EPTSC2 Ladder Analysis: Division Tiers - Page 38
| Forum Index > SC2 General |
|
Valenti
United States16 Posts
| ||
|
michaelhasanalias
Korea (South)1231 Posts
| ||
|
michaelhasanalias
Korea (South)1231 Posts
| ||
|
random user
85 Posts
Gold: Promoted recently: + Show Spoiler + http://us.battle.net/sc2/en/profile/1687175/1/Weend/ ~12 hr ago snapshot in gold: 2,890 pts 890 - 874 w/l http://us.battle.net/sc2/en/profile/435246/1/Evertras/ ~12 hr ago snapshot in gold: 2,886 pts 366 - 348 w/l http://us.battle.net/sc2/en/profile/512680/1/Schyfis/ ~12 hr ago snapshott in gold: 2,833 pts 444 - 437 w/l http://us.battle.net/sc2/en/profile/2341271/1/Shrefu/ ~12 hr ago snapshot in gold: 2,789 pts 282 - 263 w/l http://us.battle.net/sc2/en/profile/342709/1/TangoFoxtrot/ ~12 hr ago snapshot in gold: 2,787 pts 198 - 173 w/l http://us.battle.net/sc2/en/profile/1993508/1/Logwai/ ~12 hr ago snapshott in gold: 2,777 pts 204 - 178 w/l http://us.battle.net/sc2/en/profile/538237/1/kingapes/ ~12 hr ago snapshot in gold: 2,741 pts 245 - 219 w/l Not Promoted: + Show Spoiler + http://us.battle.net/sc2/en/profile/1246189/1/Imperfect/ recent snapshot in gold: 2,944 pts 1,655 - 1,637 w/l ~12 hr ago snapshot: 1,646 - 1,636 w/l http://us.battle.net/sc2/en/profile/1475494/1/Gestapo/ recent snapshot in gold: 2,874 pts 302 - 292 w/l ~12 hr ago snapshot: 296 - 287 w/l http://us.battle.net/sc2/en/profile/1044360/1/BRASKI/ recent snapshot in gold: 2,874 pts 415 - 397 w/l ~12 hr ago snapshot: 402 - 391 w/l http://us.battle.net/sc2/en/profile/963543/1/goro/ recent snapshot in gold: 2,872 pts 517 - 501 w/l ~12 hr ago snapshot: 515 - 498 w/l http://us.battle.net/sc2/en/profile/853079/1/ProDiGY/ recent snapshot in gold: 2,870 pts 565 - 539 w/l ~12 hr ago snapshot: 563 - 539 w/l http://us.battle.net/sc2/en/profile/2235145/1/Vague/ recent snapshot in gold: 2,869 pts 480 - 429 w/l ~12 hr ago snapshot: 479 - 429 w/l http://us.battle.net/sc2/en/profile/1573408/1/Simitar/ recent snapshot in gold: 2,866 pts 378 - 372 w/l ~12 hr ago snapshot: 370 - 366 w/l http://sc2ranks.com/us/1791510/fractal recent snapshot in gold: 2,866 pts 333 - 284 w/l ~12 hr ago snapshot: 332 - 284 w/l http://us.battle.net/sc2/en/profile/1910680/1/FredHuang/ recent snapshot in gold: 2,865 pts 306 - 289 w/l ~12 hr ago snapshot: 303 - 286 w/l http://us.battle.net/sc2/en/profile/1904451/1/Prozak/ recent snapshot in gold: 2,865 pts 291 - 261 w/l ~12 hr ago snapshot: 288 - 259 w/l http://us.battle.net/sc2/en/profile/2285785/1/Calfosaurus/ recent snapshot in gold: 2,861 pts 264 - 234 w/l ~12 hr ago snapshot: 252 - 226 w/l http://us.battle.net/sc2/en/profile/709143/1/ShogunRua/ recent snapshot in gold: 2,857 pts 589 - 589 w/l ~12 hr ago snapshot: 574 - 576 w/l I'm out of time for now, but I have snapshots for plat and diamond too. | ||
|
SDream
Brazil896 Posts
The divisions filling process: With the new master league we had a lot of diamond players being taken from everywhere (every tier I mean...). In the end, some E Rank had 95 players, other 91, some S rank that was filled actually had 30 or so. But look here: http://www.sc2ranks.com/div/la/diamond/1/points/0 What do you see? Well, I'll tell you what: all these divisions with 97-98 players are E Rank. I was always wondering how the system would deal with divisions that were once filled but then lost players to promotion/demotion. That's how it works: Before creating a new division, the system will fill EVERY division on that tier. That was obvious for me before already. But there is more to it. If there is 2 divisions with 30 players, 2 with 50, 2 with 95, the system will first of all fill the 2 divisions with 30 till 50, then fill the 4 with 50 till 95, then fill them "together" till they have 100 players.  | ||
|
SDream
Brazil896 Posts
On February 09 2011 18:30 michaelhasanalias wrote: Just something else to add, The KOR top 200 is so closely linked to MMR because almost everyone in the top 200 has very low or no bonus pools. SEA is a different story however. http://sea.battle.net/sc2/en/blog/122430#blog if you look at the top 200 on SEA, which is plagued by inactivity (and I only looked at a few), The #200 guy actually seems to have one of the highest MMRs at 3200 (by my calculations) on the server. The #199 guy, by comparison, his MMR is only 2430. #2's MMR (spent bonus pool) is 2957. I don't feel like parsing the SEA top 200, as I do actually want to play some games today, but I have a strong suspicion that the top 200 is in fact NOT taking the top 200 MMRs on the server or anything related to hidden MMR, and is simply ranking the top 200 players purely by points, although there could be some cut-off. There are many players with higher MMR's than 2430 who are not on this top 200, and I suspect we could find several that are lower than that guy, since SEA's top 200 has so many inactives/casuals. Maybe next Tuesday morning I can try to parse the SEA top 200 once it comes out. I have a strong suspicion there are many players who have played comparable games with higher MMRs that are left out simply because they don't meat the 2081 point cutoff (2190 by next week if adjusted for inflation). Well, I've been looking at SEA an LA top 200 for some months now, so please believe me: It's NOT top 200 points. It is a process like this: Active players only -> (probably needs to spent 60%+ of their bonus pool) Top X MMR or Y% MMR -> (that has been proven by me, don't worry) Sort by points -> Done | ||
Excalibur_Z
United States12236 Posts
I do think it could be top 200 MMR sorted by points, but I don't want to say that's for certain the case because that would potentially allow smurf accounts to rise to the top with very very few games (like 10-20) played and therefore be included in the Top 200. Notwithstanding things such as a potential MMR cap or whatever, I just think it's risky. | ||
|
SDream
Brazil896 Posts
Also, I am pretty positive that the MMR cut is not 200, it is at least 202, but potencially top 300 or even top 1% server and just then they sort by points, so no way a player that do this will succeed in to getting at top 200. | ||
|
Excalibur_Z
United States12236 Posts
On February 09 2011 14:45 michaelhasanalias wrote: it's 475 if you're in a Rank-E division. it's only 150-155 (I think 155 but the data looks to support 150 from top 200 histories with diamond/master mix) if you're in Rank-S Diamond division. Basically by eyeballing this: http://sc2ranks.com/ranks/all/diamond/1/all/pointpool I could give you more or less the exact cutoffs and division listings for every division if sc2ranks could plug those aforementioned equations in. I've been trying to get in touch with shadow for a few weeks but no luck yet. As it stands, it seems that from Bronze to Master League, I think I'm losing about 20 points (my guess is a nice round 2250 since that was the number used in WoW's arena system). Someone could take the analysis I just did twice and apply it to any division manually. The outliers are the top of global diamond are players who haven't played any or enough games since 1.2 to get promoted. You can call the system's uncertainty whatever you want. If you prefer Sigma, or moving average, or some combination of both or a third metric, you're just showing a different calculation or estimation of the level of confidence that blizzard has that your MMR is stable enough to re-evaluate you for promotion. There is an enormous difference between confidence level and moving average as promotion criteria. We know 100% that it's the moving average as well as stability of that moving average that are the key requirements. Confidence implies that your opponent range has to be a certain (small) size before you can be promoted, we know that's not the case because sigma never gets small enough. We also know it's not periodic reviews that allow for promotion. I myself would also like to know where the magic 475 number came from. (315 + 150 = 465... 475 came from where exactly?) I'm not saying the league and division offsets aren't valuable in some way, they very well may be, but I think it's foolhardy to make such a broad assumption without significant data to back it up. As random user said, first we need huge statistical evidence to prove it. | ||
|
SDream
Brazil896 Posts
There has been proved that we have 7 tiers on Diamond (S->F). But it was also proved that most platinum players get promoted to E Diamond divisions, not F ones. Why would that be so? My theory is that the moving average of most players only stabilizes enough to become diamond once they are "E rank" worth with their MMR. What's your? | ||
|
Excalibur_Z
United States12236 Posts
On February 10 2011 06:36 SDream wrote: michaelhasanalias, another thing that I want to share with you that I just remembered. There has been proved that we have 7 tiers on Diamond (S->F). But it was also proved that most platinum players get promoted to E Diamond divisions, not F ones. Why would that be so? My theory is that the moving average of most players only stabilizes enough to become diamond once they are "E rank" worth with their MMR. What's your? Another thing I noticed yesterday was that there appear to be more 3v3/4v4 tiers per league in NA, but more 1v1 tiers in KR and EU. Maybe the more popular a game type is, the more tiers need to be created? | ||
|
SDream
Brazil896 Posts
On February 10 2011 07:04 Excalibur_Z wrote: Another thing I noticed yesterday was that there appear to be more 3v3/4v4 tiers per league in NA, but more 1v1 tiers in KR and EU. Maybe the more popular a game type is, the more tiers need to be created? Are you trying to say F Rank diamond could be a "dead" tier? Now that we know that 100% of the divisions must be filled in order to another one to be made, we can more easily prove silver-platinum tiers looking at the time of creation of new divisions, is there still a 95 player division when that happened? That means 2 different tiers. So I think we can work from there to finally confirm these tiers. | ||
|
Excalibur_Z
United States12236 Posts
On February 10 2011 07:22 SDream wrote: Are you trying to say F Rank diamond could be a "dead" tier? Now that we know that 100% of the divisions most be filled in order to another one to be made, we can more easily prove silver-platinum tiers looking at the time of creation of new divisions, is there still a 95 player division when that happened? That means 2 different tiers. So I think we can work from there to finally confirm these tiers. Mm, no, not necessarily. I mean we have confirmed seven tiers on KR and EU and SEA and LA but never NA for Diamond 1v1. However, I was looking at some other leagues and Silver 3v3 and 4v4 for example have 3 distinct tiers on NA, but only 2 on KR and EU. I mean maybe it differs per region by design. | ||
|
SDream
Brazil896 Posts
So we have 2 hipothesys: 1) Depending on players skills/numbers we have different number of tiers, which means it can change over time and some tiers can just stop being born (dying). That would means that MMR meaning is not linked with player base, but that doesn't make sense to me. 2) What you are seeing is just a coincidence for the eyes, there's the same number, you just can't find a period of time where they both have less than 60 players at the same time. I bet in this one cause I actually bet the 20% worst players is ALWAYS the guy with X MMR, the guy on top 2% is ALWAYS the Y MMR guy etc. I think the meaning of the MMR changes based on the meaning of active players distribution. Also, I am pretty sure this US division is F Rank: http://www.sc2ranks.com/div/59023/division-ladranix-romeo | ||
|
random user
85 Posts
- When SC2 launched, several good players complained about not being able to get promoted to diamond. Some of them even went so far as to lose several (7 was the number most tossed about) games to get promoted even though it was only anecdotal evidence that it worked. - How does the system know that S tier is the best tier? It probably doesn't, or didn't at the start. What if it looked at the population of users, and as the bell curve started elongating as more people started playing, it started putting people into the various diamond tiers. But what if the system thought there should be a tier that was higher than S? But what if it wasn't sure, and needed a certain threshold of people to be that high before it was sure there should be another tier and to go ahead and create it? - If that is the case, then you get something like what we saw. The system wants to wait until enough people are available to create the S+ tier (or whatever you want to call the tier higher than S), but there aren't a lot of those super good people, and they get impatient and throw games to get promoted. Thus, there are never enough people to actually create the S+ division. Eventually Blizzard realizes this is a bug and hard caps S as the best tier (until Master's gets implemented months later). How does this relate to the discussion? I'm not 100% sure. The single F Rank might be a side effect of capping the best tier as S, introducing a new bug which was then quickly fixed, which is why there seems to only be 1 division. Maybe the system at the start thought there would be a need for F rank on the first day, but then realized as MMRs climbed that people with that offset needed to be in plat instead to fulfill the 20% distribution. My apologies for just typing this all out without any evidence to back any of it up. I want to emphasize again here that I'm just brainstorming, and the reason I decided to still type it all out is perhaps it will help other people get ideas about what might be going on. | ||
|
SDream
Brazil896 Posts
Even on LA server where we know 100% of diamond divisions tiers we have only 1 F Rank division and 15 E Rank divisions, almost everyone that comes from platinum seem to go to E Rank mainly, D Rank gets some as well, but others don't get much. This is the one and only F Rank (confirmed) on LA server: http://us.battle.net/sc2/pt/profile/9522/2/Magus/ladder/2264 It had 97 players 2 months ago. It has 89 players as of today, so it seems it's losing more players than getting it. But there are still some players that get promoted there, so I don't think that it's dead, there is another reason for it to seem dead, so I doubt your theory is true. But thanks for sharing. | ||
|
michaelhasanalias
Korea (South)1231 Posts
| ||
|
Excalibur_Z
United States12236 Posts
| ||
|
michaelhasanalias
Korea (South)1231 Posts
| ||
|
SDream
Brazil896 Posts
On February 10 2011 10:38 michaelhasanalias wrote: I think this is exactly the case. The division that the 3700 point Bronze player is in offers strong, albeit somewhat anecdotal, supporting evidence That division was started 3 months ago, and still only has 67 players in it. This means that they can't find enough players that bad (presumably many 0-5 players do get placed there) that actually stay in that division. I've forget to correct you before, but the division of this bronze guy is full (100 players). On February 09 2011 13:55 SDream wrote: Interesting. Why is this bronze fellow not been promoted to silver yet? http://eu.battle.net/sc2/en/profile/1652040/1/HitTheSound/ladder/19971 He's active. Edit: The link isn't working anymore, what happened? Did he changed his name?? Anyway, I saw it full on the official site (eu.battle.net) o.O Edit2: I jsut noticed that more than 20 players in that division are tied with 0 points, but it does have 100 players on it. Edit3: I found him again: http://eu.battle.net/sc2/en/profile/1652040/1/wApToofck/ladder/19971#current-rank and he has 3705 points now. I am starting to think it is a huge bug, maybe a result of negative mmr bug??? o.o I found it fun how he changed his nick just when we were discussing him haha | ||
| ||
