For the first time in quite a while, I have a massive data dump to analyze! The past weekend contained a Modern GP, an SCG Open, and the Mythic Championship. With all the shocks finally wearing off, we can start to get a handle on the new metagame. But not without first noting that this is just the first weekend of data. Metagame formation is a process of players adapting to data as it trickles in, so today's snapshot should be regarded as a starting point rather than the actual metagame.

Additionally, the starting population for each event matters significantly. The Mythic Champtionship is an invitational event with a small population of high-level players who often play team decks. Almost all of Team ChannelFireball and the MPL was on Hogaak, an observation neither random nor necessarily indicative of anything other than player preference. GP Barcelona would normally have been a random sample, but this time it was flooded by players scrubbing out of the Mythic Championship. As a result, the data will be distorted to some degree. The SCG Open and associated Classic results should therefore be given greater weight, since they are the most random and therefore statistically valid results.

MC Barcelona

As it was the most high-profile event, I'll start with the Mythic Championship. As its data is also the most skewed, I'll be treating it lightly. As mentioned, team-think and word of mouth heavily influence deck decisions at this level. There's also the impact of the draft portion. Luis Scott-Vargas only managed two match wins with Hogaak, but still made Day 2 thanks to his draft record. His luck didn't improve. Thus, any player's final placing is not necessarily indicative of their Modern deck's strength.

Top Performers

As a result, I'm discounting the Top 8, and will instead focus on only those decks that earned at least 24 points. It should be noted that Thoralf Severin, the winner of Mythic Championship IV who was playing Tron, is not in this data set. In fact, of the Top 8, only Zhiyang Zhang on Jund and Martin Müller on Hogaak received at least 24 points in constructed. The draft bias is very strong in this event.

That is an overwhelming amount of Hogaak. The picture gets worse if you consider the Dredge version just a variant rather than a separate deck. Over half the top decks contained Hogaak, Arisen Necropolis. It certainly looks like the banning of Bridge from Below has not actually released us from the rearisen menace. However, there are caveats. Look beyond the deck title and there is a lot of very high-level talent playing Hogaak, including PVDDR and Reid Duke. I have a hard time believing they wouldn't have done similarly well with another deck. Again, the effect of teams picking the same deck is strong here. Hogaak's prevalence is driving most of the discussion of Modern's metagame as a result, so that's what I will be focusing on today.

Overall Win Percentage

Trying to look at the wider picture muddies the picture further. If you look at the overall win percentages, Hogaak is doing far better than any other well represented deck. It gets even better if you lump the various versions of Hogaak together. This would serve to justify the implication from above that Hogaak is an overperforming deck. However, that assessment is complicated by the starting population. Hogaak was the most popular deck Day 1. It held roughly the same proportion of the field on Day 2. In fact, the overall metagame really didn't change over the tournament. This makes the win percentage numbers somewhat suspect.

Consider this thought experiment. At a Grand Prix, there is a metagame made of three decks: A, B, and C. Deck A beats B 75% of the time, and C 25%. The mirror is 50%. At this GP, A forms 50% of the field, B forms 20%, and C forms 30%. The expected win rate at such a tournament for deck A would be around 48%. If in practice, deck A hit the mirror seven rounds, B three times, and C five, the average win rate would be 47%. If instead it saw B five and C three, the win rate would be 53%. The prevalence of mirror matches skews the data towards the center. This tends to mask the actual strength or weakness of a deck in the actual metagame. Thus, I argue it's not necessarily accurate to say that Hogaak overperformed given it's starting prevalence.

The bottom line is that Hogaak did very well in the constructed portion of MC Barcelona. It should have done well, considering the quantity and quality of players that were running it, regardless of its actual metagame strength. Therefore, I wouldn't draw any conclusions from this chunk of data due to the various biases. As part of a wider context, it may have greater meaning by reinforcing other, more valid data...

GP Barcelona

...such as the Grand Prix that was run parallel to the Mythic Championship. Normally, GP's are the best data source since they're large, open tournaments, giving a very large and random starting population, which is critical for validity. The Barcelona results need to be taken with a grain of salt because it did receive an influx of players from the MC, and thus received some of the bias from that event. I don't expect it was enough to completely swing the results, but it still needs to be acknowledged.

Top 8

Unfortunately, being run in parallel to the MC means that the GP wasn't as well covered. Looking at the Top 8 is not enough data to really mean anything. However, that small set does have a very interesting parallel to the MC, which suggests that all is not what it seems given the data.

Hogaak is just part of the crowd in both Top 8's, and in fact neither one made it past the quarterfinals. Meanwhile, Jund is the best performing deck by Top 8 appearance and won the GP. Hogaak didn't win anything last weekend. If this was the only data to work with, the conclusion would be that Hogaak was just another deck and there's nothing really to see here. Instead, the focus would be on Jund's return to viability after disappearing for over a year. Of course, there's insufficient data to make such a declaration, but the point remains that the GP's final result appears at odds with the MC's data while agreeing with the (biased) Top 8.

Looking further into the GP's Top 16 verifies the (suspect) conclusions from the Top 8. There's no Hogaak in there. There's no Jund, either. Instead, there's Bogles and Neoform combo alongside some more standard decks. This doesn't look like a Hogaak-defined and -dominated metagame. It looks like players just running their normal decks, especially considering there's not a particularly heavy concentration of graveyard hate in the decklists. If the doomsaying was correct, then the lack of hate would have let Hogaak overrun the GP. That it didn't happen is instructive.

Day 2 Metagame

Digging more into the GP would tend to confirm that assessment. Going by the Day 2 metagame, Hogaak underperformed. It was the most popular individual deck by a decent margin (16.7% vs Eldrazi Tron's 11.1%). For being the largest single piece of the metagame, Hogaak didn't convert accordingly. Jund on the other hand started out at only 6.5% but took two slots. Once again, this pushes against the apparent absurd strength of the deck observed by the MC's constructed standings.

SCG Columbus

This brings me to the most random and therefore valid of the events, the SCG Columbus Open. This event, and its associated Modern Classic, are the best indications of last weekend's metagame because they're the least affected by the bias caused by the MC. It should be noted right off the bat that Mono-Green Tron won the Open, just like the MC. Forget Hogaak's numbers; the real story seems to be Tron winning everything.

Top 21

Since there were 21 decks with at least 24 points in my MC Barcelona sample, I'll look at the Top 21 decks from Columbus, too.

This table looks surprisingly similar to the top decks from MC Barcelona and a reversal of the GP results. Hogaak is outstripping the field by a decent, though reduced, margin. Everything else is clustered together. The composition is also more varied, though that's not necessarily surprising; pro players tend to gravitate towards known decks where enthusiasts actually create new and interesting decks.

In a further deviation, there's no Jund in this sample. This is shocking considering that it was the fourth most popular deck Day 2. Just like in Barcelona, Hogaak was the most popular individual deck in Columbus, though not by a wide margin. Given its Top 21 numbers, the result would support the Hogaak-as-overperforming narrative, suggesting that it is warping and defining Modern.

However, looking through the decklists convolutes this take. The amount of graveyard hate among the Top 21 decks varies wildly, to the point that there's no real pattern. The winning Tron deck has maindeck Relic of Progenitus and sideboard Leyline of the Void. Fourth place Izzet Phoenix has no graveyard hate. The sixth place Tron has the same maindeck as first place, but only one Grafdigger's Cage and a Tormod's Crypt in the sideboard. The 15th place Eldrazi Tron deck has Leyline of the Void, Grafdigger's Cage, and Tormod's Crypt in its sideboard. The relative success of each non-Hogaak deck doesn't appear to have been influenced by how prepared they were for Hogaak.

The Classic Asterisk

Looking at the parallel event tends to back up this assessment. In another reversal, the Classic data agrees with that of the GP.

Again, Hogaak is a good deck, but it's not an exceptional deck; just part of the pack. More than that, Hogaak didn't even Top 8, finishing 14th and 15th. That's not a great result for the supposedly metagame defining deck. Also again, there's not a particularly high amount of hate amongst the decks. Other than Hogaak itself, this appears to be very normal metagame. This just doesn't mesh with the notion that Hogaak is crushing Modern.

A Lingering Question

The implication of the Open and MC are that Hogaak is a busted deck that easily plows through hate, of which there was a particularly high concentration at the Mythic Championship. This is a strong argument that the problem that the last ban was meant to fix hasn't been fixed, Modern is still unhealthy, and another ban is necessary. Conversely, the Classic and GP indicate that it's just another deck in Modern. There wasn't a particularly high amount of hate in the observed data and Hogaak didn't do particularly well. Also worth noting, it was Tron, not Hogaak, that was the big winner of the weekend, with two trophies to none. The results all appear to contradict themselves, and that makes drawing any conclusions a dicey proposition.

Meaning Obscured

I suspect that the actual truth is that the contradiction itself is true: Hogaak is a very powerful deck that will either dominate a tournament or have no real impact. I have a theory about why that is, but I'm still gathering data to verify my thinking. I'll discuss it in detail next week. In the meantime, I've always said that graveyard hate is underplayed in Modern. Regardless of the actual threat of Hogaak, don't skimp.