The event and deck N is smaller than I would like, but it's a very transparent dataset that mostly lines up with our general understanding of the format. The author also updates it semi-regularly, at least so far. We'll see if it goes the way of the dozen other Modern metagame projects before it. I disagree with the cutoffs, tier definitions, and event pool, but the methodology is mostly sound and I'd rather draw conclusions from this transparent data than a largely subjective evaluation of the format.
Thanks for sharing. I tend to agree with most that it looks generally accurate (as much as can be anyway).
Since unban the list look something like this in order of appearance based on top 16
11 5c Humans
9 Affinity
8 Hollow One
Tier 2
Affinity
Tier 3
Humans
Hollow One
This is the part that people have a disconnect with. When you put the top performing decks at lower tiers of play, the logic doesnt work out.
It's not my list it's mtgtop8 list you fraud can you read a bit more ffs.
You guys are baddies, I've got the same damn info anyway I'm out, the list you say have more potential take the exact same data that I have except it's not only based on event post unban thus mine is more accurate for the current meta.
11 5c Humans
9 Affinity
8 Hollow One
8 Blue Moon (breach)
6 U/W Control
5 Scapeshift
5 Burn
5 Storm
5 Jund
5 GTron
4 Dredge
4 Bogles
4 Grixis Death's Shadow
3 Eldrazi
3 Ad Nauseam
But since we came to the conclusion that you guys can't read might aswell stop posting and theory craft on my side.
I've said plenty of time that this list was GP phoenix and up.
It's not my list it's mtgtop8 list you fraud can you read a bit more ffs.
...
11 5c Humans
9 Affinity
8 Hollow One
8 Blue Moon (breach)
6 U/W Control
5 Scapeshift
5 Burn
5 Storm
5 Jund
5 GTron
4 Dredge
4 Bogles
4 Grixis Death's Shadow
3 Eldrazi
3 Ad Nauseam
Here's my question. I assume this is your/MTGT8's list of major finishes from major events, as you said on a previous page. I think everyone gets that. What people don't get (myself included) are the inconsistencies. For example, why are Humans and Affinity the top decks in this list I'm quoting, but in Tier 2 and 3 on your list on the previous page?
Red Deck Wins (12-14 creature burn)
UW Control
Humans
Death's Shadow
GR Ponza
Hollow One
Similarly, Tron and Jund see fewer finishes than five other decks, and the same finishes as three other decks, and yet fall in Tier 1. These kinds of inconsistencies are what people do not understand.
It's called "The Decks to Beat - March '18" and includes Walk's rankings. But I have absolutely no idea why they choose to rank them that way when all their other breakdowns rank them differently. See the following rankings that have a totally different order:
yeah thats why i pointed out sites like mtgtop8 existing forever. if their information was so reliable, why would we need a middle-man or this thread? we would just go to those sites lol
@Walk - its pretty immature to call people baddies because they find fault with how you came to your list. the post where you explained your reasoning included deck descriptions so far removed from my own that it just made everything else irrelevant. for example:
-death shadow is a cheese strat
-hollow one is only winning because people are bad
-knightfall is a bad combo with weird filler creatures
-storm is not fast
-UWR is a burn deck with counters (totally missing the fact that the top 16 finish you listed was a control build and not tempo).
if thats how you personally see those decks then thats whatever. the point is you made a listing based on that perception then challenged people to correct it.
i know im not gonna sit here and try to change your mind on deck strengths. awesome i convinced one person to maybe revisit a list that they came up with on a whim.
Private Mod Note
():
Rollback Post to RevisionRollBack
Modern: UWGSnow-Bant Control BURGrixis Death's Shadow GWBCoCo Elves WCDeath and Taxes (sold)
Christ. Why did we want the Tier list back again? It might legitimately be the most misused and misunderstood tool available to competitive Modern players. I think the mistake might be in giving them a numerical rating. Too many people seem to identify it as how effective a deck is, which to the best of my knowledge the tier list was never really meant to directly analyze (here on MTGS, anyway). If people really need a new way of dissecting the modern meta, maybe it should be a rarity list, like... Common, Uncommon, Rare, Mythic tier decks based on how often you can expect to run into those decks?
That's a good point about the names, since they do lead to a lot of confusion and misevaluation. I know some other games also use "tier" to mean "frequency of play," but I don't know how common that is and/or if there's an accepted term for frequency that's less value laden.
That said, Humans at Common and Skred at Mythic sends an even wrong-er signal IMO, since it has the same problem with being value laden but this time in the opposite direction.
Private Mod Note
():
Rollback Post to RevisionRollBack
Playing UX Mana Denial until Modern gets the answers it needs.
WUBRG Humans BRW Mardu Pyromancer UW UW "Control" UR Blue Moon
That's a good point about the names, since they do lead to a lot of confusion and misevaluation. I know some other games also use "tier" to mean "frequency of play," but I don't know how common that is and/or if there's an accepted term for frequency that's less value laden.
That said, Humans at Common and Skred at Mythic sends an even wrong-er signal IMO, since it has the same problem with being value laden but this time in the opposite direction.
Yeah, Mythic is a bit loaded, I suppose. Still, Common, Uncommon, and Rare seem like reasonably neutral terms. Maybe common is 6% of the meta or higher, uncommon is Between 6% and 1%, and rare is 1% or less. That said, I'm pretty skeptical you can make a truly accurate tier list with the available data sets.
Christ. Why did we want the Tier list back again? It might legitimately be the most misused and misunderstood tool available to competitive Modern players. I think the mistake might be in giving them a numerical rating. Too many people seem to identify it as how effective a deck is, which to the best of my knowledge the tier list was never really meant to directly analyze (here on MTGS, anyway). If people really need a new way of dissecting the modern meta, maybe it should be a rarity list, like... Common, Uncommon, Rare, Mythic tier decks based on how often you can expect to run into those decks?
I'll take a good tier list to show just how poor a tier list really represent a meta:
In Super Smash Bros Melee, Yoshi used to be considered a low tier pick but a japanese player by the name of aMSa by himself made Yoshi raise in tier by destroying event with it. It's good to consider that Melee never received any patch since it's a game cube game and thus at one point Yoshi was considered bad while the same version is now considered a viable pick in the right hand. Yoshi was ranked 20 out of 26 and climbed to 12 out of 26 in the last version I can find.
In MTG there is a Team Phenomenon where highly skilled player sit down and decide on which deck the team should play, Channel Fireball is an example of such team commonly running up to 8+ times the same deck piloted by skilled player. This should indeed show that if such team decide to play Human, well human will thus have a really good showing on day 2. The next event they might run Jund and all of a sudden Jund gets a good showing on day 2 etc.
This thus imply that since a proper tier list is based on result (and it's probably the best neutral approach to construct a list) there could always low key have a deck like WR control that could possibly be the best deck but since no one really play that archetype it's sometimes show up at some event and people do not take notice. I do not think WR control is great it was simply an example.
The best way to really observe a viable list is then to include Match Up which are really hard to get at the moment. With Match up we could easily do the math to obtain the median winrate and say for sure that X deck against a field of 1 of each deck has 50% chances to win the event.
BUT THEN again, let's say Jund is commonly seen as having 50% win rate against the field what happen if a deck has 60% win rate against the field meta? We can never really get any competitive edge with a tier list and the sole purpose of such list should be to identify which deck are worth investing in when looking to buy a new deck and makes for easy entry point for new player to at least understand which decks are dominant and which decks are currently lacking.
I love tier list and I feel that without Match up data we cannot really build one.
I'll take a good tier list to show just how poor a tier list really represent a meta:
In Super Smash Bros Melee, Yoshi used to be considered a low tier pick but a japanese player by the name of aMSa by himself made Yoshi raise in tier by destroying event with it. It's good to consider that Melee never received any patch
~snip~
I love tier list and I feel that without Match up data we cannot really build one.
Then we've come to pretty much the same conclusion. With our current information, any sort of value-based tier list is going to be hilariously inaccurate at best. There is currently no competitive or practical point to adding one to MTGSalvation at this time, since any conclusions we would draw from it would be quite skewed. We can expect a tier list when we get access to significantly more accurate win percentage and/or metagame percentage data.
Again, we don't need to fixate on an MWP-based tier list. Prevalence-based tiering is very useful when determining what decks you are likely to face and what matchups you need to prepare for. There is still enough data to create such a tier list if anyone wants to expend the effort and time.
I'm just gonna come and talk about aMSa and Yoshi for a sec. Amsa's never dominated a tournament or the scene in general. Melee is a strange esport because of how little in the way of major upsets there are. In fact, every major tournament in the past few years has been won by one of the Melee gods or the two godslayers Leffen/Plup. That's literally 7 players who've won everything in the past 10 years barring like one exception. Amsa's great, he's currently ranked 24th, and he's a ton of fun to watch. That being said, he isn't a dominant player, and Yoshi isn't that strong of a character.
Yoshi typically ends up around 12th on a tier list. Yes, that's a lot higher than Yoshi was ranked, and yeah aMSa is the reason. Yoshi has a lot of unexplored potential that players recognize. Parry is such a powerful and unique ability that lets him do some crazy things. One big thing going for Yoshi is the fact it's impossible to find good Yoshi players to practice against. Especially since aMSa's from Japan, he can come over and spike tournaments because people aren't prepared for the matchup.
aMSa is a lot like Bogles as a deck. Both attack on different angles than others do, and both are relatively uncommon. They can spike tournaments when people don't expect it, but people always know it's a thing. Yoshi, like Bogles, are never high up on the tier lists. And honestly, that makes sense. Tier lists aren't solely "what is the best deck to play", but also "what decks are my opponent's playing". Up until very recently, you were probably okay playing a deck that would auto lose to Bogles because you wouldn't run into it.
Now, you're basically saying we shouldn't have a tier list until we have perfect matchup data. That requires us to find a perfect pilot for every deck, have the player know how to play against every deck, and then have them play a large enough sample set. That's just completely infeasible. You're basically never going to get data you can feel good about in order to show matchup data. Especially since MTGO has to be one of the primary data sources, you just aren't finding good enough data for it to be reliable.
Ken knows this last point well, as I've seen some of the posts he's made on extensive matchup testing. In fact, I recall one in the 8Rack thread where he did a pile of plays on the play/draw in pre/post-board matchups after they had determined how both sides should play the matchup. It was a great read because he put in a few hours of testing to find the data he needed.
I'm just gonna come and talk about aMSa and Yoshi for a sec. Amsa's never dominated a tournament or the scene in general. Melee is a strange esport because of how little in the way of major upsets there are. In fact, every major tournament in the past few years has been won by one of the Melee gods or the two godslayers Leffen/Plup. That's literally 7 players who've won everything in the past 10 years barring like one exception. Amsa's great, he's currently ranked 24th, and he's a ton of fun to watch. That being said, he isn't a dominant player, and Yoshi isn't that strong of a character.
Yoshi typically ends up around 12th on a tier list. Yes, that's a lot higher than Yoshi was ranked, and yeah aMSa is the reason. Yoshi has a lot of unexplored potential that players recognize. Parry is such a powerful and unique ability that lets him do some crazy things. One big thing going for Yoshi is the fact it's impossible to find good Yoshi players to practice against. Especially since aMSa's from Japan, he can come over and spike tournaments because people aren't prepared for the matchup.
aMSa is a lot like Bogles as a deck. Both attack on different angles than others do, and both are relatively uncommon. They can spike tournaments when people don't expect it, but people always know it's a thing. Yoshi, like Bogles, are never high up on the tier lists. And honestly, that makes sense. Tier lists aren't solely "what is the best deck to play", but also "what decks are my opponent's playing". Up until very recently, you were probably okay playing a deck that would auto lose to Bogles because you wouldn't run into it.
Now, you're basically saying we shouldn't have a tier list until we have perfect matchup data. That requires us to find a perfect pilot for every deck, have the player know how to play against every deck, and then have them play a large enough sample set. That's just completely infeasible. You're basically never going to get data you can feel good about in order to show matchup data. Especially since MTGO has to be one of the primary data sources, you just aren't finding good enough data for it to be reliable.
Ken knows this last point well, as I've seen some of the posts he's made on extensive matchup testing. In fact, I recall one in the 8Rack thread where he did a pile of plays on the play/draw in pre/post-board matchups after they had determined how both sides should play the matchup. It was a great read because he put in a few hours of testing to find the data he needed.
I know that amsa isnt the best melee player the point was mostly to show how he pushed a character out of a tier by his skill alone.
I agree on your idea of finding proper pilot to play decks and find match up in this matter, sadly there is about 25+ active decks that has potential of being relevant in the current meta.We could easily identify the 10 most played decks (day 2 or top 16) but we still would require a large amount of games to end up with decent data, it is indeed possible and we do have quite some time before the next unbans/bans to play it out.
Sadly I am a paper player so I don't own any deck on MTGO, I was thinking about using alternative in the form of what tappedout offer with the playtest option and play games against myself doing the optimal line of play with both decks. I used to play chess against myself a lot back then and it did help in raising my elo in chess.
I think an hundred games per match up would be a valid starting point, it might seem arbitrary but it's a large enough sample to easily work it out.
The only thing that I am afraid comes from the sideboard choice, should we only run the sample on Game 1 or only on Game 2(3)?
Jund list is pretty stock but there is some variety in the sideboard mostly per say 2 cards against X deck or 3 could change the data quite a lot.
Even with the sideboard issue I would agree that it is the best current way to obtain viable data to build a tier list from.
The problem is you need to ensure the player is of the appropriate skill for the deck. I'm talking about PT level players not us here on MTGS forums. So even if you go and compile the data, your skill level with each of the decks comes into question and so the validity of the data is uncertain. Like I said, MWP based tier lists are infeasible because of data collection requirements/issues.
Thanks for sharing. I tend to agree with most that it looks generally accurate (as much as can be anyway).
Can you tell us where you found this?
RWG Burn
GW Abzan Company
It's not my list it's mtgtop8 list you fraud can you read a bit more ffs.
You guys are baddies, I've got the same damn info anyway I'm out, the list you say have more potential take the exact same data that I have except it's not only based on event post unban thus mine is more accurate for the current meta.
11 5c Humans
9 Affinity
8 Hollow One
8 Blue Moon (breach)
6 U/W Control
5 Scapeshift
5 Burn
5 Storm
5 Jund
5 GTron
4 Dredge
4 Bogles
4 Grixis Death's Shadow
3 Eldrazi
3 Ad Nauseam
But since we came to the conclusion that you guys can't read might aswell stop posting and theory craft on my side.
I've said plenty of time that this list was GP phoenix and up.
Spirits
Here's my question. I assume this is your/MTGT8's list of major finishes from major events, as you said on a previous page. I think everyone gets that. What people don't get (myself included) are the inconsistencies. For example, why are Humans and Affinity the top decks in this list I'm quoting, but in Tier 2 and 3 on your list on the previous page?
Similarly, Tron and Jund see fewer finishes than five other decks, and the same finishes as three other decks, and yet fall in Tier 1. These kinds of inconsistencies are what people do not understand.
EDIT: To be clear, I know what page you are citing. For those that don't, it's this one:
http://mtgtop8.com/event?e=18805&f=MO
It's called "The Decks to Beat - March '18" and includes Walk's rankings. But I have absolutely no idea why they choose to rank them that way when all their other breakdowns rank them differently. See the following rankings that have a totally different order:
Last 2 months: http://mtgtop8.com/format?f=MO&meta=51
Last 2 weeks: http://mtgtop8.com/format?f=MO&meta=54
Live tourneys last 2 months: http://mtgtop8.com/format?f=MO&meta=57
I have no clue how they chose the "Decks to Beat" order/tiering based on the actual data on that site. It seems very arbitrary.
@Walk - its pretty immature to call people baddies because they find fault with how you came to your list. the post where you explained your reasoning included deck descriptions so far removed from my own that it just made everything else irrelevant. for example:
-death shadow is a cheese strat
-hollow one is only winning because people are bad
-knightfall is a bad combo with weird filler creatures
-storm is not fast
-UWR is a burn deck with counters (totally missing the fact that the top 16 finish you listed was a control build and not tempo).
if thats how you personally see those decks then thats whatever. the point is you made a listing based on that perception then challenged people to correct it.
i know im not gonna sit here and try to change your mind on deck strengths. awesome i convinced one person to maybe revisit a list that they came up with on a whim.
UWGSnow-Bant Control
BURGrixis Death's Shadow
GWBCoCo Elves
WCDeath and Taxes(sold)That said, Humans at Common and Skred at Mythic sends an even wrong-er signal IMO, since it has the same problem with being value laden but this time in the opposite direction.
WUBRG Humans
BRW Mardu Pyromancer
UW UW "Control"
UR Blue Moon
Yeah, Mythic is a bit loaded, I suppose. Still, Common, Uncommon, and Rare seem like reasonably neutral terms. Maybe common is 6% of the meta or higher, uncommon is Between 6% and 1%, and rare is 1% or less. That said, I'm pretty skeptical you can make a truly accurate tier list with the available data sets.
It's been said in the past that mtgo do not reflect big event.
11 5c Humans
9 Affinity
8 Hollow One
8 Blue Moon (breach)
6 U/W Control
5 Scapeshift
5 Burn
5 Storm
5 Jund
5 GTron
4 Dredge
4 Bogles
4 Grixis Death's Shadow
3 Eldrazi
3 Ad Nauseam
This list is the count of deck with more than 3 showing since GP Phoenix according to the same data ktkenshinx linked.
I'll take a good tier list to show just how poor a tier list really represent a meta:
In Super Smash Bros Melee, Yoshi used to be considered a low tier pick but a japanese player by the name of aMSa by himself made Yoshi raise in tier by destroying event with it. It's good to consider that Melee never received any patch since it's a game cube game and thus at one point Yoshi was considered bad while the same version is now considered a viable pick in the right hand. Yoshi was ranked 20 out of 26 and climbed to 12 out of 26 in the last version I can find.
In MTG there is a Team Phenomenon where highly skilled player sit down and decide on which deck the team should play, Channel Fireball is an example of such team commonly running up to 8+ times the same deck piloted by skilled player. This should indeed show that if such team decide to play Human, well human will thus have a really good showing on day 2. The next event they might run Jund and all of a sudden Jund gets a good showing on day 2 etc.
This thus imply that since a proper tier list is based on result (and it's probably the best neutral approach to construct a list) there could always low key have a deck like WR control that could possibly be the best deck but since no one really play that archetype it's sometimes show up at some event and people do not take notice. I do not think WR control is great it was simply an example.
The best way to really observe a viable list is then to include Match Up which are really hard to get at the moment. With Match up we could easily do the math to obtain the median winrate and say for sure that X deck against a field of 1 of each deck has 50% chances to win the event.
BUT THEN again, let's say Jund is commonly seen as having 50% win rate against the field what happen if a deck has 60% win rate against the field meta? We can never really get any competitive edge with a tier list and the sole purpose of such list should be to identify which deck are worth investing in when looking to buy a new deck and makes for easy entry point for new player to at least understand which decks are dominant and which decks are currently lacking.
I love tier list and I feel that without Match up data we cannot really build one.
Then we've come to pretty much the same conclusion. With our current information, any sort of value-based tier list is going to be hilariously inaccurate at best. There is currently no competitive or practical point to adding one to MTGSalvation at this time, since any conclusions we would draw from it would be quite skewed. We can expect a tier list when we get access to significantly more accurate win percentage and/or metagame percentage data.
Yoshi typically ends up around 12th on a tier list. Yes, that's a lot higher than Yoshi was ranked, and yeah aMSa is the reason. Yoshi has a lot of unexplored potential that players recognize. Parry is such a powerful and unique ability that lets him do some crazy things. One big thing going for Yoshi is the fact it's impossible to find good Yoshi players to practice against. Especially since aMSa's from Japan, he can come over and spike tournaments because people aren't prepared for the matchup.
aMSa is a lot like Bogles as a deck. Both attack on different angles than others do, and both are relatively uncommon. They can spike tournaments when people don't expect it, but people always know it's a thing. Yoshi, like Bogles, are never high up on the tier lists. And honestly, that makes sense. Tier lists aren't solely "what is the best deck to play", but also "what decks are my opponent's playing". Up until very recently, you were probably okay playing a deck that would auto lose to Bogles because you wouldn't run into it.
Now, you're basically saying we shouldn't have a tier list until we have perfect matchup data. That requires us to find a perfect pilot for every deck, have the player know how to play against every deck, and then have them play a large enough sample set. That's just completely infeasible. You're basically never going to get data you can feel good about in order to show matchup data. Especially since MTGO has to be one of the primary data sources, you just aren't finding good enough data for it to be reliable.
Ken knows this last point well, as I've seen some of the posts he's made on extensive matchup testing. In fact, I recall one in the 8Rack thread where he did a pile of plays on the play/draw in pre/post-board matchups after they had determined how both sides should play the matchup. It was a great read because he put in a few hours of testing to find the data he needed.
Grixis Death's Shadow, Jund, UW Tron, Jeskai Control, Storm, Counters Company, Eldrazi Tron, Affinity, Living End, Infect, Merfolk, Dredge, Ad Nauseam, Amulet, Bogles, Eldrazi Tron, Mono U Tron, Lantern, Mardu Pyromancer
I know that amsa isnt the best melee player the point was mostly to show how he pushed a character out of a tier by his skill alone.
I agree on your idea of finding proper pilot to play decks and find match up in this matter, sadly there is about 25+ active decks that has potential of being relevant in the current meta.We could easily identify the 10 most played decks (day 2 or top 16) but we still would require a large amount of games to end up with decent data, it is indeed possible and we do have quite some time before the next unbans/bans to play it out.
Sadly I am a paper player so I don't own any deck on MTGO, I was thinking about using alternative in the form of what tappedout offer with the playtest option and play games against myself doing the optimal line of play with both decks. I used to play chess against myself a lot back then and it did help in raising my elo in chess.
I think an hundred games per match up would be a valid starting point, it might seem arbitrary but it's a large enough sample to easily work it out.
The only thing that I am afraid comes from the sideboard choice, should we only run the sample on Game 1 or only on Game 2(3)?
Jund list is pretty stock but there is some variety in the sideboard mostly per say 2 cards against X deck or 3 could change the data quite a lot.
Even with the sideboard issue I would agree that it is the best current way to obtain viable data to build a tier list from.
Grixis Death's Shadow, Jund, UW Tron, Jeskai Control, Storm, Counters Company, Eldrazi Tron, Affinity, Living End, Infect, Merfolk, Dredge, Ad Nauseam, Amulet, Bogles, Eldrazi Tron, Mono U Tron, Lantern, Mardu Pyromancer