I’d have to check the spreadsheet but those first things are probably just data entry errors in the article itself. There was no easy way to copy and paste the data into the article so this kind of mistake can definitely happen.

As for the sample size, I made quite a few disclaimers on this point throughout the article. We obviously don’t have enough data for the tier 2 decks, which is a big reason why I didn’t report it. We do, however, have enough for numerous tier 1 matchups. And in instances where the N is a little low and/or the observed win rate doesn’t align with our expectations, I try to note that in the article itself. This crosschecking between the observed rate and our experiences is a strong way to shore up an otherwise smaller-N dataset, and it’s the method employed here in a number of cases.

For me, this is the MTGO player experience gap at its finest. Amulet players are very experienced with the deck: Amulet has one of the lowest ratios of unique players to appearances (and maybe THE lowest, if I remember correctly). Twin, on the other hand, has so many players with such a huge range of skills. If those experienced Amulet pilots are always at the helm of their deck, and Twin players have a much greater range of experience, then this is going to rapidly tend towards Amulet.

Re: Abzan vs. Twin
I think this is a player skill issue more than anything. Most people aren’t playing Abzan on MTGO, which suggests the remaining Abzan players are either really good with the deck (some of them) or really bad and have made a poor decision (probably more of them). It’s way easier to misplay against Twin, especially the UR versions with Moon and a more control-oriented gameplan, than it is for Twin to misplay against Abzan. I think this accounts for a big part of these results.

Re: Jund vs. Burn
This one actually seems more legitimate. At absolute worst, this should be about 50-50. Burn can race Jund, but Jund has a less painful manabase, Bolt, and potentially Huntmaster. Even the more conservative estimates I have seen put it somewhere between 50-50 and 60-40, the latter in Jund’s favor.

Hello dear kanister. So, you are stating that the Amulet players has got an opener with both Tolaria and a Titan-or-G-pact, or finds one of them in the process.
And if not?
If you just have a Titan and no Tolaria/Cavern? Especially if you run Cavern. I am playing 2 Cavern of souls just for this reason.

Abzan has a bad Twin match-up? why? is blood moon that much of deal? Abzan can pack tons of instant speed enchancement removal in the form of Abrupt Decay, Dromoka’s Command and Golgari Charm and it’s disruption paired with boltproof creatures seem more than enough to smash UR mages, I don’t know why the statistics point that

Jund has a good Burn match-up? I’ve seen this elsewhere too and i’ll make a risky assumption to explain this: Burn attracts new pilots more than any other deck due to a combination of effectiveness, simplicity and low price. Any of us who has a friend who is about to start modern would probably suggest him to start with burn and see how it goes.

on the other hand people with full BG and UR shells are most of the time seasoned players, why? because for someone close to the average (EU/US) income it takes years to collect all those cards, therefore he’s got to have learned something in all this time, also there much less young players in these archetypes as it’s simply hard to convince a parent to give you a 4 digit number of dollars/euros to buy cards

ofc good and bad/ old and new players can be found everywhere, I do not mean to underestimate anyone, but it’s natural that budget choices are much more likely to be piloted by inexperienced players, which will affect these statistics due to their high numbers, on the other hand Jund ‘newbies’ are an oddity (who would invest 1600+$ on a game he just started playing? and might not even like in the long run) and therefore unlikely to affect statistics in a significant manner

Hey,

thank you, great analysis! 🙂 regarding playr skill:

some months ago i looked up the match win % of around 20 famous pro players at 2 Modern Grand Prixs and despite entering the tournament against 3-0 players (byes) they still had a match win% of ~67% (quite similar in both grand prixs). Interestingly the % did not differ much between Day 1 and Day 2 (1-2% difference). I dont have the exact numbers at the moment. I guess that also indicates that most players who reach 3-0 at a Grandprix have chosen a solid deck and are solid players.

The Twin player is one under the pressure. When Amulet Bloom player manages to find his Cavern of Souls and gets to six mana, it trumps Twin. And it’s not like Twin’s permission is very hard or it’s damage clock is very fast. Just a bunch of remands and 3-mana 2/1s or 1/4s. Twin wins if it manages to combo off early or resolve a Blood Moon and Bloom player has no interaction with that – and remember that they could lways be holding a Slaughter Pact. In the end, the Twin has ways to easily kill the opponent, but the Bloom player has option to go both under it with a god hand or over it if he has drew a piece of interaction or two.

thanks for the reply!

What do you think makes the Bloom Titan – Twin match up so close? It really does seem like it SHOULD be a slam dunk for Twin. They have permission and blood moon to shut down Bloom Titan’s plan, and then Bloom Titan has very few tools to stop them from just comboing out. From a theoretical level, it’s really hard to imagine a way Bloom Titan wins, save for the occasional turn 2 kill and the usual lucky breaks that come with any match-up.

Esper Midrange is rocking a 58.9% MWP with 16 appearances and 56 matches. The MWP is not significantly different, with a P value of just .21 (but it’s getting there!).

Merfolk and Delver have seen 11 matches in the dataset so far. Currently, Merfolk is 8/11, which is definitely trending towards a strong matchup. The “true” win-rate is probably closer to 60/40, but Merfolk still seems like a favorite there. The only other Merfolk matchups with a large enough N to extrapolate from (and honestly, it’s still small), are Burn (9/19), and Twin (13/20).

Maybe I’m missing something you are responding to, but the MTGO average win rate is about 50%. If you are talking about specific decks, however, then the average deck win-rate (or matchup win-rate) would definitely not always be 50%.

This is a pretty solid Amulet list from SCG Worcester (2nd place):
http://sales.starcitygames.com//deckdatabase/displaydeck.php?DeckID=85210

That’s basically the list I’ve been using in testing, although I treat Oracle as a flex slot for something like Dragonlord. You can also use Simian Spirit Guide to accelerate into faster threats, but I like this less as people have become better at beating the deck.

As for boarding and other strategy, I suggest checking out the primer on MTGSalvation. Izzetmage is a very knowledgeable user there who has a great handle on Amulet
http://www.mtgsalvation.com/forums/the-game/modern/tier-2-modern/556715-amulet-bloom

We’ll probably do our own coverage of this deck at some point too!

I’m trying to do some analysis of player experience/skill. My guess is that it accounts for no more than 10 percentage points in an MWP calculation. So a true 50/50 matchup would actually be 40/60 in the hands of a skilled player. But this might cut both ways, so if the opponent was actually bad, it might widen the gap to 30/70.

Jund vs. Twin is unexpected for me too. The confidence interval around that win rate, however, is pretty wide, so it’s very possible that the true matchup win rate could be anywhere from 30/70 to 60/40. Additional data, both in this dataset and from other sources, is definitely needed here.

I suggest watching videos of people playing the deck. There are just so many interactions to explain that it’s better to see them played out in practice. Speaking of practice, the more you can play the deck (even just goldfishing!) the better you will be.

If I had to identify one critical nuance of the deck, it’s using bouncelands to return Tolaria West and Fountain to your hand. Knowing when to do that and how to sequence those turns is critical for Amulet success.

I am also finding that the Twin vs. Bloom matchup is probably not as heavily in favor as Twin players initially believed. Even with a pretty wide confidence interval, it’s still somewhere between about 48% and 85% (based on the data we have currently). That’s not even close to the 30-70 matchup some people were claiming. My guess is that it’s close to 50-50, but with player experience able to increase/decrease that by 10%+.

Also, i would really like to see Merfolk getting more spotlight. Rumors say, their Delver matchup is quite good. Is that true? How are their other matchups?

thanks, Tyrannon

Or am I missing something?

Kanister, mind if I quickly ask what are some the nuances that I should look out for as a Bloom player? I just picked up the deck about a month or two ago and the twin match up is still something I struggle with so any tips would be appreciated.

Is point 2 really valid, though?
http://www.channelfireball.com/articles/can-it-pay-off-to-play-a-high-variance-deck/
From what I have understanded from the linked article, it is just a problem coming from how people perceive the game, not an actual problem lessening your chances to win the tournament.

Homever, I believe that there is also a fourth reason. The vast majority of players perceive “fair” decks, fighting for card advantage turn by turn, as more fun to play. Even though the match can have only result in a win or a loss (or a draw, but nobody wants that), people in general feel much better when they lose after “putting up a fair fight” and die to a turn 9 Tarmogoyf swing, than they do dying to T5 Tarmogoyf swing, getting to play just a bunch of lands.