How about instead of a tier system, you just have a proven section for deck that have top whatever( the mods can decide the cut off point) and the decks have a percentage next to their name showing how often they have been in that top part of the events they get data from. If its the SCG, PTQ, and GPQ so be it. Update the percentages or have them for the last 30 days or 90 days.
How about instead of a tier system, you just have a proven section for deck that have top whatever( the mods can decide the cut off point) and the decks have a percentage next to their name showing how often they have been in that top part of the events they get data from. If its the SCG, PTQ, and GPQ so be it. Update the percentages or have them for the last 30 days or 90 days.
Isn't that still just a tiering system?
It removes the popularity aspect of the tier systems. It would be straight percentages.
I would suggest it be top 16 showings. top 8s is too narrow, and top 32 would make the list huge.
In the end if we knew the percentages were based on 3 events (SCG, PPTQs/PT, and GPQ/GP) and the cut off (top 8,16, or 32). Then when someone reads the list of proven decks, there is no other factors to sway opinion other then straight numbers.
The problem with most tiering systems is popularity comes into play. Unless we can get the numbers of the decks starting in an event, and we can compare those starting numbers to what we see day 2 and in top spots. Popularity is a baseless skewed number.
Example to explain the above..
If we know in a field of 200 players 50 decks are Affinity, and 25% of the top 16 is Affinity, we can say Affinity was represented properly.
Now if Affinity was only 10 of those 200 starting decks and Affinity was still 25% of the top spots, we can say the deck over performed.
Now if 100 decks of the starting 200 are Affinity and only 25% fall to top spots the deck under preformed, talking straight numbers.
But that is a lot more leg work and crunching of numbers if they can be found. Which I believe we cant.
How about instead of a tier system, you just have a proven section for deck that have top whatever( the mods can decide the cut off point) and the decks have a percentage next to their name showing how often they have been in that top part of the events they get data from. If its the SCG, PTQ, and GPQ so be it. Update the percentages or have them for the last 30 days or 90 days.
Isn't that still just a tiering system?
It removes the popularity aspect of the tier systems. It would be straight percentages.
I would suggest it be top 16 showings. top 8s is too narrow, and top 32 would make the list huge.
In the end if we knew the percentages were based on 3 events (SCG, PPTQs/PT, and GPQ/GP) and the cut off (top 8,16, or 32). Then when someone reads the list of proven decks, there is no other factors to sway opinion other then straight numbers.
The problem with most tiering systems is popularity comes into play. Unless we can get the numbers of the decks starting in an event, and we can compare those starting numbers to what we see day 2 and in top spots. Popularity is a baseless skewed number.
Popularity comes into play with your system too. More popular decks will top more often, and will be represented higher regardless of their actual potential.
How about instead of a tier system, you just have a proven section for deck that have top whatever( the mods can decide the cut off point) and the decks have a percentage next to their name showing how often they have been in that top part of the events they get data from. If its the SCG, PTQ, and GPQ so be it. Update the percentages or have them for the last 30 days or 90 days.
Isn't that still just a tiering system?
It removes the popularity aspect of the tier systems. It would be straight percentages.
I would suggest it be top 16 showings. top 8s is too narrow, and top 32 would make the list huge.
In the end if we knew the percentages were based on 3 events (SCG, PPTQs/PT, and GPQ/GP) and the cut off (top 8,16, or 32). Then when someone reads the list of proven decks, there is no other factors to sway opinion other then straight numbers.
The problem with most tiering systems is popularity comes into play. Unless we can get the numbers of the decks starting in an event, and we can compare those starting numbers to what we see day 2 and in top spots. Popularity is a baseless skewed number.
Popularity comes into play with your system too. More popular decks will top more often, and will be represented higher regardless of their actual potential.
Not true, you are disregarding skill level of the pilots. In reality on any given weekend for the SCG events that have 200+ players in them, maybe 50 have a realistic shot at winning the event. if they pay out to 64, that leaves 14 spots for the grinders or 'casual' players to squeeze into the money spots.
If you wanted to make a true tier system, you would only look at decks people that get paid to play, play. you could ignore the rest of the field. This would remove popularity, the skill factor, the money factor and any other negative about current tier systems. We would know the decks are playable and in the end thats all anyone really cares about since we are basing all this off the highest level events. Which I feel is wrong in itself, but we dont get the complete data from every where else so .. its doing the best we can do and removing as much of the negative from the process.
I'm just going to quote a Ktkenshinx post from a few pages back because I think it accurately addresses the concerns you are bringing up and I think it bears repeating. I don't like just verbatim quoting but he does a good enough job I don't feel any need to expand, rather just bring it back to light.
I'm not sure about the new system's specifics, but if the site uses an established, replicable, and transparent scoring/grading system, then tiers are fine. In my opinion, tiering should be as objective as possible and also use the best statistical methods available given the data. If this can happen, the subjective "established" and "proven" division should not return; Modern did this years ago when I first got here and we moved away from it.
On the other hand, if the system doesn't make sense or doesn't meet those above benchmarks, then tiering becomes much less helpful. I am not as familiar with current format data as I was with it in the past, but I sincerely doubt there are truly this many Tier 1 decks. The most I ever saw when doing updates was maybe 8-9 (estimate; don't quote me). With so many Tier 1 decks, I suspect it's because the new system uses a very broad definition of Tier 1 that might not be as helpful as people think it is. Personally, I prefer narrower definitions that establish tiers with more confidence. No system is perfect without the full dataset, but I'd be cautious about a system that had so many Tier 1 decks; it's unlikely all Tier 1 decks are truly as viable as each other if the tiering is so wide.
I'm just going to quote a Ktkenshinx post from a few pages back because I think it accurately addresses the concerns you are bringing up and I think it bears repeating. I don't like just verbatim quoting but he does a good enough job I don't feel any need to expand, rather just bring it back to light.
I'm not sure about the new system's specifics, but if the site uses an established, replicable, and transparent scoring/grading system, then tiers are fine. In my opinion, tiering should be as objective as possible and also use the best statistical methods available given the data. If this can happen, the subjective "established" and "proven" division should not return; Modern did this years ago when I first got here and we moved away from it.
On the other hand, if the system doesn't make sense or doesn't meet those above benchmarks, then tiering becomes much less helpful. I am not as familiar with current format data as I was with it in the past, but I sincerely doubt there are truly this many Tier 1 decks. The most I ever saw when doing updates was maybe 8-9 (estimate; don't quote me). With so many Tier 1 decks, I suspect it's because the new system uses a very broad definition of Tier 1 that might not be as helpful as people think it is. Personally, I prefer narrower definitions that establish tiers with more confidence. No system is perfect without the full dataset, but I'd be cautious about a system that had so many Tier 1 decks; it's unlikely all Tier 1 decks are truly as viable as each other if the tiering is so wide.
I am just going to say I didnt agree with the old system any more then I agree with the new system. Its a flawed system with incomplete data.
By defining the data, only using 'X' amount of events, and only using percentages of tops (agreed cut off). Its a much easier, and clearer picture of the performance of the decks in question.
Until we can get information such as how many of each deck type was entered to how many topped from all events a tier system is going to be skewed.
I would also say there are more T1 decks then is being shown. Simply because the events the data is being taken from is some of the highest level we have for the game. So it would mean there is better pilots, hence better results with less optimal decks. We all know Modern is a format of comfort. Play the deck you feel more comfortable playing.
First: Lantern, you are beholden to nobody. I (and others) asked for criteria just so we could make sense of what was done, but in the end this is a hobby for all or most of us.
Second: We need to try to establish what tiers are. You can see over the last several pages some discussion around this.
Are they representation?
Are they power?
Are they performance and results?
Or is it a combination? I dont think we can base it on power, as thats difficult to judge. Nor can we base it on representation, because not everyone is going to flip to Jund, from some Jank, just because they are told Jund is tier 1.
The most fair metric, will be results.
If we had a way to track data for Top 8's and 5-0's, that would be ideal to see general trends, but even those numbers as HolyDiva outlined could EASILY be skewed by a few grinders in the case of MTGO.
People need to stop being completely infatuated with tiers, and I know I do it too, but decks that are not Tier 1 can have play, even if not popular.
If we had a way to track data for Top 8's and 5-0's, that would be ideal to see general trends, but even those numbers as HolyDiva outlined could EASILY be skewed by a few grinders in the case of MTGO.
This. I play goblins in legacy, and in terms of MWP it's very very good. But because representation is low, it rarely breaks T8. Low representation decks underperform in results based metrics.
First: Lantern, you are beholden to nobody. I (and others) asked for criteria just so we could make sense of what was done, but in the end this is a hobby for all or most of us.
Second: We need to try to establish what tiers are. You can see over the last several pages some discussion around this.
Are they representation?
Are they power?
Are they performance and results?
Or is it a combination? I dont think we can base it on power, as thats difficult to judge. Nor can we base it on representation, because not everyone is going to flip to Jund, from some Jank, just because they are told Jund is tier 1.
The most fair metric, will be results.
If we had a way to track data for Top 8's and 5-0's, that would be ideal to see general trends, but even those numbers as HolyDiva outlined could EASILY be skewed by a few grinders in the case of MTGO.
People need to stop being completely infatuated with tiers, and I know I do it too, but decks that are not Tier 1 can have play, even if not popular.
This. I play goblins in legacy, and in terms of MWP it's very very good. But because representation is low, it rarely breaks T8. Low representation decks underperform in results based metrics.
There is no good dataset for MWPs. Data either comes from single players (small N, player skill can matter more than deck performance, high matchup variance, etc.) or from single events (similar issues as above, notably with sample and match variance). This would have been feasible in the days of MTGO replays and scraping software, but it isn't now.
I really don't get why people are trying to reinvent the wheel given Modern's current information sources. The prevalence-based tiering system was largely accurate and largely liked, and that has been true for years. If we had access to stuff like MWPs or matchup-specific data then great, I'm sure many would happily use it. Same if we had Day 1 to Day 2 to Top 8 conversion rates. But we don't have any of that information, especially not on a consistent basis. Because of that, prevalence-based metrics are the next best thing. Those metrics also aren't inaccurate. Even when we did get conversion rate data and MWP information, it rarely changed the top-tier picture too much, and few sleepers like Amulet Bloom and Lantern were missed.
It's a bummer that Modern Nexus is behind with consistent metagame updates. Mainphase's updates are frequent but they have small Ns and questionable (maybe even arbitrary) cutoffs. Hopefully the format can get the best of both those worlds soon, or someone steps up to fill in those gaps.
Can anyone explain to me how the Tier system has changed on this site? Does it have to do with Ktkenshinx leaving?
Apparently, Lantern has changed both the data source and the tiering cutoffs. Now it seems +1% decks are Tier 2 and +3% decks are Tier 1, roughyl. No official layout and/or explanation of changes by any mod so far.
Thank you. It would be nice if the staff were more transparent about it.
I am. Did you message me? Ive also been commenting on all the deck threads that asked about it.
I'm not a god, I'm not telepathic. I'm a Television Studio Producer. I work about 75 hours a week, and do this on the side. No one is paying me. And my main Mod, who I trained and have been working with for literally years is leaving me. Next time anyone requests me to magically fix everything, they can just quote this.
No offence. I've been modding this sub for literally a decade at this point, and its mostly people telling me how things are wrong. Sometimes I get thanked, most times I dont. I'm training 2 new mods now, so Maybe I can be free enough to actually do more work in the tiering.
I'll be as clear as day here... and again. Yall can quote me on this if anyone asks again, it would save me the leg work of people thinking Im shafting them.
THERE IS NO GOOD DATA FOR ME TO DRAW ON. Point blank. I can use MTG goldfish, but their data is incomplete and they only hold a month of data on file, and when they update they dump 3 weeks of data and start fresh every month. Thats pretty bad for a non rotating slow to change format.
I can use top 8, like I'm doing right now, that does 2 months of data, but has random data holes and weighs everything pretty equally. So theres problems there too.
I can use modern nexus... when they remember to update. They are always late, and depending on them lately has been like throwing a paper plane and hoping it comes back sometime. (If you are reading this modern nexus, sorry. I was extremely pleased by the site, but every update has been weeks late for about a half a year now.)
I can use mainphase's data. Even messaged them (because people assume I havent been trying to fix this problem) asking about the data and more information. Im waiting on replies.
To be clear. You are asking me to be on top of an issue that has no good answers while I also have to feed my family AND train 2 new mods and clean up a forum missing the best partner I ever had.
I'm not going to ask for sympathy over here, or call out selfishness, but I will ask for understanding and patience. Come on. Cut some slack.
Do you adjust for MTGO figures? The challenge is MTGO does not have top 8. As a result if you try to apply a top 8 methodology to both MTGO and paper magic equally... it's going to skew results.
For example, in Legacy D&T sees little to no play on MTGO due to the cost. As a result if you were to look at the top 8 statistics including MTGO it looks like D&T is only 3-5% of the meta (when in fact, it is far more popular in paper). When you look at paper only events, the meta of D&T raises to 7%. Which is a substantial increase.
Realistically, I know some people want tiering to be based off skill and skill alone, but we haven't had access to win rates in forever, and most tiering systems, if not all of them, still have some fraction of popularity bais too. A good example is pokemon. Pokemon tiering does go off win rates... but the exact tiering they use in smogon still has people voting for the pokemon to be promoted or demoted based on wins they need to supply test evidence with. No one really loves chancy but wont deny how good of a wall it is, but gengar is currently rank A, even though its been nerfed to hell because people still like gengar.
Our tiering still represents popularity AND wins:
Tier 1- A deck you WILL see in day 2 GP like touriments
Tier 2- A deck you have a fair chance at seeing in a day 2 GP
Proven- Decks that either were good in the past and people still might run, or decks that are rising in popularity. Its unlikely you will see them, but you should know what they are to be prepared.
Wins factor in, but so does popularity. If I, a guy who knows the Ins and outs of Zoo like pretty much no one else, take zoo, a barely tier 2 if not just proven deck to a gp... I probably will day 2 with it. Some people could see my success, start to use it and just like that the deck is winning more, based off popularity and presived "good deck"
Zoo is a great deck, but the decks it beats, delver and control arent popular right now, thus it is a bad deck... Popularity and success are intermixed there.
THERE IS NO GOOD DATA FOR ME TO DRAW ON. Point blank. I can use MTG goldfish, but their data is incomplete and they only hold a month of data on file, and when they update they dump 3 weeks of data and start fresh every month. Thats pretty bad for a non rotating slow to change format.
Would it be possible for mtgs to build it's own data pool over time? For example storing others data and using it to build trends after a few months?
Sure it could be. But we run into the same problem as last time. People dont want to spend time to do it and give it up. Im not trying to reinvent my data source every half year
First: Lantern, you are beholden to nobody. I (and others) asked for criteria just so we could make sense of what was done, but in the end this is a hobby for all or most of us.
Second: We need to try to establish what tiers are. You can see over the last several pages some discussion around this.
Are they representation?
Are they power?
Are they performance and results?
Or is it a combination? I dont think we can base it on power, as thats difficult to judge. Nor can we base it on representation, because not everyone is going to flip to Jund, from some Jank, just because they are told Jund is tier 1.
The most fair metric, will be results.
If we had a way to track data for Top 8's and 5-0's, that would be ideal to see general trends, but even those numbers as HolyDiva outlined could EASILY be skewed by a few grinders in the case of MTGO.
People need to stop being completely infatuated with tiers, and I know I do it too, but decks that are not Tier 1 can have play, even if not popular.
This. I play goblins in legacy, and in terms of MWP it's very very good. But because representation is low, it rarely breaks T8. Low representation decks underperform in results based metrics.
There is no good dataset for MWPs. Data either comes from single players (small N, player skill can matter more than deck performance, high matchup variance, etc.) or from single events (similar issues as above, notably with sample and match variance). This would have been feasible in the days of MTGO replays and scraping software, but it isn't now.
I really don't get why people are trying to reinvent the wheel given Modern's current information sources. The prevalence-based tiering system was largely accurate and largely liked, and that has been true for years. If we had access to stuff like MWPs or matchup-specific data then great, I'm sure many would happily use it. Same if we had Day 1 to Day 2 to Top 8 conversion rates. But we don't have any of that information, especially not on a consistent basis. Because of that, prevalence-based metrics are the next best thing. Those metrics also aren't inaccurate. Even when we did get conversion rate data and MWP information, it rarely changed the top-tier picture too much, and few sleepers like Amulet Bloom and Lantern were missed.
It's a bummer that Modern Nexus is behind with consistent metagame updates. Mainphase's updates are frequent but they have small Ns and questionable (maybe even arbitrary) cutoffs. Hopefully the format can get the best of both those worlds soon, or someone steps up to fill in those gaps.
I keep nexus open on my tabs. I refresh it once a day to see if it has updated. Were 3 days late at this point, and everytime its late I have to change my "when does the forum refresh" date. It sucks. I wish it was better. I like the site.
Apparently, Lantern has changed both the data source and the tiering cutoffs. Now it seems +1% decks are Tier 2 and +3% decks are Tier 1, roughyl. No official layout and/or explanation of changes by any mod so far.
Thank you. It would be nice if the staff were more transparent about it.
I am. Did you message me? Ive also been commenting on all the deck threads that asked about it.
I'm not a god, I'm not telepathic. I'm a Television Studio Producer. I work about 75 hours a week, and do this on the side. No one is paying me. And my main Mod, who I trained and have been working with for literally years is leaving me. Next time anyone requests me to magically fix everything, they can just quote this.
No offence. I've been modding this sub for literally a decade at this point, and its mostly people telling me how things are wrong. Sometimes I get thanked, most times I dont. I'm training 2 new mods now, so Maybe I can be free enough to actually do more work in the tiering.
I'll be as clear as day here... and again. Yall can quote me on this if anyone asks again, it would save me the leg work of people thinking Im shafting them.
THERE IS NO GOOD DATA FOR ME TO DRAW ON. Point blank. I can use MTG goldfish, but their data is incomplete and they only hold a month of data on file, and when they update they dump 3 weeks of data and start fresh every month. Thats pretty bad for a non rotating slow to change format.
I can use top 8, like I'm doing right now, that does 2 months of data, but has random data holes and weighs everything pretty equally. So theres problems there too.
I can use modern nexus... when they remember to update. They are always late, and depending on them lately has been like throwing a paper plane and hoping it comes back sometime. (If you are reading this modern nexus, sorry. I was extremely pleased by the site, but every update has been weeks late for about a half a year now.)
I can use mainphase's data. Even messaged them (because people assume I havent been trying to fix this problem) asking about the data and more information. Im waiting on replies.
To be clear. You are asking me to be on top of an issue that has no good answers while I also have to feed my family AND train 2 new mods and clean up a forum missing the best partner I ever had.
I'm not going to ask for sympathy over here, or call out selfishness, but I will ask for understanding and patience. Come on. Cut some slack.
Do you adjust for MTGO figures? The challenge is MTGO does not have top 8. As a result if you try to apply a top 8 methodology to both MTGO and paper magic equally... it's going to skew results.
For example, in Legacy D&T sees little to no play on MTGO due to the cost. As a result if you were to look at the top 8 statistics including MTGO it looks like D&T is only 3-5% of the meta (when in fact, it is far more popular in paper). When you look at paper only events, the meta of D&T raises to 7%. Which is a substantial increase.
Realistically, I know some people want tiering to be based off skill and skill alone, but we haven't had access to win rates in forever, and most tiering systems, if not all of them, still have some fraction of popularity bais too. A good example is pokemon. Pokemon tiering does go off win rates... but the exact tiering they use in smogon still has people voting for the pokemon to be promoted or demoted based on wins they need to supply test evidence with. No one really loves chancy but wont deny how good of a wall it is, but gengar is currently rank A, even though its been nerfed to hell because people still like gengar.
Our tiering still represents popularity AND wins:
Tier 1- A deck you WILL see in day 2 GP like touriments
Tier 2- A deck you have a fair chance at seeing in a day 2 GP
Proven- Decks that either were good in the past and people still might run, or decks that are rising in popularity. Its unlikely you will see them, but you should know what they are to be prepared.
Wins factor in, but so does popularity. If I, a guy who knows the Ins and outs of Zoo like pretty much no one else, take zoo, a barely tier 2 if not just proven deck to a gp... I probably will day 2 with it. Some people could see my success, start to use it and just like that the deck is winning more, based off popularity and presived "good deck"
Zoo is a great deck, but the decks it beats, delver and control arent popular right now, thus it is a bad deck... Popularity and success are intermixed there.
THERE IS NO GOOD DATA FOR ME TO DRAW ON. Point blank. I can use MTG goldfish, but their data is incomplete and they only hold a month of data on file, and when they update they dump 3 weeks of data and start fresh every month. Thats pretty bad for a non rotating slow to change format.
Would it be possible for mtgs to build it's own data pool over time? For example storing others data and using it to build trends after a few months?
Sure it could be. But we run into the same problem as last time. People dont want to spend time to do it and give it up. Im not trying to reinvent my data source every half year
First: Lantern, you are beholden to nobody. I (and others) asked for criteria just so we could make sense of what was done, but in the end this is a hobby for all or most of us.
Second: We need to try to establish what tiers are. You can see over the last several pages some discussion around this.
Are they representation?
Are they power?
Are they performance and results?
Or is it a combination? I dont think we can base it on power, as thats difficult to judge. Nor can we base it on representation, because not everyone is going to flip to Jund, from some Jank, just because they are told Jund is tier 1.
The most fair metric, will be results.
If we had a way to track data for Top 8's and 5-0's, that would be ideal to see general trends, but even those numbers as HolyDiva outlined could EASILY be skewed by a few grinders in the case of MTGO.
People need to stop being completely infatuated with tiers, and I know I do it too, but decks that are not Tier 1 can have play, even if not popular.
This. I play goblins in legacy, and in terms of MWP it's very very good. But because representation is low, it rarely breaks T8. Low representation decks underperform in results based metrics.
There is no good dataset for MWPs. Data either comes from single players (small N, player skill can matter more than deck performance, high matchup variance, etc.) or from single events (similar issues as above, notably with sample and match variance). This would have been feasible in the days of MTGO replays and scraping software, but it isn't now.
I really don't get why people are trying to reinvent the wheel given Modern's current information sources. The prevalence-based tiering system was largely accurate and largely liked, and that has been true for years. If we had access to stuff like MWPs or matchup-specific data then great, I'm sure many would happily use it. Same if we had Day 1 to Day 2 to Top 8 conversion rates. But we don't have any of that information, especially not on a consistent basis. Because of that, prevalence-based metrics are the next best thing. Those metrics also aren't inaccurate. Even when we did get conversion rate data and MWP information, it rarely changed the top-tier picture too much, and few sleepers like Amulet Bloom and Lantern were missed.
It's a bummer that Modern Nexus is behind with consistent metagame updates. Mainphase's updates are frequent but they have small Ns and questionable (maybe even arbitrary) cutoffs. Hopefully the format can get the best of both those worlds soon, or someone steps up to fill in those gaps.
I keep nexus open on my tabs. I refresh it once a day to see if it has updated. Were 3 days late at this point, and everytime its late I have to change my "when does the forum refresh" date. It sucks. I wish it was better. I like the site.
I think that is part of the confusion, is the new tier system is alot more lax.
I feel like the old tiering system was as follows (feel free to disagree)
Tier 1- A deck you WILL see at the top tables. As well as day 2 representation 99% of the time.
Tier 2- A deck you have a fair chance at seeing in a day 2 GP and occasional top table appearances.
Proven- Decks that either were good in the past and people still might run, or decks that are rising in popularity. Its unlikely you will see them, but you should know what they are to be prepared
I feel like the new tier system is missing the "top tables" element. But due to the lack of data, there may not be anyway around that.
What if the reason WotC is trying to hard to keep data from the players is that MTGO data and even some of the tournament data (particularly some things like W/L ratio of some decks) shows that the game is actually rather badly balanced, and some decks are far and away better than the others, and it's only luck that few enough players are playing those decks that they get weeded out in tournaments and such. They aren't quite good enough on WotC's metrics to get bans, other decks leak through to top results due to pure luck reliably enough to hide some aspects of their performance, but WotC has seen the way the secondary market reacts to consistently winning decks, and their analysts have indicated that if players had the data, the deck prices of top decks would skyrocket so high that WotC wouldn't be able to keep up with reprints and it would create a really unfriendly metagame due to more competitive people shifting en-mass to those decks, and it would push out of the game people with more limited budgets or somewhat less reliable decks whose decks would look a lot worse when facing up against a higher ratio of top performing decks, and also make the prices of the secondary market possibly reach unreasonable levels, especially with the long delay WotC has on designing a products until it actually reaches the market, and how quickly the secondary market reacts just to the data we already have like top 8 tournament results.
In this sense, WotC could be trying to avoid data that would shape the meta.
It's also possible that if that data does exist, it would likely force results so impressive that WotC would be caught in a cycle of endless bans, because the format is actually rather poorly balanced, and only luck and currently lower metagame shares keeps decks with higher w/l ratios from exploding out of control in metagame shares due to more and more spikes picking them up, potentially leading to WotC forced to do chains of bans that would be highly unpopular every time a new deck reaches the top, since the format doesn't have the tools to self-correct against many of the secret top decks WotC has data indicating.
On using MTGO data: It seems MTGO is littered with weird and random brews that get wins because of being weird and random. Like I just played against UW deck that tries to win on the back of life gain lands and Mulldrifters. Then played against some Bant Glittering Wish deck that I still don't really know what it does other than cast Search for Tomorrow, Spreading Seas, and Auriok Champion, while Wishing for Dragonlord Dromoka and Teferi's Moat, of all things. It's definitely not as prevalent as you would see in the Leagues (and until I can mirror my paper decks online without breaking the bank I stick to the 2-man queues), but lots of people are playing with weird decks, and even Leagues are relatively small events (akin to a large 5-round FNM). It creates pockets of data that can be swayed wildly by variance, matchups, and repetitious grinders. It seems lots of people are trying to play "gotcha" decks and specifically plan to play decks people don't have sideboard plans for. If nothing else, it makes for interesting games in an interesting format, but not exactly helpful for judging and classifying decks.
In this sense, WotC could be trying to avoid data that would shape the meta.
It's also possible that if that data does exist, it would likely force results so impressive that WotC would be caught in a cycle of endless bans, because the format is actually rather poorly balanced, and only luck and currently lower metagame shares keeps decks with higher w/l ratios from exploding out of control in metagame shares due to more and more spikes picking them up, potentially leading to WotC forced to do chains of bans that would be highly unpopular every time a new deck reaches the top, since the format doesn't have the tools to self-correct against many of the secret top decks WotC has data indicating.
That's basically it. Watch any Modern deck tech and you'll have 10 different players, on 10 different decks, all saying their deck is the best positioned deck in the meta. WotC loves that. They want people to think they've figured it out, and for them to be wrong. It keeps a market alive as people switch decks and make guesses. It also means things don't stagnate into solid data where cards are overpowered based purely on recorded results. With a less clear meta, a card has to really overperform to be ban worthy.
This is certainly an interesting topic and there are multiple approaches to the data we work with. But I would really like if we shifted the discussion away from tinfoil hats and conspiracy theories. At least share your sources and evidence with the thread guys.
To me, it seems everyone has their own needs for the tierings, and this requires both different datasets and cutoffs. USA vs non-USA regional metagames, online vs paper metagames, FNM vs GP metagames, it goes on and on.
The solution isn't so simple. Nexus is comprehensive but too far behind, Goldfish only tracks last 30 days and is heavily skewed towards online and USA events, MainPhase just tracks Goldfish weekly, MtgTop8 has some weird classification issues, I personally just track online stuff.
Point is, nothing is perfect, but you can always choose to use whatever fits best your own needs. If this specific forum chooses to use X dataset with Y cutoffs, don't make it such a big deal. If it doesn't fit your needs, you are free to look elsewhere or do it yourself. And that's fine:)
My only problem with the new tiers here was the lack of the explanation, not the changes themselves. And I think I wasn't the only one, as many here expressed they couldn't make sense of it. Whatever you finally do Lantern, just make it clear to everyone (dataset and cutoffs) so everyone knows what's happening.
Random thought: can't we just have all primers in one single section? As long as people can easily search for the deck thread they want to post in, we should be fine. Is it that necessary?
I agree with the one section for all the primers that are consistently so called Tier 1 and 2 at least.
The idea has been around for a while. There are a new form of deck and primer thingy mtgsal is looking to implement in the future that might make that the best solution at some point, and if the system changes, well likely adpot that model.
I'm not trying to be vague on that one. I literally dont know the details and its been in the works back burner for a while.
On using MTGO data: It seems MTGO is littered with weird and random brews that get wins because of being weird and random. Like I just played against UW deck that tries to win on the back of life gain lands and Mulldrifters. Then played against some Bant Glittering Wish deck that I still don't really know what it does other than cast Search for Tomorrow, Spreading Seas, and Auriok Champion, while Wishing for Dragonlord Dromoka and Teferi's Moat, of all things. It's definitely not as prevalent as you would see in the Leagues (and until I can mirror my paper decks online without breaking the bank I stick to the 2-man queues), but lots of people are playing with weird decks, and even Leagues are relatively small events (akin to a large 5-round FNM). It creates pockets of data that can be swayed wildly by variance, matchups, and repetitious grinders. It seems lots of people are trying to play "gotcha" decks and specifically plan to play decks people don't have sideboard plans for. If nothing else, it makes for interesting games in an interesting format, but not exactly helpful for judging and classifying decks.
I know some dont want to hear it or even believe it, but this is exactly what the game of Magic use to be. An established pecking order of decks, and players brewing to beat that pecking order, keeping the pecking order always in a state of change. Now the player base has gotten lazy and only plays proven decks and sits at events playing mirror matches complaining about this or that. That is why I dont believe in the Tier system. Never have. Its nice to know what you may see, and what is the flavor of the week so you can tweak your deck to be competitive, but nothing more, nothing less then that.
On using MTGO data: It seems MTGO is littered with weird and random brews that get wins because of being weird and random. Like I just played against UW deck that tries to win on the back of life gain lands and Mulldrifters. Then played against some Bant Glittering Wish deck that I still don't really know what it does other than cast Search for Tomorrow, Spreading Seas, and Auriok Champion, while Wishing for Dragonlord Dromoka and Teferi's Moat, of all things. It's definitely not as prevalent as you would see in the Leagues (and until I can mirror my paper decks online without breaking the bank I stick to the 2-man queues), but lots of people are playing with weird decks, and even Leagues are relatively small events (akin to a large 5-round FNM). It creates pockets of data that can be swayed wildly by variance, matchups, and repetitious grinders. It seems lots of people are trying to play "gotcha" decks and specifically plan to play decks people don't have sideboard plans for. If nothing else, it makes for interesting games in an interesting format, but not exactly helpful for judging and classifying decks.
I know some dont want to hear it or even believe it, but this is exactly what the game of Magic use to be. An established pecking order of decks, and players brewing to beat that pecking order, keeping the pecking order always in a state of change. Now the player base has gotten lazy and only plays proven decks and sits at events playing mirror matches complaining about this or that. That is why I dont believe in the Tier system. Never have. Its nice to know what you may see, and what is the flavor of the week so you can tweak your deck to be competitive, but nothing more, nothing less then that.
To Many players want to be followers and not enough want to forge there own destiny. It's not just being lazy it's being able to have instant access to information all the time. The internet is changing people in good ways and bad. I totally agree with you.
To Many players want to be followers and not enough want to forge there own destiny. It's not just being lazy it's being able to have instant access to information all the time. The internet is changing people in good ways and bad. I totally agree with you.
It's bad to want your deck to be good when you drop a few hundred dollars building it? Even for the best deck builders in the world, only 10% of their ideas are any good. Some random person is going to have a much lower rate than that (if they're even capable of judging their deck as good/bad). Knowing you have a good deck lets you focus on playing the game.
No, But a Burn list from a year ago would still be a viable deck.
That's only true of some deck types though. I love to build my own lists (for example, I've got a pretty good Mardu DS right now), but I still need a starting point for them. That could be building off of an existing archetype to try and do it better, or it could simply be in building a meta deck to beat what other people are doing. In either case, I feel as a deck builder that having all of this information access only helps me rather than hinders me. It hasn't made me lazy, it's given me a better starting point for anything I choose to build. And it has the advantage that when I want something safe, I can filter my options to those decks and make a choice.
Yep, exactly. Here's some of the decks that I've built since they came out and still have now.
Tron - still viable and strong
Jund - still viable and a bit less strong right now
Pod - banned, but some pieces used in Abzan Company
Twin - banned, but the other cards are staples that I've used in everything from Grixis Control to UR Possibility Storm
Bloom Titan - banned, but usable in a weaker version: Amulet Titan
Affinity - I actually completed this around 1-2 years ago. I didn't do it initially because it was a deck style that I don't prefer, but then later on I realized how close I was to the complete deck list, so I bought the remaining cards.
That's basically 3 of 6 decks that have been or had been viable for years that are all right now. I'm not diminishing the losses of those who went all-in on 1-3 of those decks and had bans wreck their deck(s). A ban is always a failure on Wizard's part. I'm just pointing out that only some of the "original decks" are still around.
Legacy - Sneak Show, BR Reanimator, Miracles, UW Stoneblade
Premodern - Trix, RecSur, Enchantress, Reanimator, Elves https://www.facebook.com/groups/PremodernUSA/ Modern - Neobrand, Hogaak Vine, Elves
Standard - Mono Red (6-2 and 5-3 in 2 McQ)
Draft - (I wish I had more time for limited...)
Commander - Norin the Wary, Grimgrin, Adun Oakenshield (taking forever to build) (dead format for me)
I can understand the argument that tiers based on popularity is useless if all you care about is measuring deck strength, but I don't agree with it. Popularity very often reflects the strength that decks have, as people naturally gravitate towards the winningest decks. But tiers also serve another purpose, and that's to let players know what they can expect to face at a given event. Prevalance-based tiers are ideal for this purpose.
the card itself really isn't worth the card board its printed on...does it serve a purpose? Sure its a blue 1 drop that might not be a 1/1. I would not put it in a list and expect to win a PTQ or GP though.
Isn't that still just a tiering system?
It removes the popularity aspect of the tier systems. It would be straight percentages.
I would suggest it be top 16 showings. top 8s is too narrow, and top 32 would make the list huge.
In the end if we knew the percentages were based on 3 events (SCG, PPTQs/PT, and GPQ/GP) and the cut off (top 8,16, or 32). Then when someone reads the list of proven decks, there is no other factors to sway opinion other then straight numbers.
The problem with most tiering systems is popularity comes into play. Unless we can get the numbers of the decks starting in an event, and we can compare those starting numbers to what we see day 2 and in top spots. Popularity is a baseless skewed number.
Example to explain the above..
If we know in a field of 200 players 50 decks are Affinity, and 25% of the top 16 is Affinity, we can say Affinity was represented properly.
Now if Affinity was only 10 of those 200 starting decks and Affinity was still 25% of the top spots, we can say the deck over performed.
Now if 100 decks of the starting 200 are Affinity and only 25% fall to top spots the deck under preformed, talking straight numbers.
But that is a lot more leg work and crunching of numbers if they can be found. Which I believe we cant.
Popularity comes into play with your system too. More popular decks will top more often, and will be represented higher regardless of their actual potential.
Not true, you are disregarding skill level of the pilots. In reality on any given weekend for the SCG events that have 200+ players in them, maybe 50 have a realistic shot at winning the event. if they pay out to 64, that leaves 14 spots for the grinders or 'casual' players to squeeze into the money spots.
If you wanted to make a true tier system, you would only look at decks people that get paid to play, play. you could ignore the rest of the field. This would remove popularity, the skill factor, the money factor and any other negative about current tier systems. We would know the decks are playable and in the end thats all anyone really cares about since we are basing all this off the highest level events. Which I feel is wrong in itself, but we dont get the complete data from every where else so .. its doing the best we can do and removing as much of the negative from the process.
I am just going to say I didnt agree with the old system any more then I agree with the new system. Its a flawed system with incomplete data.
By defining the data, only using 'X' amount of events, and only using percentages of tops (agreed cut off). Its a much easier, and clearer picture of the performance of the decks in question.
Until we can get information such as how many of each deck type was entered to how many topped from all events a tier system is going to be skewed.
I would also say there are more T1 decks then is being shown. Simply because the events the data is being taken from is some of the highest level we have for the game. So it would mean there is better pilots, hence better results with less optimal decks. We all know Modern is a format of comfort. Play the deck you feel more comfortable playing.
Second: We need to try to establish what tiers are. You can see over the last several pages some discussion around this.
Are they representation?
Are they power?
Are they performance and results?
Or is it a combination? I dont think we can base it on power, as thats difficult to judge. Nor can we base it on representation, because not everyone is going to flip to Jund, from some Jank, just because they are told Jund is tier 1.
The most fair metric, will be results.
If we had a way to track data for Top 8's and 5-0's, that would be ideal to see general trends, but even those numbers as HolyDiva outlined could EASILY be skewed by a few grinders in the case of MTGO.
People need to stop being completely infatuated with tiers, and I know I do it too, but decks that are not Tier 1 can have play, even if not popular.
Spirits
This. I play goblins in legacy, and in terms of MWP it's very very good. But because representation is low, it rarely breaks T8. Low representation decks underperform in results based metrics.
There is no good dataset for MWPs. Data either comes from single players (small N, player skill can matter more than deck performance, high matchup variance, etc.) or from single events (similar issues as above, notably with sample and match variance). This would have been feasible in the days of MTGO replays and scraping software, but it isn't now.
I really don't get why people are trying to reinvent the wheel given Modern's current information sources. The prevalence-based tiering system was largely accurate and largely liked, and that has been true for years. If we had access to stuff like MWPs or matchup-specific data then great, I'm sure many would happily use it. Same if we had Day 1 to Day 2 to Top 8 conversion rates. But we don't have any of that information, especially not on a consistent basis. Because of that, prevalence-based metrics are the next best thing. Those metrics also aren't inaccurate. Even when we did get conversion rate data and MWP information, it rarely changed the top-tier picture too much, and few sleepers like Amulet Bloom and Lantern were missed.
It's a bummer that Modern Nexus is behind with consistent metagame updates. Mainphase's updates are frequent but they have small Ns and questionable (maybe even arbitrary) cutoffs. Hopefully the format can get the best of both those worlds soon, or someone steps up to fill in those gaps.
Realistically, I know some people want tiering to be based off skill and skill alone, but we haven't had access to win rates in forever, and most tiering systems, if not all of them, still have some fraction of popularity bais too. A good example is pokemon. Pokemon tiering does go off win rates... but the exact tiering they use in smogon still has people voting for the pokemon to be promoted or demoted based on wins they need to supply test evidence with. No one really loves chancy but wont deny how good of a wall it is, but gengar is currently rank A, even though its been nerfed to hell because people still like gengar.
Our tiering still represents popularity AND wins:
Zoo is a great deck, but the decks it beats, delver and control arent popular right now, thus it is a bad deck... Popularity and success are intermixed there.
Sure it could be. But we run into the same problem as last time. People dont want to spend time to do it and give it up. Im not trying to reinvent my data source every half year
I keep nexus open on my tabs. I refresh it once a day to see if it has updated. Were 3 days late at this point, and everytime its late I have to change my "when does the forum refresh" date. It sucks. I wish it was better. I like the site.
I think that is part of the confusion, is the new tier system is alot more lax.
I feel like the old tiering system was as follows (feel free to disagree)
Tier 1- A deck you WILL see at the top tables. As well as day 2 representation 99% of the time.
Tier 2- A deck you have a fair chance at seeing in a day 2 GP and occasional top table appearances.
Proven- Decks that either were good in the past and people still might run, or decks that are rising in popularity. Its unlikely you will see them, but you should know what they are to be prepared
I feel like the new tier system is missing the "top tables" element. But due to the lack of data, there may not be anyway around that.
Twitter: twitter.com/axmanonline
Stream: twitch.tv/axman
Current Decks
Modern: Affinity
Standard: BW Control
Legacy: Death and Taxes :symw::symr:
Vintage: NA
In this sense, WotC could be trying to avoid data that would shape the meta.
It's also possible that if that data does exist, it would likely force results so impressive that WotC would be caught in a cycle of endless bans, because the format is actually rather poorly balanced, and only luck and currently lower metagame shares keeps decks with higher w/l ratios from exploding out of control in metagame shares due to more and more spikes picking them up, potentially leading to WotC forced to do chains of bans that would be highly unpopular every time a new deck reaches the top, since the format doesn't have the tools to self-correct against many of the secret top decks WotC has data indicating.
UR ....... WUBR ........... WB ............. RGW ........ UBR ....... WUB .... BGU
Spells / Blink & Combo / Token Grind / Dino Tribal / Draw Cards / Zombies / Reanimate
That's basically it. Watch any Modern deck tech and you'll have 10 different players, on 10 different decks, all saying their deck is the best positioned deck in the meta. WotC loves that. They want people to think they've figured it out, and for them to be wrong. It keeps a market alive as people switch decks and make guesses. It also means things don't stagnate into solid data where cards are overpowered based purely on recorded results. With a less clear meta, a card has to really overperform to be ban worthy.
To me, it seems everyone has their own needs for the tierings, and this requires both different datasets and cutoffs. USA vs non-USA regional metagames, online vs paper metagames, FNM vs GP metagames, it goes on and on.
The solution isn't so simple. Nexus is comprehensive but too far behind, Goldfish only tracks last 30 days and is heavily skewed towards online and USA events, MainPhase just tracks Goldfish weekly, MtgTop8 has some weird classification issues, I personally just track online stuff.
Point is, nothing is perfect, but you can always choose to use whatever fits best your own needs. If this specific forum chooses to use X dataset with Y cutoffs, don't make it such a big deal. If it doesn't fit your needs, you are free to look elsewhere or do it yourself. And that's fine:)
My only problem with the new tiers here was the lack of the explanation, not the changes themselves. And I think I wasn't the only one, as many here expressed they couldn't make sense of it. Whatever you finally do Lantern, just make it clear to everyone (dataset and cutoffs) so everyone knows what's happening.
Random thought: can't we just have all primers in one single section? As long as people can easily search for the deck thread they want to post in, we should be fine. Is it that necessary?
Frontier: UBR Grixis Control | BRG Jund Delirium
Modern - Burn
EDH - Neheb the Eternal
The idea has been around for a while. There are a new form of deck and primer thingy mtgsal is looking to implement in the future that might make that the best solution at some point, and if the system changes, well likely adpot that model.
I'm not trying to be vague on that one. I literally dont know the details and its been in the works back burner for a while.
Modern - Burn
EDH - Neheb the Eternal
I know some dont want to hear it or even believe it, but this is exactly what the game of Magic use to be. An established pecking order of decks, and players brewing to beat that pecking order, keeping the pecking order always in a state of change. Now the player base has gotten lazy and only plays proven decks and sits at events playing mirror matches complaining about this or that. That is why I dont believe in the Tier system. Never have. Its nice to know what you may see, and what is the flavor of the week so you can tweak your deck to be competitive, but nothing more, nothing less then that.
It's bad to want your deck to be good when you drop a few hundred dollars building it? Even for the best deck builders in the world, only 10% of their ideas are any good. Some random person is going to have a much lower rate than that (if they're even capable of judging their deck as good/bad). Knowing you have a good deck lets you focus on playing the game.
That's only true of some deck types though. I love to build my own lists (for example, I've got a pretty good Mardu DS right now), but I still need a starting point for them. That could be building off of an existing archetype to try and do it better, or it could simply be in building a meta deck to beat what other people are doing. In either case, I feel as a deck builder that having all of this information access only helps me rather than hinders me. It hasn't made me lazy, it's given me a better starting point for anything I choose to build. And it has the advantage that when I want something safe, I can filter my options to those decks and make a choice.
Tron - still viable and strong
Jund - still viable and a bit less strong right now
Pod - banned, but some pieces used in Abzan Company
Twin - banned, but the other cards are staples that I've used in everything from Grixis Control to UR Possibility Storm
Bloom Titan - banned, but usable in a weaker version: Amulet Titan
Affinity - I actually completed this around 1-2 years ago. I didn't do it initially because it was a deck style that I don't prefer, but then later on I realized how close I was to the complete deck list, so I bought the remaining cards.
That's basically 3 of 6 decks that have been or had been viable for years that are all right now. I'm not diminishing the losses of those who went all-in on 1-3 of those decks and had bans wreck their deck(s). A ban is always a failure on Wizard's part. I'm just pointing out that only some of the "original decks" are still around.
Premodern - Trix, RecSur, Enchantress, Reanimator, Elves https://www.facebook.com/groups/PremodernUSA/
Modern - Neobrand, Hogaak Vine, Elves
Standard - Mono Red (6-2 and 5-3 in 2 McQ)
Draft - (I wish I had more time for limited...)
Commander -
Norin the Wary, Grimgrin, Adun Oakenshield (taking forever to build)(dead format for me)Counter-Cat
Colorless Eldrazi Stompy