Another Look at 2v2 Super Titan

GreatGonzales · (This post was last modified: 03-21-2013 04:41 AM by GreatGonzales.)

(03-21-2013 03:42 AM)CombatEX Wrote: I respect that, but in that case this discussion does not concern me. What I would like to discuss is separating 2v2 random from arranged, but until that is done I feel any conversation regarding 2v2 arranged distribution is overshadowed by the large number of random players skewing the results.

I believe Eijolend and I have encountered random players in only 1 game and I imagine this is similar for STs in general. As such it doesn't really have an impact on us as I imagine is the case with your ST team too, but as a whole 2v2 random teams have a huge impact on 2v2 arranged distributions since 2v2 arranged teams play 2v2 randoms whether you want to ignore it or not.

I see, you are interested in a different sort of conversation. Anyway, I agree that 2v2 random teams impact 2v2 arranged league promotion (merely because they share the same player pool in league play), however I believe that 2v2 random teams do not impact 2v2 arranged league composition (because their leagues are mutually exclusive). This thread was specifically concerning league promotion thresholds, which is more to do with the composition of leagues.

Sidebar, I agree that 2v2 random and 2v2 arranged shouldn't be playing together...kind of like apples and oranges. Perhaps there aren't enough of one population or the other (or both) to sustain swift matchmaking. That might be why OML designed the system this way.

Actually CombatEX, come to think of it, the shared player pool between 2v2 random and 2v2 arranged may help explain why 2v2 arranged seems so "top heavy" compared to the 1v1 population. That is, about 23% of arranged players are either Masters or ST, compared to just 5% in 1v1.

CombatEX · 03-21-2013, 04:49 AM

(03-21-2013 04:35 AM)GreatGonzales Wrote: Perhaps there aren't enough of one population or the other (or both) to sustain swift matchmaking. That might be why OML designed the system this way.

I speculated that may be the case in Pinga Pinga's (sp?) thread where s/he complained about 2v2 matchmaking. Hopefully this isn't the case though and we can eventually separate them so Apples don't have to play Oranges as you put it =)

GreatGonzales Wrote:Actually CombatEX, come to think of is, the shared player pool between 2v2 random and 2v2 arranged may help explain why 2v2 arranged seems so "top heavy" compared to the 1v1 population. That is, about 23% of arranged players are either Masters or ST, compared to just 5% in 1v1.

I agree. As you said previously, as an arranged team you have an advantage over someone playing in a random team and so it should be easier to for arranged teams to rise to higher leagues than a random team player. This will influence the arranged distribution to be more top heavy than 1v1 due to the 2v2 random players which is the underlying reason I suggested we consider 2v2 as a whole. Perhaps what we should be requesting to OML is for them to make the distributions independent. I suspect that the source of your problem with 2v2 arranged distributions is because 2v2 distributions are currently considered as a whole by OML. If the distributions were separated this could resolve the issue (supposing I am correct in my assumptions).

**Harti** · 03-21-2013, 05:05 AM

What would the "perfect distribution" among tiers look like in your opinion?

It's a tough thing to design - I bet it took Alex and Adam quite a while to figure a sensible way for these league systems out. That said, there can always be unforeseen flaws as to smaller player base or other variables to behave differently from what you would expect.

A short summary of your thoughts might help to understand why the current system is broke and how you would fix it.

GreatGonzales · 03-21-2013, 05:13 AM

(03-21-2013 05:05 AM)Harti Wrote: What would the "perfect distribution" among tiers look like in your opinion?

It's a tough thing to design - I bet it took Alex and Adam quite a while to figure a sensible way for these league systems out. That said, there can always be unforeseen flaws as to smaller player base or other variables to behave differently from what you would expect.

A short summary of your thoughts might help to understand why the current system is broke and how you would fix it.

I think that the 1v1 ST league feels right, and we can use it as a model. That being said, my findings by looking at the numbers are inconclusive. And there's the added complexity of the 2v2 random league players influencing the league promotion...without knowing exactly how that works, it's hard to make a specific recommendation. However, I do think that the current promotion threshold for 2v2 ST is probably too loose, merely by looking at the skill level of some of the teams who have made it in so far (and of the teams who may make it in soon, like my secondary team!).

CombatEX · (This post was last modified: 03-21-2013 05:39 AM by CombatEX.)

(03-21-2013 05:13 AM)GreatGonzales Wrote: I think that the 1v1 ST league feels right, and we can use it as a model. That being said, my findings by looking at the numbers are inconclusive. And there's the added complexity of the 2v2 random league players influencing the league promotion...without knowing exactly how that works, it's hard to make a specific recommendation. However, I do think that the current promotion threshold for 2v2 ST is probably too loose, merely by looking at the skill level of some of the teams who have made it in so far (and of the teams who may make it in soon, like my secondary team!).

I think 1v1 ST league feels right too. Though I think Gifted, Clever, and Fluffy are off for reasons discussed in IMPERIAL's thread a while back. Looking at that again I'm not so sure about the distribution IMPERIAL suggested but I think it is a step in the right direction. Really I think Fluffy is too demeaning and should be even smaller than the suggested 20% (currently it's at 40% though which is far worse). Recently Blizzard decreased Bronze from 20%->8% for reasons discussed here. Though it may not be directly comparable to Outwitters I wouldn't be surprised if the same is essentially true here. This is getting off track though so I'll stop for now.

Pharmafan · 03-21-2013, 08:41 AM

I completely agree with GG. With my arranged team with HolySab we were long-time 2v2 masters and always aimed to get good enough to perform the necessary winning streak to get promoted to ST....one day....right before they loosened the promotion threshold we had again a good streak but never knew if it will be sufficient....so we always weren't good enough to get over....and then someone gave us and many others a hand.....and it feels a bit like "come on in.....we lowered the barriers for you.....so now you are good enough"..... Bad taste....

ST means best of the best and if we would have made it regarding the barriers form former days I would have been proud telling that we are part of the best of the best. Don't get me wrong. I really like to be a ST in 2v2 but perhaps you know what I mean especially when I commit that I KNOW, that we make mistakes and I'm sure that there are better teams out there. So I'm not really sure if we deserve the honor of being in.

BUT I also think, that it is absolutely not easy to find a correct threshold for that. The threshold in 1v1 feels right. And the threshold for 2v2 ST felt a bit high. I was OK with that (loving high thresholds) and had no wishes for a change. Others had and it took place. I never before took part in a discussion about that so I have to respect the new threshold and our promotion which feels a bit too early. Hope that our new clothes will fit one day :-)

amoffett11 · 03-21-2013, 09:45 AM

Your assumption that the percentage of 1v1 ST's should be equal to the number of 2v2 ST's is only correct if you assume that the distribution of skill in the two formats is equal. Due to a number of factors that make 2v2 a very different game than 1v1, this is not the case.

The reason there's 40 STs at 2v2 is because with the exception of numbers 1 and 2, the rest of you at the top are so close to each other that is impossible to find a threshold (right now) where you will get the percentage that you're looking for.

GreatGonzales · 03-21-2013, 10:24 AM

(03-21-2013 09:45 AM)amoffett Wrote: Your assumption that the percentage of 1v1 ST's should be equal to the number of 2v2 ST's is only correct if you assume that the distribution of skill in the two formats is equal. Due to a number of factors that make 2v2 a very different game than 1v1, this is not the case.

The reason there's 40 STs at 2v2 is because with the exception of numbers 1 and 2, the rest of you at the top are so close to each other that is impossible to find a threshold (right now) where you will get the percentage that you're looking for.

I assume nothing here; I make that clear in the OP. My contention we can use 1v1 as a model for 2v2 is only because that league has been successful and it feels like the right level of exclusivity. I recognize the differences between the two.

I don't understand why you believe it's "impossible" to find the right threshold for ST promotion. Why should it be impossible? Difficult, sure, but OML at least knows that the first threshold was too high, this one may be too low, so the proper threshold is probably somewhere between the two, right? Also, your statement that the top 40 2v2 teams are very close together in skill level is largely erroneous.

amoffett11 · 03-21-2013, 10:55 AM

(03-21-2013 10:24 AM)GreatGonzales Wrote:
(03-21-2013 09:45 AM)amoffett Wrote: Your assumption that the percentage of 1v1 ST's should be equal to the number of 2v2 ST's is only correct if you assume that the distribution of skill in the two formats is equal. Due to a number of factors that make 2v2 a very different game than 1v1, this is not the case.

The reason there's 40 STs at 2v2 is because with the exception of numbers 1 and 2, the rest of you at the top are so close to each other that is impossible to find a threshold (right now) where you will get the percentage that you're looking for.

I assume nothing here; I make that clear in the OP. My contention we can use 1v1 as a model for 2v2 is only because that league has been successful and it feels like the right level of exclusivity. I recognize the differences between the two.

I don't understand why you believe it's "impossible" to find the right threshold for ST promotion. Why should it be impossible? Difficult, sure, but OML at least knows that the first threshold was too high, this one may be too low, so the proper threshold is probably somewhere between the two, right? Also, your statement that the top 40 2v2 teams are very close together in skill level is largely erroneous.

Here's our problem: you're not reading what I'm saying.

What I said: it's impossible (right now) to find a threshold where the percentage of 1v1 STs is the same as the percentage of 2v2 STs.

What you apparently read: it's impossible to find the right threshold.

What I'm saying is that the distribution of skill is such that by lowering the threshold a little, like they did, they lowered it below at least 30 teams, this is because these 30 teams skill rating is very close together. There is no in between, at this point in time.

I recognize that you believe, based on nothing but the games you've played against various teams, that you and about 15 other teams stand above the rest. However, with the evidence we have given the two league distributions (and this, along with the top 200 2v2 lists is the only evidence we have) supports the theory that you're all closer together than you think. Just because someone is #3 and someone else is #69 doesn't mean that their skill ratings are not actually close together.

And you do make an assumption: you assume that two skill distributions in two different formats should be equal. There is nothing to suggest that this would be the case. It's apples and oranges.

GreatGonzales · (This post was last modified: 03-21-2013 11:33 AM by GreatGonzales.)

Man come on, you're picking this fight with me again? Alllllrighty then....

Quote:Here's our problem: you're not reading what I'm saying.

What I said: it's impossible (right now) to find a threshold where the percentage of 1v1 STs is the same as the percentage of 2v2 STs.

What you apparently read: it's impossible to find the right threshold.

To be fair, that's not an exact quote of what you said, but I see what you meant now. I do not intend to knock down a straw man. Sure, it's going to be an inexact process due to the inherent differences between the two populations; my point is, however, that OML can re-draw the boundary to improve the threshold (assuming current scenario is not ideal) knowing nothing more than what they know right now. But could they precisely locate the same threshold utilized by 1v1 ST? No, probably not.

Quote:What I'm saying is that the distribution of skill is such that by lowering the threshold a little, like they did, they lowered it below at least 30 teams, this is because these 30 teams skill rating is very close together. There is no in between, at this point in time.

You're apparently fond of claiming that I make assumptions....so if I may be so bold, you are making an important assumption here. Namely, you presume to know, or at least have some idea about, the degree that the promotion threshold for 2v2 ST was lowered. It could be that Alex gave it a very minor tweak, which would support your contention that the top 40 teams are all very close together in terms of skill level. But it could also be that Alex gave it a large nudge, or something in between. We don't know; all we know is that once the threshold was changed, the number of ST teams skyrocketed from 2 to 33 (and probably will land somewhere around 40 if current trend holds). These facts are also potentially consistent with a top 40 2v2 teams of fairly diverse skill levels.

Quote:I recognize that you believe, based on nothing but the games you've played against various teams, that you and about 15 other teams stand above the rest. However, with the evidence we have given the two league distributions (and this, along with the top 200 2v2 lists is the only evidence we have) supports the theory that you're all closer together than you think. Just because someone is #3 and someone else is #69 doesn't mean that their skill ratings are not actually close together.

I believe that there is a fairly severe drop-off in skill after the top 10 or so teams, yes, and it's true that much of that belief is based on my own personal experience. But I think Rawkhawk and I are probably one of the top 10 teams, so I feel well-qualified to make that call.

But anyway, can you please clarify what you mean when you say that the two league distributions and the top 200 lists support the theory that the top teams are closer in skill than I think? I don't understand. The top 200 lists do not tell us the distance between player skill, only their rank. I'm also not sure what the league distributions, by which I take you to mean the breakdown of the player population by league, can tell us about the relative skill of players within a particular league. You're right that just because a team is ranked 3rd overall does not necessarily mean that there is a significant difference in skill between them and the team ranked 69th, but it does suggest as much.

Quote:And you do make an assumption: you assume that two skill distributions in two different formats should be equal. There is nothing to suggest that this would be the case. It's apples and oranges.

No, I don't believe that the skill distributions between 1v1 and 2v2 should be equal, nor do I think they in fact are equal. I'm merely suggesting that we use the 1v1 ST league as a model for 2v2 since it feels right. Absent another model, it's the best one we have.