I know, I know—my timing is a little off. The Modern community has moved on from the land of ports to the land of oaks. But for last week’s article, I was working under the threat that the bot behavior on MTG Arena could change at any time. So I simply had to get my solution to its Draft format out there while the take was still hot/true.
As you might be able to imagine, sorting and analyzing massive amounts of data also takes a while, so I can’t focus on data from Grand Prix Oakland quite yet anyway. Don’t worry, I’ll get to that next. In the meantime, let’s look at matchup results from the previous Modern Grand Prix in Portland, which took place about a month ago. As an added benefit, this should leave us with an excellent baseline with which to compare Oakland’s data!
I’m again working off of the same data set that I used two weeks ago: 971 main event deck lists, more than 55% of GP Portland’s total entries. For this article, I cross referenced results with decks to figure out against what the major archetypes won and lost.
All in all, I know for 7,187 matches across Day 1 and Day 2 which deck the winning player ran. For 4,224 of them, I know the deck choice of both players. I’ll focus on the latter and reference a deck’s overall record whenever that’s noteworthy.
Note that Modern has a ton of supposedly viable decks, and they look viable precisely because anything can win on a good draw. Many pairings only happened a couple of times. These match results barely provide any insight with regard to the underlying matchup.
For example, I could list Spirits’ 1-2 record versus 8-Rack, its 1-2 against Abzan Midrange, its 3-3 against Affinity, and its 2-4 versus Amulet Titan, and we’re only down to the first letter of the alphabet here. But I don’t think we should draw any conclusions based on that. Because I needed to make a cut somewhere, I decided to disregard all results from three or fewer encounters. Likewise, for pairings that occurred between four and nine times, I only included the results if they favored one deck over the other by upward of 70%.
This leaves us with matchups that appear to have a clear winner based on a somewhat reasonable sample size, with matchups that have a likely but unclear winner based on more fragile evidence, and with matchups that look about even based on a double-digit number of matches.
The Spirits of Portland
Spirits was by far the most represented archetype in my sample and, by reasonable extension, very likely to have been the most popular deck in the tournament as a whole. As such, it generated the most match data too.
The top half of the spectrum from good to bad doesn’t hold many surprises. The combination of a fast clock with disruption and punishing sideboard options makes Spirits a deck Ironworks players don’t want to face too often. That Spirits controls the tempo versus White-Blue Control is similarly in line with expectations, as is that it has a harder time versus the Jeskai counterpart.
We have to look for the news at the bottom here, but boy do we find news. All of the decks with Arclight Phoenix gave Spirits a more or less severe beating!
When I looked at metagame shares, I was surprised to find green-based Tron holding out in second place, even after what seems like years of new hate cards added to the Modern card pool. But its results don’t look all that surprising to me.
The deck had always been known to prey on traditional Jund and control. During Death’s Shadow’s heyday it was famous as a problematic matchup, and I don’t consider its perfect record versus Spellstutter Sprite to be much of a spike either. Against White-Blue Control with its Field of Ruin, by the way, Tron went 4-4.
Tron’s limited options for interaction leave it the ideal victim for Storm’s game plan. Fringe strategies like Ponza, Mill, and Merfolk, meanwhile, bring their own crucial interaction. They already run some land destruction main, even if it’s just Ghost Quarter, Field of Ruin, and/or Spreading Seas. Nowadays, the same is true for Black-Green Rock and Assassin’s Trophy is making its presence felt.
I can see how Tron might eke out an even record versus Ironworks, although my experience had me expect differently. The only thing that leaves me truly baffled is that Tron won almost half of its matches versus Infect.
Watch the World Burn
Burn’s results, in contrast, are all over the place, and I can’t make sense of a lot of this:
For instance, against Jund’s closest relatives, Abzan Midrange and B/G Rock, Burn posted winning records, at 2-1 and 3-2 respectively. Versus White-Blue Control, which has a less painful mana base than Jeskai but will never target Lightning Helix with Snapcaster Mage, Burn went 3-4.
The above numbers don’t explain how the deck managed to put less than 14% of its pilots from our sample into the second day. So I feel the need to point out that overall Burn went 166-194 against the field.
Harder, Faster, Scalier
We’re only down to our fourth most popular deck and already the total numbers are dwindling. Four to eleven matches aren’t proper evidence.
This time, not even calling close relatives of listed decks to the witness stand corroborates or calls into question much of anything. Hardened Scales went fifty-fifty against Abzan Midrange and B/G Rock, as well as 2-3 versus Jeskai Control.
Nonetheless, the absence of too many outliers offers insight in its own right. That is, Hardened Scales appears to be an all-rounder, with no more than a few lopsided matchups on either side of the divide. The good news: the most important matchup, Spirits, went favorably with the most convincing sample size.
Overall, Hardened Scales won a solid 54.3% of the 381 of its matches on which I have data.
The League of Extraordinary Phoenixes
The Izzet League sent a new contender into the ring at Grand Prix Portland. Here are the notable matchups for all blue-red builds featuring Arclight Phoenix:
There’s no doubt that Izzet Phoenix’s win rates against Hollow One and Infect would become less pronounced in a larger sample, just as I’m sure the deck would beat Dredge at some point. With the exception of these and the aforementioned strong Spirits matchups, we get little notable data.
My suspicion: Izzet Phoenix doesn’t benefit from many extreme matchups, and suffers from fewer. Its 57.1% winning record in 401 recorded matches is in fact a record, and comes with a two-pronged implication for the future or, in the case of GP Oakland, the recent past. Izzet Phoenix’s metagame share should rise, and its win rate should go down. Needless to say, checking up on these predictions is high up on my to-do list as soon as I get a look at Oakland’s data.
Walk Through the Valley of the Shadow of Death
And we’re back on track with sensible, informative results! For some time a while ago, mana expenditure in Modern was locked in a downward spiral. Decks were looking to go ever cheaper, and Death’s Shadow was both the epitome of this development as well as that environment’s logical champion. From there, the only way onward was up, and Death’s Shadow didn’t match up too well against strategies aiming for a bigger and more solid endgame rather than speed susceptible to pinpoint disruption.
While the deck finished in first place at GP Portland, its overall performance lacked in luster. It only won 49.3% of the matches about which I know, and this includes Tyler Putnam’s 3-0 run through the Top 8.
In addition to the things mentioned above, two things stand out. One, in a world where Dredge has received new toys, relying on a pair of Surgical Extraction for graveyard removal—as Putnam did—maybe doesn’t cut it anymore. Two, if Ponza can’t even beat a deck that relies on seven lands to produce all of its mana, then that’s really something. Probably an indictment.
Most of KCI’s notable stats I already talked about when they showed up earlier. I repeat the full list here for your viewing pleasure and to illustrate a larger point:
Ironworks exhibits one of the more lopsided matchup spreads we’ve seen so far. When the deck is good, it’s really good. Though everything could change at moment’s notice if people turned the knobs of their sideboards, the thumbscrews, accordingly.
If you’re looking for an explanation of why Ironworks proved so much more successful against Hollow Phoenix than it was against traditional Hollow One: One (pun acknowledged) harbors an Ancient Grudge and Leyline of the Void, while the other typically doesn’t (or didn’t as of Portland anyway).
Black Blood Braid
For something completely different, more even across the board, we should be able to turn to Jund, Modern’s original midrange and middle-of-the-road gangster …
Well, I can’t explain either why the deck would do well against Burn and Amulet Titan, or why it lost every match against Humans. I basically suspect all of Jund’s matchups to lie closer to 50% than this, and am willing to attribute the outliers to the low match count.
I did my best to make sense of the results so far, or you could say I did my worst. “Make” is indeed the correct word. Depending on your fault tolerance, “too little data” would be an equally valid or superior interpretation for a number of matchups discussed above. I’m obviously a big believer in comparing actual outcomes with expected outcomes, or of looking for reasons why matches went the way they did. Your mileage may absolutely vary.
If you think any of these results differ massively from how a matchup should play out, I can only encourage you to leave a comment below! Contrary to popular wisdom, I don’t think it’s hard to argue with results. I believe it is essential. As I never tire of repeating: Reality is but a woefully imperfect approximation of mathematics. To reverse engineer definite truth from observation, we’d have to observe way more matches. The best we can settle for is probable truth.
The Best of the Rest
Either way, we have reached the point where I simply dump all of the remaining matchup tables. Enjoy!
Next in line, the thirteenth most common archetype in my sample was Valakut, the Molten Pinnacle, meaning TitanShift/Breach. But under the same criteria as before (decks paired against each other ten times or one deck defeating the other with 70+ percent in at least four encounters total) there’s nothing more to add here. Valakut didn’t meet/beat/lose to any further decks often enough.
In fact, among the less popular strategies only four more pairings in all of the field rose to that level. Black-Green Rock beat Amulet Titan 3-1 with Assassin’s Trophy, Hollow Phoenix beat Amulet Titan 3-1 with luck, while Devoted Vizier went 5-0 versus non-Hollow Red Phoenix and 1-3 versus Storm.