SC2 Ladder Analysis: Division Tiers - Page 3
Forum Index > SC2 General |
god deezy yo
49 Posts
| ||
Shadowed
United States679 Posts
![]() | ||
strongandbig
United States4858 Posts
Additionally, if I've ready ExcaliburZ's MMR posts correctly, the game promotes your league as soon as it's sure that your MMR is high enough with low enough uncertainty - it doesn't wait for you to totally even out to 50-50, just for you to be a certain number of sigmas over the threshold. If that's true, then without movement between divisions in Diamond there couldn't be collection of more skilled players in some divisions than others. It's possible that divisions can get "promoted," but I think it's unlikely since it's almost certain that all of the division weighting happens when you lose points at promotion; given that the gaps are always multiples of 63, there's really no way to abruptly change players' point totals during a "season" when their league gets promoted. | ||
Catreina
United States304 Posts
This means that I am really a 1400 point diamond, and that Minigun is really 2700 not almost 3000 As always, awesome info ExcaliburZ - When I noticed the points on the top 200 I initially thought MMR, and found this post almost instantly afterwards. Always good to know more about the game we all love. | ||
Wargizmo
Australia1237 Posts
http://sc2ranks.com/div/us/diamond/1/players/393 http://sc2ranks.com/div/eu/diamond/1/players/245 http://sc2ranks.com/div/kr/diamond/1/players/80 http://sc2ranks.com/div/sea/diamond/1/players/0 you can see that in every single region there are 5 or 6 divisions that haven't filled up to 100* players yet. It seems likely that these divisions each have a different grading and that it's assigning players to a particular non-full division based on their MMR. If it just filled up divisions arbitrarily you'd expect there to be just one division with less than 100 players and then when that was full it would just create a new division and start chucking players in there. From this you can extrapolate that the top ranked divisions are likely to have the best players in, which is very consistent with what you can see from the top 200 list, I would even theorise that of the 500 or so diamond divisions in North America, that there is a fairly even spread of each 'class' but because mostly S-Class ones are represented in the top 200 (since they have the best players) the original post by Excalibur Z includes mostly S-Class divisions. * I'm assuming that the many divisions with 98/99 people were full but had players getting demoted or banned without the slot having been refilled yet. | ||
![]()
Excalibur_Z
United States12224 Posts
On November 18 2010 10:06 Wargizmo wrote: Take a look at this http://sc2ranks.com/div/us/diamond/1/players/393 http://sc2ranks.com/div/eu/diamond/1/players/245 http://sc2ranks.com/div/kr/diamond/1/players/80 http://sc2ranks.com/div/sea/diamond/1/players/0 you can see that in every single region there are 5 or 6 divisions that haven't filled up to 100* players yet. It seems likely that these divisions each have a different grading and that it's assigning players to a particular non-full division based on their MMR. If it just filled up divisions arbitrarily you'd expect there to be just one division with less than 100 players and then when that was full it would just create a new division and start chucking players in there. From this you can extrapolate that the top ranked divisions are likely to have the best players in, which is very consistent with what you can see from the top 200 list, I would even theorise that of the 500 or so diamond divisions in North America, that there is a fairly even spread of each 'class' but because mostly S-Class ones are represented in the top 200 (since they have the best players) the original post by Excalibur Z includes mostly S-Class divisions. * I'm assuming that the many divisions with 98/99 people were full but had players getting demoted or banned without the slot having been refilled yet. That is interesting. I wonder how many tiers exist per league, then? According to the NA Diamond list, there are only 5 that are currently not full or close to full. I think you're probably right that each of these represents a different tier, but out of how many, I wonder? 10? 20? I've noticed that divisions also tend to fill up fairly quickly (within a day or two), so perhaps promotions are put on hold for some players until a new division tier can be opened up if one that isn't full doesn't already exist? I guess it's also possible that the activity level is just that high, but I find that harder to believe. | ||
Pokebunny
United States10654 Posts
| ||
Wargizmo
Australia1237 Posts
On November 18 2010 10:31 Excalibur_Z wrote: That is interesting. I wonder how many tiers exist per league, then? According to the NA Diamond list, there are only 5 that are currently not full or close to full. I think you're probably right that each of these represents a different tier, but out of how many, I wonder? 10? 20? I've noticed that divisions also tend to fill up fairly quickly (within a day or two), so perhaps promotions are put on hold for some players until a new division tier can be opened up if one that isn't full doesn't already exist? I guess it's also possible that the activity level is just that high, but I find that harder to believe. I would hope there aren't too many tiers that are lower than E-Class, once you start inflating people's points by 400 or 500 they become even more meaningless than they already are. Or are you suggesting that perhaps there are more classes inbetween?, the 63 point thing doesn't seem like it's a coincidence though. Having a division fill up in two days doesn't seem that fast, when you consider there are 86,000 platinum players in NA (according to sc2 ranks), I don't think it's too unreasonable that a few hundred of those would be getting promoted every few days. In fact doing the math if 6 divisions (one for each tier) filled up every two days since Sc2's release you'd have 342 divisions, which is actually less than the 500 we currently have... so if you assume there were a lot of divisions created very fast initially after release (like 150 or so) that seems like about the exact speed you'd be seeing divisions fill up, once every couple of days per tier. | ||
Minigun
619 Posts
| ||
starcraft2leverage
United States207 Posts
On November 18 2010 11:23 Minigun wrote: One thing I don't understand is, can't you play people from other divisions? aka, wouldn't they all be similar? 0-0 Everyone gains and loses points the same from wins and losses, the difference is the starting point when you join diamond. You got an extra 300 or so points when you joined, others got 63 or none or whatevs. All else equal then, someone in a different division with the same record as you beating the same exact opponents at the same time will have a different amount of points given the difference in the starting points. | ||
![]()
Excalibur_Z
United States12224 Posts
On November 18 2010 10:53 Wargizmo wrote: I would hope there aren't too many tiers that are lower than E-Class, once you start inflating people's points by 400 or 500 they become even more meaningless than they already are. Or are you suggesting that perhaps there are more classes inbetween?, the 63 point thing doesn't seem like it's a coincidence though. Having a division fill up in two days doesn't seem that fast, when you consider there are 86,000 platinum players in NA (according to sc2 ranks), I don't think it's too unreasonable that a few hundred of those would be getting promoted every few days. In fact doing the math if 6 divisions (one for each tier) filled up every two days since Sc2's release you'd have 342 divisions, which is actually less than the 500 we currently have... so if you assume there were a lot of divisions created very fast initially after release (like 150 or so) that seems like about the exact speed you'd be seeing divisions fill up, once every couple of days per tier. Remember that leagues are split 20% each among active players, but the required activity period is not known. If we said 7 days, that's only 20,000 Plat players that are active. Maybe those people (and the 100k other active players from the past week) get promoted with a frequency enough to fill a division tier every 2 days, but that seems like a lot of league hopping. I guess it's not impossible, though. I don't think there are tiers in between what I found, but it wouldn't surprise me if there were additional ones beyond those. I don't think Blizzard necessarily needs to be concerned with the value of points because a) they're constantly inflating anyway and b) the larger the division modifier, the lower the skill of the division, which only serves to produce a smoother gradient for the entire ladder. If you had a tier that had a division modifier of +630, but the players therein were just that bad such that they were averaging 1300 (current bonus pool max is over 1500), then on paper they would still be competitive with many players even though in reality they're far behind. Side note: I wonder if division tiers interact with the bonus pool in any way? For example, maybe a new tier is introduced every time the bonus pool max increases by 63? Baseless speculation there, but I wonder. | ||
vanick
United States53 Posts
On November 18 2010 05:58 Lightspeed wrote: I'm not sure I understand correctly. If you were promoted to Scout Kilo you would end up with 315 more ladderpoints than if you were promoted to Medic Mu? Where's the sense in this, what would be the reasoning? The intent behind the design of the system is you only focus on your own division. Your ranking is correct within your division but you were never intended to find your place within a league because for most players that could be discouraging. Keeping things limited to 100 makes progressing in a ladder much easier to quantify. At the top end obviously people care a lot more about inter-division rankings. The Masters and Grand Masters leagues will address a lot of this desire I think, but we're kind of stuck until then. | ||
StrifeCro
United States69 Posts
On November 18 2010 11:38 starcraft2leverage wrote: Everyone gains and loses points the same from wins and losses, the difference is the starting point when you join diamond. You got an extra 300 or so points when you joined, others got 63 or none or whatevs. All else equal then, someone in a different division with the same record as you beating the same exact opponents at the same time will have a different amount of points given the difference in the starting points. you might be right on the starting points when u get promoted to diamond but this is by far the smallest difference. people have played so many games some ppl up to 1000-2000 so this is just about meaningless. by far the biggest difference is you get more points per win and lose less points per lose versus the same people. maybe you are 3000 points in your division but you might gain/lose points like a 2500 player from a "s-class division", which is why you were able to get 3000 points in this first place. | ||
Minigun
619 Posts
On November 18 2010 13:13 StrifeCro wrote: you might be right on the starting points when u get promoted to diamond but this is by far the smallest difference. people have played so many games some ppl up to 1000-2000 so this is just about meaningless. by far the biggest difference is you get more points per win and lose less points per lose versus the same people. maybe you are 3000 points in your division but you might gain/lose points like a 2500 player from a "s-class division", which is why you were able to get 3000 points in this first place. I see. Well I can't wait for grand masters league then, to put all this to rest ^_^ . | ||
StrifeCro
United States69 Posts
On November 18 2010 13:25 Minigun wrote: I see. Well I can't wait for grand masters league then, to put all this to rest ^_^ . i just made those numbers up btw and wasn't really referring to you or anything just as an example, but i believe excalibur's numbers here are pretty much spot on so scout kilo is only -315 anyways so more like 3000 same as 2685 in an s-class division. | ||
Minigun
619 Posts
On November 18 2010 13:28 StrifeCro wrote: i just made those numbers up btw and wasn't really referring to you or anything just as an example, but i believe excalibur's numbers here are pretty much spot on so scout kilo is only -315 anyways so more like 3000 same as 2685 in an s-class division. Oh sorry I didn't mean that the way it sounded. Yeah I know what you are saying. I know I am not near #1 player because my points are that high. That was my fault ![]() Thanks for explaining it. | ||
Wfat
Australia108 Posts
I started on SEA and was terrible for the first ~50 ladder games. My win percentage was very good >60% on SEA up until recently when I've been having more difficulty beyond 2000 points. I think that this can be attributed to playing bad players early on such that my skill improved much faster than my MMR. As I got better I began playing on my NA account (as SEA players got free access at some stage). At this stage I was significantly better than when I had started on SEA and so I began playing good players much more quickly, so my MMR increased rapidly and then flat-lined, at a rate much faster than my skill. My win percentage on NA is thus much worse ~50%, and I have far less points than I do on my SEA account (1543), even though both accounts have relatively low amounts of bonus points. I started playing about 1 month or so after release, so was placed in brand new divisions on both SEA and NA regions. Summary: So in SEA I benefitted from having my skill improve much faster than my MMR >60% win ratio, whereas in NA my skill was relatively constant so my win rate has always been approximately 50%. Thus, how can points accurately represent skill when it does not account for differential increases in player skill? My accounts for reference: SEA account - Studyharder NA account - Wfat | ||
StrifeCro
United States69 Posts
On November 18 2010 14:05 Wfat wrote: I don't understand how a ranking system just based on points can provide an accurate measure of skill. Take myself for example: I started on SEA and was terrible for the first ~50 ladder games. My win percentage was very good >60% on SEA up until recently when I've been having more difficulty beyond 2000 points. I think that this can be attributed to playing bad players early on such that my skill improved much faster than my MMR. As I got better I began playing on my NA account (as SEA players got free access at some stage). At this stage I was significantly better than when I had started on SEA and so I began playing good players much more quickly, so my MMR increased rapidly and then flat-lined, at a rate much faster than my skill. My win percentage on NA is thus much worse ~50%, and I have far less points than I do on my SEA account (1543), even though both accounts have relatively low amounts of bonus points. I started playing about 1 month or so after release, so was placed in brand new divisions on both SEA and NA regions. Summary: So in SEA I benefitted from having my skill improve much faster than my MMR >60% win ratio, whereas in NA my skill was relatively constant so my win rate has always been approximately 50%. Thus, how can points accurately represent skill when it does not account for differential increases in player skill? My accounts for reference: SEA account - Studyharder NA account - Wfat points can very accurately reflect skill level, at least if your playing enough and trying hard. its just because sea server is much weaker than NA server, you cant compare points across different servers. | ||
Amber[LighT]
United States5078 Posts
I'll break back into top 10 soon enough! So now is this (below) technically correct? I'm trying to understand exactly what this means... On November 18 2010 08:25 Wargizmo wrote: Not really, because when people in an E-class division are placed initially they get given 315 free bonus points, so their rating is always going to be 315 higher than what it should be no matter what. That's why some people lose a ton of points when they get promoted and others don't lose much at all. If Dayvie for example had been placed in Medic Mu and played exactly the same games as he has done already, then he would have exactly 315 points less now than what he currently has but would be in exactly the same position on the top 200 list (because of the division). Also I'm assuming this (below) can't be answered because we don't have the ability to interpret all of the data? On November 18 2010 08:27 Shadowed wrote: The part I'm wondering is, how do they define ranks? As far as I know, divisions are created on the fly based on demand, so the assignment of rank makes little sense. And are the modifiers arbitrary? How was 63 determined? Sry it's late and I might have missed this ![]() I'll pose this question as well... In division K-Xray the bolded players were in the top 200, and the non-bolded players did not appear.. Why and how did this happen? 1 Katari (2,856) 2 MasterAsia (2,682) 3 EGMachine (2,513) 4 Bamboocha (2,419) 5 FreshShock (2,276) 6 RaNGirLxD (2,267) | ||
gdTyrael
49 Posts
I mean, supposing your really "bad" and when you got into Diamond you were placed at a junk division, but after some time your MMR got really high and you became a pro player, and you just stand out from everyone in your division, it doesnt make too much sense for you to stay there.. and to the fact that blizzard seems to care about divisions. | ||
| ||