Over the course of the year thus far, the Lands community has gathered data on 944 matches played. Below we crunch the numbers on that data.

First a couple disclaimers –

1. No fudging – all losses are losses, all wins are wins. While this may seem obvious, anyone who has tracked their data knows the temptation to put down a loss from a stupid misplay as a win. But stupid misplay losses are still losses, and should be counted as such. Same goes for when your opponent just ‘got super lucky.’
2. Skill Level – relatedly, this data does not take into account what might be called the ‘skill level’ of the players, whether the opponents or the Lands pilot. It includes data from FNMs where your buddy is playing a meme deck, and it includes data from playing against pros on MTGO. As to the skill level of the pilots, I can’t speak to that to preserve anonymity. I can say however that about 2/3s of the data came from my own matches, and I picked up the deck in February and started tracking about a week or two after getting my Tabernacle. That said, there’s a certain self-selection bias here – people who track their data are probably taking their play more seriously on some level, and those willing to submit it are probably happy with their performance.
3. This data spans the whole year (well, at least from February), so it includes information from the Breach & Companion meta. I consider that a feature, not a bug, as it is interesting to see how Lands did against those decks.

The overall idea is that all the differences in luck or skill come out in the wash when one has enough data.

With all that said, let’s get down to it! First, the question on everyone’s lips – is Lands ban-worthy??

The average winrate over these 944 matches was 56.51%.

So by WotC’s criteria, Lands is a broken deck and you should probably be playing it.

Let’s take a look at how that winrate breaks down relative to opposing archetypes. For a chart of how decks were categorized, look here. Some decisions are perhaps questionable (is Urza Echo really a Chalice deck?) and others make sense only from a Lands-player perspective (a whole category for Knight of the Reliquary decks), but at the end of the day some categorization had to be made, no system was perfect, and we did the best we could.

In the graph below, we see the winrates against different major archetypes. Archetypes are ordered by meta share, which can be found on each archetype’s label.

Because the data is ordered by meta share, we can see that Lands is favored against the top 5 decks in the meta, or roughly 60% of the meta (though of course the category of ‘Other’ varies a lot).

On the level of individual archetypes, this data more or less confirms what most Lands players already know. We are favored in fair matchups (see tribal, delver, and control) while less favored against combo (Storm, Show & Tell). The delver matchup is less favored than many might like, just about 55%. Control is more favored than some might have thought, though straight UW control is actually worse for us. The worst form of control to be paired against is combo-control (think Food Chain and Aluren) as those decks don’t need to compete with us for inevitability, and instead can simply threaten a win at any moment.

With regard to combo, the Storm and Show & Tell matchups seem less bad than many Lands pilots feared, with Storm even coming out as about 50/50. The graveyard-based combo decks (Reanimator, Dredge, Hogaak) come out as generally positive matchups. Even Doomsday turns out to be about a 50/50 matchup.

Our best matchups by far are against Big Mana decks like Post and tribal decks like Elves and Goblins. This makes sense since tribal decks rely on creatures and are thus extra vulnerable to Punishing Fire and Tabernacle. Big Mana decks are extra soft to wasteland. Both decks tend to have few answers to Marit Lage. That all combines to make for great matchups on the Lands side.

Now let’s see which version of Lands seems to perform the best.

Before discussing these results, it’s worth noting that Jund was by far the most popular variant, with almost 450 matches to its name. UG, RG, and BUG each had about 100-120. The other variants were all around 40-50.

Looking at the results, it seems like while Jund may have been very popular, BUG is secretly the best-performing archetype. One could try to explain this away by saying that Jund’s lower average is just a function of its popularity, with Jund coming down to the true average where BUG’s 121 matches didn’t normalize as much. Still, BUG seems like a good avenue to explore in the future, as does UG. Both of these overperformed, and had a decent number of matches to make their performance meaningful. BG is another point we could consider looking at – it had a 60% winrate. This was over only 43 matches, but it is still promising.

The remainder of this article looks more closely at certain opposing archetypes to see exactly what kind of Delver (for example) is the most difficult.

From the Delver breakdown we see that Death’s Shadow is a quite good matchup. This makes sense, since they run few basics and even just giving them life with Grove can impact their gameplan. Conversely, BUG, Grixis, and UR are less good. Luckily, BUG and Grixis are much more rare (and their numbers here are not high, so the data might be misleading). UR however, has the advantage because of its basics and its ability to run Blood Moon.

Let’s look at Storm next:

In the eternal battle between ANT and TES, it seems ANT has the edge in the Lands matchup. This is probably because of their stronger ability to play a long game; post-board games against Storm tend to drag on under a sphere or two. Indeed the fastest, most all-in storm deck (Belcher) is actually a favored matchup for us. This is probably because they can lose to Tabernacle when they go for goblins and have so little interaction overall. A turn 1 sphere on the play is often enough to win the game on the spot against them, for example, as is mulliganing to Mindbreak Trap on the draw. The data here also confirms what everyone already knew – Breach was a broken deck.

Next is control:

Here we find the continuation of a theme. Decks with stable mana and answers to Marit Lage are harder to beat than the decks without. UWx control decks, for example, are tough, while BUG and Stryfo Pile are relatively easier to beat. Snowko straddles the center, since some versions play fewer Swords to Plowshares.

It’s also interesting that decks like Food Chain or Aluren are quite difficult for Lands. It make sense, since these decks (especially RiP Helm, which runs Swords), can attack from two angles, and their combos can be hard for Lands to interact with.

Depths is next – this is a matchup that has traditionally been described as favored, but our overall winrate was below 50%. Let’s see why.

Well here we can see the culprit – Slow Depths. For some reason Slow Depths (which includes traditional BG as well as BUG and GW Depths) is quite a difficult matchup for us. GW depths has Knight of the Reliquary, which has always been hard for us, and in general it can be hard to fight off the combo while also answering threats like Dark Confidant. I suspect that the abysmally low winrate there is at least in part due to small sample size (9 matches), but it does show a general pattern.

Turbo Depths, on the other hand, is strongly favored, as many probably already knew. Playing against Lands is a 50/50 matchup, which may surprise you until you realize that we’re playing Lands ourselves so…

The last category we’ll look at is the Tribal decks.

Perhaps nothing too surprising here. Humans and Slivers, the most straightforward beatdown decks with the fewest card advantage engines and the most fragile manabases, are the best matchups for us. Elves is the worst tribal deck to be paired against because of it has its own combo and can pressure us pretty well. Merfolk and Goblins both have their angles of attack that make them slightly more difficult than Humans or Slivers.

Overall though, no deck in this category is below a 70% winrate, so Lands players paired against tribal decks should feel good about the matchup.

And there you have it! Thanks to everyone who shared their data and made this collection possible. If you are interested in the raw data or the methods used to manipulate it, you can find the data and the computations themselves here. I encourage anyone to work on the data and create any other graphs they find interesting; it would be a pleasure to post them here or in their own article.

Thanks for reading! – aslidsiksoraksi