The Speed of the Metagame - Tracking the Fall of BG Via Metagame Prediction Survey

The Magic Online Championships is in the books, and so is the Prediction Game. How well did the community predict the metagame at the Magic Online Championships when $50 in Cardhoarder bot credit was on the line? Um… not so well.

You can check out an in-depth breakdown of the submissions in my previous blog post, but in short, the community overwhelming picked the BG decks to both be most popular and to win the tournament. Mardu Vehicles came in second, and Saheeli a distant third. And at the actual tournament? The full decklists are available on the mothership, but here’s the archetype breakdown:

Ouch. BG was actually a distant third, at 2 decks split between 2 archetypes, coming behind a solid showing from 4-color Saheeli and an even more solid showing from Mardu Vehicles, which comprised fully half of the field. Mardu was also the choice of all three of the players tied for best standard record: Lee Shi Tian, Marcio Carvalho, and Piotr Glogowski, all at 6-2.

All in all a pretty poor showing from our community prediction survey. It would be one thing if the numbers were close and BG was just barely edged out by Mardu. But Mardu was fully half the field, and even second-most-common Saheeli had double the metagame share of BG. So what happened? Well, let’s look at the breakdown of predicted metagame share of each of the top decks by date of survey submission. Note that this is a 100% stacked column chart, so even though the vast majority of submissions came in over the weekend I’ve normalized the subsequent days to show how the proportions compare over time.

As I mentioned in my previous post, it does appear as if the community was just a step behind the metagame. Most of the submissions came in over the weekend, when GP Utrecht was still in progress and the results from the GP Pittsburgh, at which BG did dominate, were fresh on people’s minds. However, once the results from Utrecht became public on Monday, there was an immediate metagame shift: the “Mardu Ballista” or “Big Mardu” technology demonstrated that Mardu had fresh legs. BG previously was popular because it feasted on a favorable matchup against the “best deck” coming out of the Pro Tour, Mardu. But if Mardu Ballista tech could reverse that matchup - and it appeared from the Utrecht results that it did - then that would leave BG without a place in the metagame, and room for Saheeli to rise to number 2.

As the community digested the results from GP Utrecht, there were quite a few articles publicly available that spelled out the metagame implications. For example, at Channel Fireball, Josh Silvestri wrote an article headlined “B/G is Out, Saheeli is In.” A slightly less high-profile example is my aforementioned blog post, in which I expressed my belief that the survey was a few days behind the metagame’s shift away from B/G and towards bigger Mardu.

The submissions over time graph seems to confirm that it took time for people to recognize the shift. Granted, the sample sizes on some of the later days of the survey are quite small, as we only got a trickle of submissions later in the week. Even so, for what it’s worth we do see a trend towards greater predicted metagame share from Mardu, and decreasing predicted metagame share from BG.

Honestly, dynamics like this hammer home just how quickly the competitive metagame can shift. If you’re still jamming the same BG deck you built after the Pro Tour and GP Pittsburg because it seemed strong then, at this point I think we can safely say you’re a step behind the field. This constant churn even intra-rotation is what makes Standard so interesting from a pure gameplay point of view - but we must admit that this grind can become exhausting or cost-prohibitive for more casual players.

Since this is the MTGO Championships though, I think it’s worth pointing out that the ability to quickly switch decks is one reason why, for all its flaws, MTGO is nevertheless the most cost- and time-efficient outlet for a competitive-oriented player hoping to hone their skills. Swapping cards in real life is process that carries high transaction costs. Either you sell to your local store at half price (to be fair, they need these margins in order to pay the rent), or you binder-grind constantly to recoup as much value as you can from trade partners. In contrast, the online economy makes transactions far less painful as the vendors (such as our contest sponsor, the all-around cool dudes at Cardhoarder) can be far more efficient in their operations. If you wanted to audible from one tier 1 deck to another the day before the PTQ, you can totally do this online and lose just 10% of your deck’s equity just by conveniently selling and buying cards from a vendor. That's setting aside the fact that decks are typically a flat 25% cheaper online to begin with.

I hope everyone enjoyed taking part in the Magic Online Prediction Game, and that the results were instructive. The winners of the contest have been contacted by email.

If you're interested in staying in touch and participating in future free games like this one, leave your email address. I'll only use email addresses collected this way to notify you when I create a Magic Pro Tour Prediction Game.

Community Predictions for the Magic Online Championships

Earlier this week, I invited the paper and online MtG communities to fill out a metagame prediction survey for the Magic Online Championship being held this weekend, with $50 in Cardhoarder/Isle of Cards credit on the line for the most accurate submissions. With over 200 submissions collected so far, let’s take a look at how the community predicts this event to go down.

(Update: some additional submissions came in after the post, but did not significantly affect the numbers below. You can scroll all the way down for a link to a live dashboard with the complete data set.)

Picking Winners

The MOC has an elite 16-player invitation list: the defending champion, the Magic World Champion, 2 players from each MOCS Quarterly Playoff, and 8 players from the MOCS points leaderboard. Note however that in 2016, Gold and Platinum level pros received significant edges in the form of invites to Monthly and Quarterly MOC Events. This system means that the field is a mix of 11 established pros and 5 comparatively unknown online players. The “tiebreaker” questions on the survey asked for predictions on the makeup of the Top 4 and the eventual winner.

Our survey respondents clearly considered strong pros to be favorites to win the tournament, with the 5 non-pro players together combining for less than 1/3rd of the responses compared to incumbent Player of the Year Owen Turtenwald. Brian Braun-Duin also gets a lot of love from the community. This makes a lot of sense as the structure of the MOC, with its small yet elite field, actually closely resembles the Worlds tournament which BBD took down just a few months ago. There was a little bit of drama in the Pro community both about the “legitimacy” of BBD’s invite to Worlds - he earned the “grinder’s invite” as Grand Prix Master - and whether or not BBD was truly the “best” player of the tournament or the beneficiary of some variance, but it appears that in the community’s eyes, at least, BBD is a legit powerhouse.

The position of Marcio Carvalho in the middle of the pack merits additional discussion. Carvalho has been on an absolute tear lately, being the runner-up at Worlds and the current leader in the 2017 Player of the Year race. Judging purely from his results we would expect Carvalho to be a more popular pick. Unfortunately, we need to acknowledge that Carvalho has long been under a cloud of suspicion for sketchy play. Although Marcio’s only serious past sanctions have been a 6-month ban a decade ago and a recent Worlds DQ, there are a number of Pros that continue to “call out” Marcio to this day and publicly accuse him of continuing to cheat.

Of course even his detractors will admit that Carvalho is very good at magic, so why the low ranking? Perhaps the community buys the unproven allegations and is discounting Carvalho’s recent performances, particularly as the Magic Online engine will completely negate any possible cheating. Or perhaps people just don’t feel right picking a suspected cheater to win it all. The sad truth is that Carvalho is either an honest player with some past indiscretions that is unfairly being maligned, or an unrepentant and continuing cheater that is actively stealing from his fellow competitors, and neither scenario is ideal.

Predicting the Standard Metagame

One of the more interesting questions to what extent is the current Standard metagame a “3-deck-meta” of Mardu, Black-Green, and Saheeli Combo as the best Aggro, Midrange, and Combo decks in the format. Here’s how many rogue, ie not-Mardu/BG/Saheeli decks the survey expected to show up:

With a median of 4 or 25% rogue decks, there were slightly less “Standard-Optimists” expecting greater than 4 rogue decks, than there were “Standard-Pessimists” expecting fewer than 4 rogue decks.

As for the decks themselves, people were very taken with Black-Green, with Mardu and Saheeli Combo the picks for 2nd and 3rd most popular:

Not only was BG Midrange the consensus pick to be the “most popular” deck in the tournament, it was also the majority choice for the “most successful” deck of the tournament. These are not at all synonymous, and often the “best” deck of the tournament is the one that “next-levels” and preys on the “most popular” deck. But here, many people thought that - for this weekend at least - the “best” and “most popular” would be one and the same.

That said, I think the survey respondents are at least a day or two behind the latest meta. The survey opened (and the bulk of the submissions came in) over the weekend while GP Ultrecht was still in progress, so they were largely informed by the results of the previous major event, GP Pittsburg. There, the Black-Green decks put up strong performances thanks to a favorable Mardu Vehicles matchup; the GB decks were just a little bigger than Mardu and so could keep up with the fast aggressive vehicles decks and then eventually turn the corner with more powerful spells. But the latest tech out of GP Ultrecht was “Mardu Ballista,” a vehicles variant that itself went slightly bigger still, including an anti-Black-Green sideboard plan of planeswalkers and Fumigate. The latest consensus, from what I can gather, is that this new tech even flips the matchup in Mardu’s favor.

At this point it’s worth reminding anyone who’s filled out the survey that you can go back and edit your submission as long as you log into your google account again. It’s very possible that these responses will change dramatically as people come to appreciate the implications of GP Ultrecht’s results.

We also see some interesting things when we narrow the charts to display only those responses from “Standard Optimists” and “Standard Pessimists,” as defined by their outlook on “number of rogue decks” as mentioned in our earlier chart. Here’s what the optimists think:

The makeup of the “most popular” decks doesn’t actually change all that much - the optimists are actually even more confident that Black-Green will be the most popular deck - but the standard optimists are more hopeful that a non-meta deck such as Dynavolt Tower, Emerge, or some other archetype will be able to attack the format and win.

What about how the pessimists think?

At least from what I see here, I must admit the pessimists appear savvier than the optimists in their metagame predictions. Mardu - which as we cover above is the latest tech - is still second but pessimists almost twice as likely to have picked it to be most popular, and slightly more likely to have picked it to be the winningest.

As for the moderates, no one cares what the moderates picked. Their picks were dull in how closely they matched the survey as a whole. Pick a side, moderates, you’re boring! (Ok, if you’re really interested in how the moderates picked you can scroll down and investigate the complete interactive dashboard for yourself.)

That’s what the community expects from the MOC at least. My personal prediction: the Mardu Vehicles’ “Ballista” tech is very real. I’ve actually seen it at work already in some quite high-level events, and the matchup against Black Green truly seems to be as good as advertised, while the matchup against slower combo and control decks remains quite good. I think it’s about to show up in a big way on one of Magic’s biggest stages. I'm less confident about which deck will actually be the most successful - it could be Mardu Ballista, but there are a lot of very talented players competing this weekend for a large prize pool. The rewards would be huge for any player who successfully identifies new tech. For the most winning Standard deck, I wouldn't be surprised to see some real innovations, either in the form of another novel re-tweaking of one of the big existing archetypes, or a rogue build.

Of course, we’ll need to watch the event itself to see what happens for real. Until then, fill out the survey if you haven’t already to be in the running for the prize, and check out the live dashboard if you want to do a deep dive into the survey data. I will update the dashboard with live data at least once or twice until the event begins and further entries are locked. Hope everyone has fun playing the MOC Metagame Prediction Game!