are just too many hidden influences below the T8 surface to draw meaningful conclusions. Here are just a few dynamics behind the T8 curtain you need to consider:
- Player luck: Magic is a card game with variance. Anyone can topdeck an out or flood for ten turns. You can mulligan to 4 in your Day 2 decider or never mulligan a hand all event. Luck is why players can draw perfect double Bolt to win an otherwise lost MC4 win-and-in. Luck is also a factor as to why Reid Duke can go 1-3 drop on Day 1 of 2018 GP Stockholm. Sometimes it's your day and sometimes it's not; lone T8s magnify this effect in big ways, even if multiple events will still show Duke to be one of the best players on the planet.
- Matchup luck: Over many events, you will beat the impact of matchup variance and reach your "true" performance level. Fun fact: the top 40 players have slightly higher Modern win rates than in Standard (61% vs. 58.4% as of late 2018), despite allegations of Modern being more variable. This is because over many events, you minimize the effects of bad matchups and reach your performance baseline, which for pros is high in both formats. This normalization doesn't happen at a single event, where you can register Jund and hit three Gx Trons and drop. T8s amplify matchup sensitivity.
- Human performance factors: It's hard to perform at competitions alone. It's harder still to perform when you're tired, hungry, thirsty, cramped from sitting in a chair all day, and/or just anxious about performance on a big stage. Add in thousands of hyper-critical viewers on Twitch that you know are watching and even the best can fall apart, committing heartbreaking misplays like poor Yam Wing Chun they would never repeat from the comfort of their living room. T8 standings can sometimes reflect who had the better breakfast or who performs best in the spotlight.
- Narrow tiebreakers: Players don't always win into T8z. Many get there based on tiebreakers to separate contenders with the same final point totals. For example, 2017 GP Las Vegas saw a 14-way tie for spots 6, 7, and 8 in that T8. Tiny margins can separate an ignominious 9th place exit from T8 glory, and this can change headlines if a T8 has no control decks but a UW Control and Blue Moon lurking at 9th and 11th behind breakers.
- Event format: Although GP and SCG Opens are examples of Modern-only events, tournaments like the MC or an SCG Invitational do not share that purity. Multi-format events may mean Modern decks reach T8 through player performance in Standard (an Invitational) or Limited (the MC). An MC player can go 6-0 in Limited and then 7-3 in Modern to beat out a 2-4 Limited/10-0 Moderner to T8. Similarly, Team events don't post individual records even if the whole team makes T4. A weak deck can get carried by stronger partners, sneaking to T8 without merit.
- Small sample size: This is a common thread throughout these dynamics, but it's always worth emphasizing. A single T8 is just one dataset of eight players. Anything can happen on any given day, and you only normalize data around these small N samples by aggregating multiple events. Eight decks will never make a large sample, even if they might make a compelling story.
Between these dynamics and the dozens of others I haven't listed (for examples, byes are hugely impactful on performance but don't apply to MCs), we simply cannot ever trust T8s at face value. They are important from an optics perspective, and may even suggest larger format conclusions, but they can be notoriously unreliable at telling the full story. Indeed, Modern's history is packed with examples of misleading T8s from all tournaments, especially MCs/PTs. Here are just a few examples you can show your Magic friends the next time they try to sell you a bill of T8 goods:
- Eldrazi at 2016 GP Melbourne: Remember Eldrazi Winter? It was bad. Really bad. When a single Eye of Ugin and Eldrazi Temple-powered monster makes up 25% of all major event T8s and 54% (!!) of all major event Day 2s in a 1-2 month period, you know you have a problem. Except, you wouldn't necessarily know if you just looked at 2016's GP Melbourne. In a winter overrun by Cthulhu's offspring, GP Melbourne saw just three Eldrazi variants in its T8; no worse than 2017 GP Las Vegas and its three Affinities. Never mind how heavily the other decks in that T8 warped around Eldrazi. Also, never mind GP Bologna and GP Detroit, held the same weekend, with five and six Eldrazi decks respectively. Wizards banned Eye a month later, but you might have underestimated Eldrazi's dominance looking at GP Melbourne's T8 alone.
- Dredge at 2016 GP Dallas: Golgari Grave-Troll was another stinker to come out of 2016, but it wasn't quite Troll's fault. The real blame goes to Wizards who, not content to give Dredge just one new potent toy in Prized Amalgam, forgot the whole discard-then-draw lesson of Bazaar of Baghdad in Vintage. Insolent Neonate and Cathartic Reunion joined the recurring Zombie and poor Troll, resuling in a righteous 2017 ban and a fall/winter 2016 of graveyard hate. But that outcome was not predictable based on the GP Dallas T8 just two months before Wizards swung the hammer. Dallas saw a single Grave-Troll in the T8 in the hands of Brian Braun-Duin, who sputtered out in the quarterfinals. You might have used the GP Dallas standings to predict the simultaneous Probe ban (three Infect in the T8 with 4 Gitaxian Probe each), but GP Dallas completely missed the Dredge mark.
- Tron at 2018 GP Lyon: 2018 was a notable year for Modern in that it saw no bans. It was also the first year since 2012 this happened (2019 is off to a rough start). But what if you were in February 2018 looking at recent Modern T8 standings trying to predict a ban? What if you saw a GP with a T8 so hideous it actually had more representatives of a single, top-tier deck than two of Eldrazi Winter's GP and any GP during Splinter Twin's 2015 run (max 3 at GP Pittsburgh)? Readers, I remind you of GP Lyon's T8 and its four Green Tron representatives. Of course, nothing from Tron got banned in 2018 or 2019, even if cries of WOW broke the internet. Tron also declined following GP Lyon, posting just three GP T8 appearances from July through December 2018 after posting seven through June. Again, the lone Lyon T8 missed the trend.
Given these dynamics and the historical examples, critical format consumers must always look beyond the T8 to assess a format. This is true regardless of the assessment's purpose. Predicting a ban? Look past the T8 because Wizards definitely has been since 2016. Trying to decide if a deck is successful? Don't put too much stuck in its lone T8 appearance; cross-check other events and lower brackets. The best times to consider a T8 are a) if you're trying to predict what shiny decks players might gravitate towards, b) if you're trying to weigh the public pressure on Wizards taking action due to visible data, and c) if you're aggregating multiple T8s across a longer time span.
Limitations of Mythic Championship Data
Important takeaway number two: just as Top 8s have limitations, so too do MCs. MCs, not to mention their Pro Tour predecessors, revolve around a number of unique dynamics that can make them poor predictors of the overall format. This may seem weird at first glance. If you get 400+ of Magic's best players in a room together and force them to play Modern for significant social and financial gain, you'd think the MC/PT picture would represent the pinnacle of Modern performance and deck selection. Historically, however, this has not been the case. Before we get into the reasons for this, let's look at two classic MC/PT misses of many pros getting it wrong.
- Abzan at 2015 PT Fate: Wizards banned Splinter Twin and the broken Summer Bloom a year after this PT, and both were legal at this Pro Tour in more or less their banned form. Twin even went on to win the PT. Amulet Bloom lost in the finals. So what was the pro deck of choice going into the tournament? What strategy had the dominant share all weekend? Not Twin, not Amulet Bloom, but Abzan. Abzan made up an outrageous 28.3% of Day 1 and a perfect, expected 28.3% of Day 2. Although two Abzan pilots did make T8, Abzan quickly lost relevance, posting just a single GP T8 all year after February 2015. Twin and Bloom, however, went on to define Modern, ultimately resulting in bans. Of course, nothing from Abzan got banned, despite its 28% Day 1/Day 2 shares and its two T8 representatives.
- Eldrazi at 2016 PT Oath: We already know Eldrazi Winter was a nightmare, but you wouldn't know just how bad it would get looking at PT Oath's Day 1 and Day 2 standings. In an event with a comical six Eldrazi variants in the T8, you'd expect a comparable 50%+ of players to be running the deck into that T8. This was not the case. More players brought Affinity (51), Burn (51), and Infect (32) on Day 1 than Eldrazi (32). More players even played BGx Midrange variants (Abzan 20 + Jund 18) than were on Eldrazi variants! The Day 2 picture didn't change much, with Affinity (33) and Burn (30) still ahead of Eldrazi (26), which was virtually tied with Infect (25) and the BGx collective (Abzan 10 + Jund 13). Those pros likely felt pretty silly missing a deck ultimately more dominant in Modern than Caw Blade in 2011 Standard. The Day 1 and even Day 2 statistics alone did not predict just how broken Eldrazi would be.
As we see, pros can miss the mark, especially at MC/PT-level events. That doesn't mean we shouldn't observe MC/PT results and it doesn't mean we shouldn't follow their potential trends. It does mean we need to be aware of the limitations of these events and the reasons pros choose decks or perform in certain ways. Here are the dynamics which we must consider when extrapolating MC results to the broader Modern picture:
- Draft rounds: This is the biggest issue with MC-level data. Limited accounts for 6 rounds of the rough 13+ wins needed to hit T8. It takes just a cursory look at the 24+ point Modern decklists to realize we'd have 4+ Hogaaks in the MC4 T8, not one, if we excluded Limited rounds. These rounds even affect who gets to Day 2 to begin with, especially if you go 0-3 in Limited and need to now win at least 5 in a row to make the cut, vs a pro who went 3-0 and now just needs to finish 1-4 in Modern.
- Event size: Only 400-500 players attend MCs, compared to GP scenes with thousands of players. They then only play 10 rounds of Modern instead of 15. Fewer players mean a more predictable field of known personalities. Fewer rounds means you can run hot shorter and still make it to the top. As with T8s, sample size matters and MCs have a smaller player and round sample.
- Team testing: Many top players attend the MC as part of teams. They also test and sharpen decks in team settings leading up to the MC. This results in refined, highly iterated lists that entire teams might register for the event. The team effect homogenizes metagames towards decks that whole teams agree to play and hone.
- Expected field: It is virtually impossible to predict a GP or Open field. You simply don't know all the personalities and what they might show up with on any given weekend. This is not the case at MCs, where you know exactly who is attending before you even register a list. You also know what teams are forming, and you might have a good sense of what lists they are testing. This lets pros tailor their maindecks and sideboards to expected breakout strategies, narrowing fields relative to broader public events.
- Modern expertise: Some pros vocally dislike Modern. Other pros enjoy Modern but have to split time between Modern, Standard, Limited (necessary for the MC), Arena, team events, etc. With more non-Modern than Modern events in a year, pros must commit less time to Modern to maintain success. This means many pros go into Modern without the rich and deep experience of players like MTGO's shoktroopa, who has played Mono U Tron almost exclusively for years. Pros in this category will gravitate towards higher profile decks that are obviously powerful, and are more willing to defer to a team's (or even the vocal public's) evaluation of a strategy.
- Incentives: If you fall flat at one of the half dozen Modern GP in a year, there's always next tournament. This means you can try out new tech and brews you've been testing on MTGO (a venue where there's even lower stakes and higher incentive to innovate). Some pros just need to post a certain GP record just to gain planeswalker points. MCs have higher financial and social stakes on the line. There's a big difference between winning a GP and earning $10,000 and winning an MC and earning $50,000. This pushes pros to play decks they are supremely confident in, even if that confidence is misplaced, or perceived safe bets, if they can't decide on a breakout strategy.
- Open decklists: This is a huge change at the MC relative to the PT with major implications for how decks are selected and built. MC4 players had approximately one minute to look at an opponent's full maindeck 60 and the cards in their sideboard (names but not exact quantities). This affects mulliganing decisions and, when coupled with the London Mulligan, allows more informed mulligans to even maindecked hate. It also makes rogue decks significantly worse because you can't steal that Game 1 off a weird strategy. GP and other events don't reveal this kind of critical information.
Just as with T8 limitations, there are many other factors that likely impact MCs (e.g. overall higher skill level, pro gravitation towards certain deck styles, etc.). This list is still a good starting point to illustrate the tricky dynamics influencing everything from Day 1 deck choices to final MC standings. As we explore our Hogaak and graveyard hate takeaways today, we need to keep the above limitations in mind when translating MC4 conclusions to the broader Modern format.
MC4 Datasets and Sources
Last section before we get into the takeaways, and it's a short one. I cite a lot of numbers throughout my conclusions, and here are the various datasets those numbers draw from. I strongly encourage you to just steal my citations and reference these sources heavily in your own arguments. We need to have a numbers-driven debate on big Modern issues. Only by drawing on sources like these will we get out of the one-liner meme wars and reductionist arguments. Also, give credit to these content creators when citing their work. Data collection and presentation is time consuming; the least these folks deserve is a shoutout.
- Day 1 and Day 2 breakdowns by Frank Karsten
- All Day 1 Modern decklists by Wizards event coverage
- 8-2 or better Modern decks by Adam Styborski
- 7-3 Modern decks by Adam Styborski
- 6-4 Modern decks by Adam Styborski
- Matchup win percentage (MWP) breakdown by Frank Karsten
- Matchup MWP (MMWP) matrix by Redditor heyzeto OR MMWP matrix by Redditor flaflafl.
- Deep MWP analysis by Redditor flaflafl
Top 8 and MC limitations? Check. Datasets? Many checks. Now let's get into the takeaways.
Hogaak Dominated. Full Stop.
As Redditor bamzing so eloquently stated, maybe Wizards should have gone for the head. Quality crap-posting aside, Hogaak strategies decisively outperformed the competition at MC4 by most available metrics. By now, this takeaway may not be surprising to anyone who has engaged any Modern media in any online form, but it is still important for us to understand exactly which numbers indicate Hogaak's overperformance. This will ground future Hogaak discussion in auditable statistics we can widely share with others when making arguments. If Wizards ultimately bans Hogaak, these numbers also serve as a benchmark for future red flags.
Highest prevalence at most MC4 levels
If you've read my articles before, you know I prefer performance metrics (i.e. how often decks win/lose) vs. prevalence metrics (i.e. how often decks show up) when we have performance numbers available. Performance is a much better indicator of success than mere prevalence; the latter often reflects perceptions of popularity and power, rather than actual power. But high prevalence can be correlated with high performance, and in Hogaak's MC4 case, its prevalence was very high. Hogaak's Day 1 share of 21.4% more than doubled the next most-played strategy (Izzet Phoenix at 10.5%), maintaining that lead going into Day 2 (24.2% vs. 10.7%). This continued into the 6-4+ and higher Modern-only standings, where the prevalence of all decks with 5+ pilots speaks for itself:
HOGAAK SMASH. Commanding shares like this are heavily correlated with high performance, as we're going to see in a second. It's also easy to chop these shares into really disgusting quotables . At the 6-4+ level, Hogaak's share was virtually equal to that of the next three most-played decks combined (Izzet Phoenix, E-Tron, and Jund). At 7-3+, that share was now almost equal to that of the next four most-played strategies (the same three plus Urza ThopterSword). Finally, at 8-2+, Hogaak and Hogaak Dredge made up 55% of all 20 decks. All of this more than outweighs the one Hogaak in the T8, particularly with our established T8 limitations. Want one more quotable? There was more Hogaak at 8-2+ at MC4 than Eldrazi variants in the same bracket for PT Oath 2016: 55% vs. 40%.
Best Day 1 to Day 2 conversion
Hogaak started MC4 with a 21.4% (98 players) Day 1 share, jumping to 24.2% (70) on Day 2. This means 71.4% of the Day 1 players converted to Day 2 off both their Modern and Limited records. It also represents a +2.8% point increase over expected conversion (expected conversion meaning 21.4% on both Day 1 and Day 2). The next best converters of decks with more than 10 pilots on Day 1 were Eldrazi Tron (9.2% --> 9.7%: 66.7% conversion, +.5% points) and Burn (3.9% --> 4.2%: 66.7% conversion, +.5% points). For decks with 20+ pilots, it was Eldrazi Tron and Izzet Phoenix (10.5% --> 10.7%: 64.6% conversion, +.2% points). Hogaak converted +5% more pilots than the next best deck, and overperformed its Day 1 expectation by +2.8% points relative to E-Tron at just +.5% points.
Best conversion through winner's metagame
I define the winner's metagame as all decks with 18+ points/6-4+ records in the Modern-only rounds. We've already seen Hogaak's prevalence across those brackets above. Now let's look at its conversions from Day 2 all the way up to the 8-2 standings. The smaller, grayed columns show prevalence percentages for reference, but pay attention to the conversion rates. Try to find the two decks that positively converted in every single bracket from Day 2 all the way to the top:
One of the best converters is Urza ThopterSword, which I'll note here and later in this article was a secret winner in this event. The other one? HOGAAK SMASH. AGAIN. Not only did Hogaak have three positive conversions from Day 2 all the way to the 8-2+ bracket, but the conversions get better each time! Hogaak was the deck to play if you wanted to win at Modern. After Urza and Hogaak, the next most consistent converters were Jund and UW Control. The worst converters were Eldrazi Tron (which lost conversion points at every increasing bracket) and Burn (same problem).
Highest overall match win percentage
Quick note on match win percentage (MWP) before we get started: all of the major MWP sources have slightly different numbers due to slightly different coding/tabulation methods. They're all in the same ballpark so I'm going to use them interchangeably, citing sources for specific points. It shouldn't be surprising Hogaak had the best MWP at MC4. Of decks with 20+ pilots, Hogaak was the dominant frontrunner at 56.2% followed by Jund (52.3%), Izzet Phoenix (50.8%) and E-Tron (50.4%). Even if we extend to the decks with 10+ or even 5+ pilots, decks with smaller Ns that could see higher MWP just by virtue of a lower sample, Hogaak is still at the head of the pack. In fact, Karsten's dataset shows only one deck with a higher MWP than Hogaak at any sample >5 pilots, and it's still a Hogaak deck! Hogaak Dredge posted a 60.4% MWP which literally made me laugh out loud. The only deck in the entire tournament with a better MWP was Esper Control (61.4%) with a statistically meaningless N=2.
Readers with even a little statistics background will know MC4 MWPs don't necessarily represent the "true" MWP of a deck over a larger, N=10,000 dataset. There's a +/- margin around these MWPs, and Redditors heyzeto and flaflafl include the confidence interval around their own MWP calculations, estimating with 95% confidence the "true" MWP lies within a certain range around the calculated MWP. In both of their datasets, Hogaak was the only deck where the lower end of their confidence interval was greater than 50%: for heyzeto, 52.1%, for flaflafl, 50.5%+. This is particularly dominant because all other decks could statistically hover in the 45%-55% range or even wider. Not Hogaak, which is statistically likely to have a >50% win-rate against the field.
Most favorable matchup spectrum
I've always called matchup match win percentages (MMWP) the gold standard of Magic data analysis. Assuming a high enough N, it is the best predictor of deck success in a known or predictable field. It's normally difficult to find this data, leading users like myself and other content creators to build their own datasets to crunch MWP numbers. MC4 is an exception, with comprehensive MMWP data available to anyone who can cross reference Day 1 decklists with all the pairings/results pages. Thankfully, our enterprising Reddit heroes have done this work for us. Less thankfully, Hogaak is still on top.
The table below crudely combines all <40% matches ("bad matchups"), 40% <--> 60% matches ("even"), and 60%+ ("good") for all decks. Of course, many of these matches have wide confidence intervals so we don't truly know if the calculated MMWP is accurate. For instance, the Burn vs. Dredge matchup could be anywhere from 7% up to 70% in the 2-4 (n=6) sample. But given what we already know about MC4, we're still going to see a familiar pattern that a sampling issue can't distract from:
Heeeere's Hogaak! Coming in with no calculated bad matchups and every matchup in the even or better range, Hogaak again emerges as the top Modern contender. Also, as I mentioned in the conversion section, Urza ThopterSword is looking pretty hot too. Sleeper MC4 champion! If we look at discrete MMWP calculations, Hogaak is also near the top for decks with the most 50%+ matchups: six (51.6% vs. Izzet Phoenix at worst up to 69.6% vs. Dredge). Modern's traditional top decks like Tron, Humans, Izzet Phoenix, and UW Control weren't even close, all posting much less favorable MMWP bands.
Conclusion: Hogaak is scary
Individually, no one of these statistics would be cause for alarm, and may even be blunted by the MC4 limitations we discussed earlier. Collectively, however, these numbers paint a disturbing picture of Hogaak dominance at an event where many pros knew it was coming and packed absurd amounts of graveyard hate (more on that right around the corner). I'm going to withhold opinions on whether or not this is a format problem until the end, but it was indisputably dominant by every measure except the meaningless T8.
Rampant Graveyard Hate
We already know Leyline of the Void was MC4's most-played card (836 total copies), second only to graveyard enabler Faithless Looting (747). We also know graveyard hate is one of the better mechanisms for fighting Hogaak and Looting decks generally, so assessing the level of MC4 graveyard hate can give important context to Hogaak's performance and the overall event. Although Wizards does not provide sideboard numbers for all Day 1 decks (they just indicate 1 copy of every card), they do provide full maindecks. They also give sideboard numbers for the 6-4+ decks. We can analyze those to assess how much graveyard hate was present at MC4 at various levels. Before we get started, here are the top graveyard hate cards we'll look at:
Top decks with maindeck hate had mixed performance
Let's start with maindeck graveyard hate. There's a narrative that MC4 saw unprecedented levels of maindeck hate critical to deck success, but the numbers only give mixed support for this theory. Only a few decks in the entire MC4 used maindeck hate, and only a few of these were top strategies with a large enough N to draw conclusions from. Given this, I'm going to focus on the following decks and their common maindecked hate: Hogaak and Leyline of the Void, UW Control/Izzet Phoenix and Surgical Extraction, Jund and Nihil Spellbomb, and G Tron and Relic of Progenitus. This excludes random Mardu Shadows and Titanshifts with too small of a sample, and some outlier lists with alternate graveyard hate. For example, two UW Control players maindecked 4 Rest in Peace but never made 6-4 or better. It also excludes stuff like Eldrazi Tron with bullet Scavenging Grounds.
Let's start with the decks benefitting from maindecked graveyard hate. The tables below show the average hate for decks choosing to maindeck them, the number of Day 1 decks with that hate, and the percentage of total Day 1 decks within that overall strategy which used maindeck hate. So for Hogaak, you'd read the table as: Hogaak decks with MD Leyline averaged 2.5 copies. 24 Hogaak players maindecked 1+ Leyline on Day 1, representing 24.5% of all Hogaak pilots. From there, the tables show the percentage of 6-4+, 7-3+, and 8-2+ decks of that type using maindecked hate. For Hogaak, that would mean 34.2% of 6-4+ Hogaak decks had maindecked Leyline, up to 36% for 7-3+ decks, up again to 44.4% for 8-2+ decks. Given that, here are maindeck graveyard hate winners:
Hogaak, Izzet Phoenix, and G Tron decks all benefited from maindeck hate at all tournament levels. At each increasing performance level (i.e. Day 1 to 6-4+ to 7-3+ to 8-2+), we see increased shares of graveyard hate. This overrepresentation generally signals it was the more competitve choice. Using Hogaak as an example, only 24.5% of Day 1 Hogaak pilots had maindecked Leyline. But if we look at the 8-2+ bracket, Leyline Hogaak pilots make up 44.4% of the total share. This is a 20% point overrepresentation, suggesting this maindeck hate was the way to go. The same conclusions apply for Phoenix and Tron, admitting those decks had high Day 1 maindecked hate to begin with. If you play these decks, play maindeck hate (assuming similar MC4 metagame conditions; remember the MC4 limitations!). A notable MC4 limitation with Hogaak is open decklists. If you're playing the best deck in an open decklist environment, you can afford to maindeck 2-3 Leyline because you know when to mulligan for it in the first game. This is not true in most other events, so this is one of the least translateable conclusions of this article.
Not all decks unambiguously benefited from high shares of maindecked hate at MC4:
Jund and UW Control did not necessarily need the kind of specific, maindecked hate as Hogaak and company (we'll talk about sideboards shortly, and it's a very different story). For UW Control, maindeck Extraction builds had expected conversion from Day 1 to 6-4+, but then saw decreased relevance en route to 8-2+. For Jund, the drop was sharper: 31% of Day 1 decks used maindeck Spellbomb down to 16.7% in 7-3+ and 0% in 8-2+. I'll note Jund has non-Spellbomb hate in Scavenging Ooze which distorts this picture, so I'll just say Jund mages should probably stick to the maindecked Ooze and maybe avoid adding additional graveyard bullets. As for UW Control, those Extractions might be less necessary than you think. Final note on maindecked graveyard hate: open decklists played a huge role at MC4 in the decision to play these cards. I imagine the "true" maindeck hate number, or at least the correct one, for non-MC4 events should be lower. Sideboarded and total graveyard hate, however, matter at all events.
Increased total graveyard hate increased success (to a point)
Let's add that sideboard picture and combine it with the maindeck hate. To do this, we have to limit our analysis to 6-4+ deck standings; we don't have exact sideboard numbers for Day 1. That's fine with me because I'm more interested in the winner's metagame anyway. Before we talk about how different levels of total graveyard hate performed, here's the overall breakdown of average maindeck, sideboard, and total graveyard hate pieces across all decks in the different 6-4+ or better brackets:
These descriptive statistics shouldn't be too surprising. On average, decks had roughly 1 piece of maindeck hate, 3 pieces of sideboard hate, and 5 pieces of total graveyard hate in their 75, with piece #4 more frequently in the board than the main 60. Of course, averages only tell part of the story. We really need to see a frequency distribution to know if this is an average clustered around the middle, or a big spread that just averages to that median value. Here's the frequency of total graveyard hate (main plus sideboard) for all decks in the winner's metagame:
Note these are not even brackets: I've left 4 hate cards distinct and consolidated 0-3 hate, along with 5-6, 7-8, and 9+. This just gives the most graphically clear depiction of the breakdown. As we see, taken in aggregate, most decks chose to use 4 pieces of total hate. This was increasingly important in the 8-2+ bracket, where 65% of decks went with the full 4+ pieces. That said, there were slight diminishing returns on pieces of hate 5+, as we see more 6-4+ or 7-3+ decks using those packages relative to the 8-2+ success stories. This suggests 4 hate pieces is the sweet spot for most decks.
But even this doesn't capture the full picture because MC4's 6-4+ decks aren't all from the same archetypes. I hypothesized aggro/combo/control/midrange/big mana decks would all require different levels of graveyard hate to succeed at the MC4, splitting up the frequency distribution by decktype. I loosely combined the decks into group. I defined aggro/combo as predominantly proactive, critical mass strategies. I defined control/midrange as mixed reactive/proactive strategies that try to answer opposing threats while also protecting their own. I defined big mana as decks going over the top of other strategies through mana advantage. Based on that, here are the groups I created:
Humans (Humans has interactive components like Meddling Mage, Kitesail Freebooter, Reflector Mage, and Phantasmal Image to copy these effects, but it's primarily a proactive, critical mass aggro tribal deck)
Urza ThopterSword (I gave more weight to its combo elements and primarily proactive gameplan than its reactive elements; switching it to another category didn't seriously change the final picture)
Izzet Phoenix (I'm aware this deck can play more reactive, tempo-style Magic, but its primary gameplan is often very proactive with no discard or countermagic to back up only limited burn spells)
Mono R Phoenix
Eldrazi Tron (This deck can play a midrangey game plan, but between Temple and Tron assembly, it's very much a big mana spin on a midrange secondary theme)
Given those definitions, here are the graveyard hate breakdowns by macro-archetype:
Aggro/combo is heavily clustered around the 4-card total hate strategy, with fully 80% of the 8-2+ decks in this category using this package. Aggro/combo decks with exactly 4 hate pieces did better in increasing brackets from 6-4+ all the way to 8-2+, further suggesting this is the optimal strategy going forward (assuming MC4 conditions, of course). What about control/midrange?
Totally different picture. The majority of control/midrange players were on 5-6+ pieces of total hate across their 75, and the data suggests the sweet spot was actually a whopping 7-8 hate pieces. Decks with 7-8 pieces only made up 25% of the 6-4+ decks as a whole, but when we narrow it to the 8-2+ decks, they represent 50% of all the interactive strategies that made it that far. Remember the control/midrange population is only N=20, so take these conclusions with a sample size caveat. This all points to significant graveyard pressure on interactive control/midrange strategies if they are to succeed in MC4-like environments. Finally, we end with big mana:
The N=21 big mana decks (again, admitting small sample size) needed a full 5-6+ hate pieces to even be competitive in the 6-4+ bracket. The winning strategy seems to be 7-8 pieces, which saw increasing conversions from 6-4+ into 7-3+ all the way to 8-2+. This makes sense to any longtime big mana player who knows graveyard matchups can be difficult. I am personally comfortable with this pressure because graveyard strategies and similar aggressive builds are natural checks on big mana, but 7-8 may also be a lot higher today than in past metagames.
In summary, total graveyard hate definitely mattered at the MC4, but it mattered in different ways for different decks. Aggro/combo strategies were able to get away with 4 pieces of hate in most cases. Other strategies needed 7-8 to maximize their success. This is a testament to just how much pressure graveyards placed on decks at MC4.
I cannot emphasize the limitations of MC4 data enough, which is why I have not just one but two whole sections on those limitations. MC statistics and conclusions may not translate to public metagames. August's GP/MTGO scene may look totally different than the MC4. That said, the overwhelming Hogaak dominance at MC4 and the heavy graveyard pressure at that event is unlikely to fade soon. I suspect we'll see a million Hogaaks, Carrion Feeders, and friends at GP all month long, and we'll see this dominance continue in varying degrees. But remember, even in Eldrazi Winter, GP Detroit saw just three Eldrazi in the T8 of an otherwise horrific triple GP weekend. Don't cherrypick single T8s from the coming month as evidence of anything, as we could also see similar scenarios play out this month even if Hogaak remains monstrously broken.
Finally, just to re-re-re-emphasize a disclaimer I've made several times in this article, sample sizes must limit the confidence of our conclusions. MC4s already have smaller samples to begin with, and once we start cutting them down into smaller analysis units, N drops even more. In case it was unclear in my conclusions, all conclusions are specific to available MC4 data, admitting the limitations of the sample.
On Bans and Next Steps
It would be disingenuous for me to wrap up this article without discussing some banlist discussion points running through the community. Both Hogaak, Arisen Necropolis and Faithless Looting have seen significant player and pundit attention these last weeks. Twitter user Emsizz hilariously captured some of the anti/pro-Looting chatter in the highest quality crap-post I've seen in months.
Twitter hilarity aside, these are real questions to discuss. Personally, I am withholding all judgment on non-Hogaak elements until after the Hogaak situation resolves. We have no idea what post-WAR/MH1/M20 Modern looks like because Hogaak has been slobbering corpses all over the format since late May. Anyone who says comments like "If you ban Hogaak, Deck X will rise up in its place using Looting and we'll have to ban something there too!" is selling a bill of ban mania goods. People are bad at trying to predict and next-level the metagame at single tournaments, let alone after potential bans and multiple new sets. We need to reserve all non-Hogaak ban or format health judgment until after August.
Then there's Hogaak. The numbers today are extremely convincing but limited by the MC4, even if they have many shades of Eldrazi Winter take two - Hogaak Summer: A Bridge to Horror. Given the many MC4 limitations, and wary of setting a precedent to ban anything based solely on limited/niche MC4 results, I still recommend we wait until late August to assess Hogaak's influence on the metagame. I suspect it will not be different from the MC4 picture, because the deck is either going to continue its dominance or push decks too heavily to graveyard hate. And even then, as we saw at MC4, it might still dominate despite concentrated hate. I predict another ban in late August at this rate, ideally at the primary offender Hogaak itself, and then we can figure out where to go in Modern after that.
Finally, for those who want a non-Hogaak or non-graveyard takeaway today, I'll strongly endorse Urza ThopterSword as a sleeper hit at the tournament. It posted strong conversion rates, an overall MWP comparable to Hogaak's, a positive matchup spectrum comparable to Hogaak's, and an overall gameplan lending itself to varied axes of attack. I expect this deck to continue to perform in the future.
Thanks for tuning in today to parse through this lengthy data deep-dive into Modern's biggest issues. Feel free to PM me here, find me in the MTG Nexus or Reddit threads, or Tweet me (https://twitter.com/ktkenshinx) to discuss the article and takeaways further. Also, if you notice any errors in this long article brimming with numbers, please notify me! Try to enjoy August Modern and don't let the Necropolises bite...