I think it’s mostly a function of who plays Bloom and who plays Twin. Many of the Bloom players on MTGO are incredibly experienced with their decks: I’ve seen them playing those decks for 2-3 years now. This is reflected in the data itself, where many of the Bloom results were on the backs of a relatively few number of pilots. Not so with Twin, a deck that attracts both veteran players and also newer/less experienced ones who are just trying to play the “best” tier 1 deck in Modern. But those players are going to struggle against the Bloom pilots who are more experienced with the deck. Also, they will struggle because Bloom is a harder deck to play against if you don’t know what to disrupt. We only needed to see PT FRF coverage for evidence of that, and that was at the highest level of the format. All of those factors are probably at play in that particular matchup.

5 more dailies have been added to the dataset since this article, so you can expect updated numbers either this Wednesday or next Monday! It might even be up to 6-7 by then.

Email Sheridan.

In line with the other comments, and as a Jund player/BGx player in general, we can get away with Bob in Jund for a generally lower curve, a slightly less painful mana base, more lifegain cards up the curve, and somewhere in the range of only 3-5 cards with a CMC of four or more.

Most Jund decks are running 2-3 Scavening Ooze, 2-4 Kitchen Finks, 0-2 Huntmaster of the Fells, and 0-1 Batterskull. Also TS has become much more of a 0-2 of in the main deck of most BGx builds lately causing even less main deck life loss. There are definitely still games where you flip Tasigur or Batterskull for a loss, but even in a long game those odds are pretty slim…but I’m not going to lie, I’ve found myself bolting/terminating/abruptly decaying my own Dark Confidant on occasion just to avoid the chance in a few tight games 😀

Thanks man and again great work!

Greetings,
Kathal

As I mention in the article, I agree with you on many of these points. It’s a matter of checking our quantitative results against experiences. For instance, with just 10 matches total and a win rate of 6/10, we can’t safely say that Abzan’s MWP against Affinity is just 60%. It’s probably anywhere from 40%-60%, all things considered. Think of it this way: If one Affinity pilot made a single misplay in one of their games (e.g. bad combat math, which isn’t that hard to screw up), then that immediately switched the matchup from 5/10 (50% – “even”) to 6/10 (60% – “favorable”). So the only thing we would want to definitely conclude for Abzan vs. Affinity is that the true MWP is somewhere around 50% and maybe trending positive.

For Grixis Delver, I mention that Abzan’s true win rate is almost definitely not 90% and is probably closer to 60%-70%. But that’s still a very favorable matchup for the Abzan player, all things considered: most matchups are about 45%-55% in Modern, so getting up to 60%+ is huge. Again, the takeaway would not be that the deck has a 90% win rate, but rather that it has a “very favorable” matchup here that is probably at least 60%.

We can use a similar framework for analyzing the different Burn matchups with Abzan and Jund. In reality, Jund’s Burn matchup is probably around 60%/40% (Again, trending the observed MWP back to 50%). But that’s still pretty solid for a BGx deck, especially if it could be as high as 65%/35% or better. This one makes more sense than the Abzan vs. Affinity MWP because we can identify more qualitative reasons to explain it (e.g. Bolt, manabase, etc.), because the N is larger, and because we have seen a rise in Jund online and this could be a big factor.

So overall, I think you are right to question quantitative results like these based on your experience. The key is not to think of them as concrete benchmarks, which is something I talk about. Instead, it’s to use them as checks on your own experiences and then to kind of build an MWP interval out from 50/50 based on your experiences and the size of N.

Re: Delver/Bloom
I think this is true of Amulet Bloom but not necessarily true of Grixis Delver. That deck really is everywhere, and I wouldn’t be surprised if the true prevalence were anywhere between 7%-9% if you had a full picture of the Daily metagame. But for Amulet, your theory makes a lot of sense. We already know this deck doesn’t have that many unique pilots (at least, not relative to other decks), and that it has a crazy MWP. This suggests it is indeed overperforming based on a few players’ successes.

Re: Junk/Jund
Because the Ns are a bit smaller for Jund, the only swingy matchup we can reliably comment on is that Burn matchup. This actually makes a lot of sense in my experience, because Bolt is so much better than Path here, and because the manabase gives you 2-3 free life over the course of a game. This is the sort of difference that really adds up over the course of a match.

As for the metagame sheet, I’m working on getting something together to make the numbers more viewable for our readers. No ETA yet, but it’s a work in progress! I have some other Top Deck page/metagame sheet updates coming tomorrow too.

of course i recognize that i might be simply wrong, i do not claim to be better than the testers in fact i might as well be a worse magic player/deckbuilder

Abzan 60% vs affinity due to stony silence for instance? game 1 is most of the times lost and Stony Silence is 3? sideboard slots, i will mulligan medicore hands without it, but aggressive mulligan is not an option unless we run 4 (never seen anyone doing this so far), also Abzan’s 90% vs Grixis Delver seems too much : UR delver is indeed a walk in the park but now they have big creatures of their own to rival Goyf/Rhino and relevant black removal, it’s still favourable but in my experience (which if of course not that much of a sample like 60-40 in Abzan’s favour, lastly 50% vs Twin seems to low, i don’t feel that this deck can win without Blood Moon, their combo is vurnerable and i am certainly not afraid of Keranos which mostly provides Bolts vs a Boltproof deck

lastly Jund’s match-up with burn seems abnormal, if that’s true i am switching to Jund on the spot, but burn has always been a bad match-up for Jund and now it’s stronger than ever, in my meta i am facing lists running 4 Atarka’s Command and 2-3 Skullcracks and life gain has never been more unreliable against them and will generally not work unless paired with disruption aiming for these effects

Abzan Liege is coming up next week! It has some very decent matchups across the board and is one of MTGO’s more viable decks.

Glad to hear that you enjoy the content! Let us know if there’s anything you want to see more/less of, or any general changes.

It definitely comes up, and until the most recent dataset update, was actually the deck with the most statistically significant MWP of all decks with a reasonable N. It’s gone down a little since then, but it still remains a very viable deck on MTGO along with Abzan Liege, Infect, RUG Twin, and Esper Mentor. Expect to see more on these decks in the next article in the series!

I would agree with Daniel’s assessment. The big reasons you can’t run Bob in Abzan are the 4 Rhino on top of the 1-2 Tas, and the manabase. With respect to the creatures, that’s just too many cards that can randomly kill you. With respect to the lands, It’s not very likely to resolve a turn 2 Bob with more than 16-17 life as Abzan, just because you have to run shocks and fetches to enable that turn 1 IoK, Path, or Hierarch. With Jund, it’s totally plausible to resolve a Bob on just 18-19 life, which makes a big difference against decks that take an aggressive posture. Jund has the added benefit here of being able to open turn 1 Bolt into turn to Bob against an opponent who played an aggressive one drop. Abzan has to Path that turn 1 play, which just guarantees a turn 2 play that is ahead of the curve. These are just some of the reasons Bob is a much more viable card in Jund than in Abzan.

a very interesting article about the match-ups from various decks based on pure data (although for some match-ups the N is quite low, but I still think, you can take those values as guidelines for those match-ups).

There are 2 things which surprised me:

1) The prevalence of both Grixis Delver and Amulet Bloom. If I’m correct, both decks have only a “few” players (relative low percentage of the deep dive metagame) compared to the top end finishes (which is the first prevalence of the data). This means for me, that those 2 decks have more 4-0/3-1 finishes than other decks with a lower player base (so the player base from Grixis Delver and Amulet Bloom 4-0/3-1 more dailies than the player bases from other decks). The exact opposite of this is Burn (10,4% to 9%) and Affinity (7,7% to 6,9%).

2) The variations of the match-ups between Jund and Junk. I never thought that some match-ups can swing that much around.

In the end, I have a small question. Would it be possible to create a metagame sheet form the “deep dive” metagame, since I think there would be several “huge” shifts compared to the high end finishes metagame spreadsheet (would be especially interesting because of Sideboard cards on MODO).

Greetings,
Kathal

I think with Jund’s total average curve being lower that playing bob with 2 tasigur is ok. Abzan curve average is brought up by the full playset of rhino’s. I think the general consensus is that jund is better against he field, where junk is better in the BGx mirror, but having access to bob definitely improves the mirror match. He’s bad in the burn matchup, but can still trade with a goblin guide. Sure lingering souls is a great card in grindy matchups, but against a bolt-less opponent (junk) bob still puts in great work, at the very worst giving your opponent 1 less path/decay to point at your goyfs/scoozes.