964 players made Grand Prix Niagara Falls the largest Legacy tournament of the year so far. I was able to find out, with a lot of help from the awesome Legacy community, what decks 751 of the competitors decided to play. This equals 78% of the field and a huge amount of data to analyze. So here we go.
The Starting Field Composition
Next in line were nine players each on Merfolk and Delverblade, eight players each who dug into their graves with some Arclight Phoenix build or Dredge, as well as seven players each who ran Omni-Tell and R/U/G Delver.
As you can tell, when it came to archetype categorization, I erred on the side of caution. That is, I used a fine comb to split hairs rather than the somewhat broader brush for painting similar pictures in other formats. It makes sense too, because between Brainstorm and Ponder, even a few cards difference can make a whole lot of difference in Legacy game play.
Yet some animals are more equal than others. A typical Grixis Delver main deck differed from its more popular Izzet cousin by a handful of black spells at most: a couple of Dark Confidant or Gurmag Angler, maybe a singleton Bitterblossom, Fatal Push, or Kolaghan’s Command—rarely Cabal Therapy nowadays to go with the ubiquitous Young Pyromancer. The straight blue-red versions didn’t have many cards uniquely their own either: an abundance of Pteramander, more burn and bounce, and the odd Light Up the Stage. Of course, sideboards tell another story entirely.
A model is only as good as the utility it provides, and different interests may require different models. If all you want to know is how likely it was for someone at GP Niagara Falls to face an early Insectile Aberration followed by mostly blue and red spells, then the answer isn’t 8.9% but 12.6% or 13.6% (when counting R/U/G as well).
So the Human Insect was hardly an aberration at all. Comparably common was White-Blue Stoneblade, and this one crossed the 10% threshold too if you add in the White-Blue Delverblade decks. My friend Julian insists that the two belong in separate categories.
Admittedly, one additional reason for separating them is the simple fact that their numbers allowed the luxury to do so. In contrast, the above doesn’t differentiate between “Ad Nauseam Tendrils” and “The Perfect/Epic Storm” or “Fetchland Tendrils” or any of the finer notes the trained palate of a consummate combo connoisseur can distinguish. I just called everything “Storm”—except for the Mono-Red Storm deck powered by Ruby Medallion and played by two attendees to a Day 1 finish.
Comparison with Top Finishers
When we compare these numbers to the full Day 2 metagame, a few things stand out. Blue-Red Delver fell from almost 9% to about 5%, indicating a very poor performance on Day 1. Grixis Delver, on the other hand, climbed from 3.7% to about 9%.
Took most of the day, but we have a full Day 2 metagame breakdown for #MTGNiagaraFalls!
– Just one deck eclipsed 10% of the field.
– Stoneforge Mystic is having a very strong weekend.
– Delver strategies span the full spectrum, from Grixis to Temur to UW. pic.twitter.com/XAQ2kRXra4
— ChannelFireball (@ChannelFireball) April 21, 2019
Such a reversal is unbelievable. In fact, we literally shouldn’t believe it, because it also isn’t quite true. My data show that the metagame breakdown must have mislabeled at least some blue-red decks as Grixis. This is not a knock on the work the one-man coverage crew put in during the very limited time available on the weekend. Rather, I like to view it as proof of the value of in-depth post-event coverage such as what you’re reading right now.
Minor inaccuracies notwithstanding, Grixis Delver indeed outperformed Blue-Red Delver by a lot. Among the 751 whose deck choice I know, 11 of 28 Grixis players (39%) reached the second day, but only 15 of 67 Izzet players (22%) did.
Furthermore, the complete Day 2 breakdown suggests that the metagame share of Stoneblade and Turbo Depths increased after the cut, whereas Miracles, Sneak and Show, Death and Taxes, and especially Burn lost a bit of their stake. All of it is consistent with what the incomplete 751-player sample says about advancement.
Some of these trends continued on Sunday…
Stoneblade improved its share a little more at the top of the field, and Turbo Depths really put on the turbo for the homestretch. Grixis Delver again outperformed Blue-Red Delver.
At the same time, Sunday’s rounds introduced at least one new trend: When it came to placing inside the Top 32, both Grixis Control as well as Miracles excelled. Specifically, their success here exceeded their Day 1 performance, which echoes the truism that control strategies are at a fundamental advantage in a well-defined environment compared to a wider field.
Alas, when working off of incomplete Day 1 information, looking at conversion rates always leaves a bit of uncertainty. It’s much better to just calculate the win percentages for the known portion of the field.
Depending on how wide a net you cast, the top finishers variously suggest that Stoneblade claiming the trophy, Death and Taxes with two copies in the Top 8, or Miracles with five copies in the Top 32 were the weekend’s winners. But the best performance among mainstream decks delivered Grixis Delver. The Statistics 101 lecturer agrees. Winning 147 of 248 matches is highly significant. A deck with an even win rate wins this much or more only with a probability of p=0.002. (A p-value of 0.05 is the customary benchmark for statistical significance.)
It’s harder to make a definitive statement regarding the merits of Death’s Shadow or Maverick. Both won 57% of their matches within the sample. That much is undeniable. But we can’t just award a stamp of approval here, because it’s way too likely that these records resulted from variance. Neither deck’s performance was significant, at p=0.094 and p=0.097. In fact, most decks’ results didn’t clear that hurdle. The only decks whose records did pass muster were the aforementioned Grixis Delver as well as Sneak and Show, Stoneblade, Reanimator, and poor Burn.
Usually it proves futile to extend the quest for significant findings into the fringes of the metagame. It’s rare for a niche strategy to cobble together a meaningful record. In Niagara Falls, one minor deck did win in a major way though. I only know of four players who picked Steel Stompy as their weapon of choice, but they ended the tournament on 36, 33, 30, and 15 match points. Success on this level seems incredible, but a combined record of 34-15 is credible. At least the corresponding p-value of 0.0047 suggests something highly significant.
Or rather, it would be highly significant, if all of these wins were unrelated. With just four players contributing, we instead face a very strong connection. So maybe it wasn’t so much Steel Stompy itself but the Steel Stompy pilots that overperformed. I mean, this Jamie fellow who finished ninth isn’t just a regular Dekin but a proper Archdekin.
Then again, this whole field of study is predicated on the idea that well performing players must have chosen good decks and that it’s worthwhile to follow their lead. So maybe take a look at Jamie Archdekin’s 12-3 deck list and at Max Gilmore’s 11-4 deck list below.
Max Gilmore, 45th Place
4 Ancient Tomb 3 Blinkmoth Nexus 4 City of Traitors 1 Inventors' Fair 2 Karakas 4 Mishra's Factory 1 Tower of the Magistrate 4 Wasteland 4 Arcbound Ravager 1 Hangarback Walker 1 Phyrexian Metamorph 3 Phyrexian Revoker 4 Steel Overseer 4 Thought-Knot Seer 2 Vault Skirge 4 Walking Ballista 4 Chalice of the Void 4 Lotus Petal 2 Mox Opal 4 Thorn of Amethyst Sideboard 3 Karn, Scion of Urza 1 Hangarback Walker 1 Dismember 3 Ratchet Bomb 4 Leyline of the Void 1 Warping Wail 2 Ensnaring Bridge
I also checked the records from the first eight rounds and the following seven rounds for differences. Some notable improvements:
- Death and Taxes won 47.5% of 265 matches before the cut, then 60% of 50 matches.
- Sneak and Show won 53.8% of 292 matches before the cut, then 61.5% of 78 matches.
- Grixis Control won 48.9% of 186 matches before the cut, then 54.4% of 57 matches.
- Miracles won 51.3% of 304 matches before the cut, then 56.2% of 73 matches.
Turbo Depths, despite its strong presence in the Top 32, actually did worse on Day 2. During the first eight rounds the deck earned victories in 55.7% of 230 matches, whereas it barely went positive throughout the following seven, with a record of 43-42. These wins and losses clearly weren’t distributed evenly.
Next week, I’m going to use the data from Niagara Falls to answer further questions. Like: Who won in the battle between Reanimator and Turbo Depths? Or: what was Sneak and Show’s worst matchup?
When I did similar work for other formats in the past, I could always draw upon a wealth of expectations with which to compare the findings. For example, I was able to confirm again and again that Modern Hardened Scales has a problem with Izzet Phoenix. Conversely, I learned that Mono-Blue Tempo was in fact not favored at all against Esper Control at the Standard Mythic Championship. Since I rarely get to cover Legacy, I don’t know what conforms to conventional wisdom, what would constitute a surprise, and what’s a contested topic of debate in this format.
If you want to help me out, post a comment below and tell me what you consider to be some clear matchups. Which Legacy deck beats which other deck reliably? What do you expect the data from Niagara Falls to show? Then check back next week and cash in your hard-earned “told you so” equity.