It's been... wow, almost three years since last I analyzed the results from a major Magic weekend. The entire environment of tournament Magic has changed in that time. Where I used to report on Star City Tours, Grands Prix, and the occasional Pro Tour, I'm now limited to Regional Championships (RCs). The pandemic has done a number on everything. That said, Regional Championships are currently ongoing, so let's dive in and see where Pioneer is headed.

Note that at the time of writing, the results I have to work with are the Atlanta, Sao Paulo, and Sofia Championships. Results for the weekend's Pacific region championships were not readily available. There's nothing about them on the official Wizards site, and Google gave me info sheets, but not tournament coverage. I know a few snippets from Twitter, but that's all. I'm going to skip over that partial information and focus on the more complete data.

Disclaimer #1: Avoid Reading Too Much

As long as I'm doing disclaimers, here's a big one: don't read into these results too much. This goes for the results I'm covering today as well as the rest of the Regional Championships. It isn't that I don't trust the results or that the events aren't useful. The problem is that the data isn't truly random, and therefore isn't fully valid.

Full validity requires a large sample size and data representative of the total population. The first condition is not a problem. The Atlanta Championship alone had 923 players. That's more than a lot of Grand Prix in 2019.

However, this isn't truly representative data. The problem is that the RCs aren't open events; they're invitationals. An open event is an opportunity for any player to be represented, and is therefore more representative of every player. These RCs are only representative of the spikiest of players who won their invites. Therefore, the data I'm working with is representative not of the overall population, but of tournament grinders specifically.

Disclaimer #2: There Was Metagaming

The other thing to keep in mind is that there were a lot of strongly held beliefs about Pioneer's metagame going into these events which definitely affected player choices. In addition to a crescendo of the usual "Pioneer Sucks" vs. "Pioneer Rocks" activity, there was a ton of chatter on decks being utterly unplayable vs. hopelessly broken.

While I cannot say how all this chatter affected player choices, it certainly did, because players were constantly tweeting about changing their minds as the perceived metagame shifted. I can't quantify this effect, but it does warrant mentioning.

Also, remember that all this data came from a single weekend. It represents what the players of several regions thought about Pioneer at the time and for the purpose of a single, large tournament. That isn't necessarily indicative of what is currently happening ahead of other RCs or of what happened this past weekend.

Comprehensive Data

Alright, enough disclaimers, and on to the actual data! I had intended to gather all the data myself, but as it turned out I didn't need to. Frank Karsten already did everything I was going to do and a whole lot more for Wizards. I assume that he's got actual programs helping him count up the match wins for his winrate calculations, because I can't fathom doing that otherwise. Well done sir and thank you. I'll go through the population data first, then the win rates.

The Big Green Elephant

First things first: Mono-Green Devotion was almost 21% of the field. That is both unsurprising and very surprising. Unsurprising because Devotion has been the assumed default best deck in Pioneer for some time now. Less so because according to the social media chatter, it should have been higher. I recall some claiming that Devotion should be 50% of the field and all the Top 8. While obviously hyperbolic, it does show how big a shadow Devotion casts over the format, both in reality and in players' minds.

What is more interesting is that Devotion's metagame share in these three RCs is significantly higher than its actual share. At the time of writing, Mono-Green Devotion is 16.3% of Pioneer according to MTGGoldfish and 12% on MTGTop8. Which looks suspiciously like all the predictions of Mono-Green's domination were accurate. Again, this is an invitational event for spikes, so it does make sense for many to be on the presumed best deck.

However, that may be deceptive. Many might have chosen Devotion for that reason, but it may also have been the deck that got them there and the only one they were comfortable with. This is an invitational that players had to win an invite to, after all, and if the deck was good enough to get there it's probably good enough to run. Trying to audible to an unknown deck is usually a bad idea, after all. Thus, Devotion's specific metagame share isn't all that meaningful.

An Indicative Metagame

However, it does fit into the overall picture of this RC metagame quite nicely. According to Frank's data, the Top 5 Pioneer decks at the RCs were Devotion, Rakdos Rock, Izzet Phoenix, Mono-White Humans, and UW Control. Both Goldfish and Top8 have these the exact same decks as the Top 5 too, though Top8 has Rakdos above Devotion for the top slot. Additionally, ignoring differences in specific metagame share, the overall metagame from each source looks similar. This strongly suggests that thanks to the high population of the three events, they were ultimately able to accurately model the metagame.

However, that is a bit deceptive. There were huge differences in results between each tournament. It is only thanks to aggregation that it ended up looking similar to the "real" metagames from the websites. For example, Devotion was 21.9% of the Atlanta RC but only 15.8% in Sao Paulo. I don't know what it did in Sofia, but it had to have been around 25% for the overall field to produce a total metagame share of 21%.

Thus there were enormous, but compensating, regional biases present in the data. If the Pacific RCs' data were available, it would throw off the metagame picture. I don't know how severe the shift would be, but it would be a very different look.

Metagame By Win Rate

What is more interesting is that Frank has provided the numbers on each deck's win rate. I don't do win rates because 1) the starting field data is usually unavailable, so I only know that Top 32 deck's win rates, which doesn't model anything about the overall field; and 2) many events only report standings, not records, making win rate calculations impossible. Frank had all the data from all the events (and a math PhD), so he could and did do them. And the results are unexpected.

Lotus Field Triumphant

The headline result is that Lotus Field combo had the best overall win rate of these RCs. This is surprising for a number of reasons. The first is that Field did so well despite not winning any event. In fact, it put exactly one deck into the Top 8 of Atlanta and nowhere else. A high overall win rate doesn't translate into an event win.

The second is that Field showed up at all. Back when the new qualification system was announced, Field was considered a major player in Pioneer. However, it has been declining steadily since then. I know that back in June there were at least three Field players in my local Pioneer metagame and now there are none. The deck is an awkward and slow combo to pull off and punishes mistakes. This leaves it as a fairly niche deck.

However, that second point was critical to Field's success at the RCs. Field's combo is easily hated out, but players didn't expect it. Damping Sphere hits all aspects of the combo and can be played in any deck, but is also the most easily answered. Tomik, Distinguished Advokist and Archon of Emeria are more effective, but didn't see much if any play. As the only prominent pure combo deck in Pioneer, Field thrives mostly when players are unprepared. Now that players are aware that Field is still a deck, they'll have the hate in the future, and Field should decline.

Leading from the Bottom

Looking at the standings overall, what might be surprising to many is that none of the Top 5 decks from the population have exceptional win rates. Phoenix did the best at 53.3% and is in 7th place. The rest are in the middle of the pack with rates around 50%. Which isn't spectacular.

In fact, all the best decks by win rate are mid-Tier 2 at best. Field was only 1.8% of the field while runner-up Selesnya Auras was a measly 0.5%. This shouldn't be surprising to regular readers of my Modern metagame articles. The top of the average power charts is always filled with Tier 3 decks, frequently low-placing Tier 3 decks to boot. Low-population decks tend to be run by enthusiasts and so their win rates are less likely to be dragged down by inexperienced players misplaying the deck, which is what kills the performance metrics of Tier 1 decks. Thus, an unexpected low-tier deck like Lotus Field Combo should do much better than the expected decks.

What Does It Mean?

Consequently, the win rate data doesn't actually measure what deck is best in Pioneer, just which decks performed best at the RCs. This is what I was alluding to above in the population data. Players were prepared for Mono-Green, and there are tons of anti-Devotion cards in players' sideboards. This fact, plus the high population, dragged down Devotion's win rate. The fact that Devotion did so well in a very hostile field is impressive.

Ultimately, there is a reason that top-tier decks are top-tier. They have the power to fight through hate and players being otherwise prepared. Mono-Green did exactly that. Lower-tier decks face greater struggles, but can do well against an unprepared field. This is how Lotus Field performed so well yet didn't convert an RC. Win rates should never be read as hard indictors of which deck is best. What they are is indicative of what the field expected and prepared for.

On that note, it's also important to look beyond the overall win percentage to the specific matchups. Frank didn't include them in his article, but did tweet them out. This is where the disparity between overall metagame place and specific event win rates starts to make sense. Devotion's worst matchups were all from relatively low-population decks, while its matchups against the more popular ones were average at worst. Devotion was such a large proportion of the field that it couldn't avoid terrible matchups, and so its win rate tanked despite good placings overall.

Is Pioneer Healthy?

The big question after any large event is what it says about the format. Given all the buzz about Devotion prior to the start of these RCs, there's a decidedly mixed message. Devotion was a huge percentage of the metagame, but it didn't take up too many slots at the high tables (if the Atlanta data is indicative of all the events). There is a wide range of viable decks and the cost is low relative to Modern. Whether the gameplay is good is a matter of opinion, but the same is true of any format. Thus, overall, the data indicates that Pioneer is healthy.

However, there is a problem that the data can't reveal. While I've never noticed it myself, there has been a chorus of complaints about Pioneer matchups being determined more by who wins the play or draw than by matchups. Even Pioneer's defenders are acknowledging the problem, which is present in all formats but seems particularly pronounced in Pioneer. I don't have much of an opinion on this myself, as I've never personally noticed it and nobody's provided data one way or another. However, if this known problem is uniquely bad in Pioneer, then that indicated that there is a deeper problem with the format that the metagame can't fix. I'll keep an eye on this issue.

Proceeding as Expected

Ultimately, I don't think the data coming from the RCs will change anyone's mind about Pioneer, nor will it greatly shake up the metagame. However, it is comforting to know that Pioneer's metagame does appear able to keep even its best decks in check. Hopefully, the first set of RCs is indicative of how the rest will go, and Pioneer is as healthy as it looks.