EPTAugust 2011 KeSPA Ranking - Page 5
| Forum Index > BW General |
|
xarthaz
1704 Posts
| ||
|
TwoToneTerran
United States8841 Posts
| ||
|
cascades
Singapore6122 Posts
| ||
|
Sypuku
Russian Federation20 Posts
On August 01 2011 22:44 xarthaz wrote: yes, the formula is biased, lets be honest.. Lets stop trolling LOL ;-) U have big buttheart from Flashs records everybody knows it. | ||
|
xarthaz
1704 Posts
On August 01 2011 22:46 TwoToneTerran wrote: By your definition, TL Elo is biased. Since the only way to be unbiased is to treat all games equally, then Prelims should be counted (Kespa Ranking does count prelims). You misunderstand. Elo IS unbiased, with regard to the games that are a subject of it. The games that are not a subject of it, it does not even attempt to make a statement about, hence it not being bias. This is completely different in the case of the Kespa rating, as it has bias even within the subject whom it describes: the differentiation of ratings of the games that it is calculated from. | ||
|
Ryo
8787 Posts
On August 01 2011 22:05 Winechu wrote: Slight differences from the ranking posted on Kespa's website, strange. http://www.e-sports.or.kr/static/Eng/ranking.kea?m_code=&imgSel=sel21 Roo with the biggest upward surge, by 22. Fuck yeah kt rooster! Hoeja impressive too with a 13 place jump. KT doing well. Stats being in top 10 nearly on the back on his performance in teamleagues alone shows how much of an asset he is in PL, don't think any team has a better 1-2 punch than KT. Coupled with a rejuvenated zerg line following the acquisition of coach SES, don't think anyone should be surprised they made it this far. Lol bisang rising 13 places from his solitary win over hero. The ranking on the original Korean site matches the ones released by Fomos and Dailyesports: http://www.e-sports.or.kr/news/notice_view.kea?m_code=news_30&PageNo=1&seq=836&where=&query= Kespa's English site is the one with the errors. | ||
|
Gamegene
United States8308 Posts
T^T! | ||
|
Sypuku
Russian Federation20 Posts
On August 01 2011 23:05 Gamegene wrote: Flash 500 points up on Jaedong... T^T! Anybody cares of it? | ||
|
moopie
12605 Posts
On August 01 2011 23:03 Ryo wrote: The ranking on the original Korean site matches the ones released by Fomos and Dailyesports: http://www.e-sports.or.kr/news/notice_view.kea?m_code=news_30&PageNo=1&seq=836&where=&query= Kespa's English site is the one with the errors. Its not just the English site. Their database(1) and their news notice do not match up. One of them is incorrect. For now I left both data sets in the OP. | ||
|
Winechu
Singapore1186 Posts
On August 01 2011 22:44 xarthaz wrote: yes, the formula is biased, lets be honest. Because there exists bias regarding the valuation of games: not all games are valued equally. Hence the Kespa rating being biased, unlike ELO, where Bisu's rank is more reasonable. Why should games be valued equally? Obviously some games are going to have more significance over other games and hence should accordingly have a higher weightage. The two rating systems differ in what they measure too, elo more of skill and Kespa more of a player's value in terms of his success and value to his team. For the latter, ascribing weightage makes sense. On August 01 2011 23:02 xarthaz wrote: You misunderstand. Elo IS unbiased, with regard to the games that are a subject of it. The games that are not a subject of it, it does not even attempt to make a statement about, hence it not being bias. This is completely different in the case of the Kespa rating, as it has bias even within the subject whom it describes: the differentiation of ratings of the games that it is calculated from. The point TTT was trying to make is that if all games are treated equally, then prelim games are equal to that of a regular PL match or SL match and hence should be included in the calculation of elo. If you assume them to be equal, excluding those matches would be arbitrary. Hence Elo WOULD be biased, in spite of not arbitrarily ascribing weightage to games unlike ELO. You can't subtype it and call it a completely different case. Bias is bias. | ||
|
Koka-kola
241 Posts
| ||
Shikyo
Finland33997 Posts
On August 01 2011 23:02 xarthaz wrote: You misunderstand. Elo IS unbiased, with regard to the games that are a subject of it. The games that are not a subject of it, it does not even attempt to make a statement about, hence it not being bias. This is completely different in the case of the Kespa rating, as it has bias even within the subject whom it describes: the differentiation of ratings of the games that it is calculated from. So you're saying that a random proleague game should be valued as much as a starleague finals, and that prelims aren't supposed to be counted? I'm pretty sure Kespa rating is a much better ranking system, it has much less fluctuation due to variance and values individual leagues more, as one should. | ||
|
SaetZero
United States855 Posts
| ||
|
vectorix108
United States4633 Posts
| ||
|
shadymmj
1906 Posts
In individual leagues you know who you're playing against in advance and can prepare accordingly. For that reason individual leagues should have greater importance. | ||
|
TwoToneTerran
United States8841 Posts
On August 01 2011 23:02 xarthaz wrote: You misunderstand. Elo IS unbiased, with regard to the games that are a subject of it. The games that are not a subject of it, it does not even attempt to make a statement about, hence it not being bias. This is completely different in the case of the Kespa rating, as it has bias even within the subject whom it describes: the differentiation of ratings of the games that it is calculated from. Doesn't matter, Kespa Ranking atleast acknowledges all serious, non-exhibition games. Elo biasedly excludes other games (because Elo has a problem with inflation, so it excludes prelims which add a very high amount of games and therefore speeds the inflation. Also maybe the TL staff cares less about prelims) and puts exact value on PL Round 6 garbage time games over game 5 of a finals. It's silly. Elo is excellent in depicting consistency. It is not a clear, accurate ranking for who is clearly achieving the most. | ||
|
SoulSever
Canada779 Posts
| ||
|
TwoToneTerran
United States8841 Posts
On August 02 2011 01:35 SoulSever wrote: JD will be back on top in no time. Hydra and Zero won't win OSL so it won't take too much to get that #1 ranking back. And who the hell says "where did Soulkey come from?"?????? Was beastly in proleague and is ripping up MSL Jaedong won't win the OSL, either... | ||
|
naneri
Kyrgyzstan165 Posts
On August 02 2011 00:18 shadymmj wrote: A factor called "the coach" in proleague does exist, and SKT has got a very good one. Bisu may be the ace, but if there's a terran that needs taking out, Best can do it. With the team depth that SKT has it is hard to prepare specifically for Bisu, but that is obviously not the case for KT or OZ. Let's face it, we don't hear of a "Bisu got sniped" very much, and there's a reason why. In individual leagues you know who you're playing against in advance and can prepare accordingly. For that reason individual leagues should have greater importance. Acctually it is not his PvT that is lacking but his PvP now. Cause Bisus' PvT winrate is even higher than his PvZ. He only lost 5 games this year and 3 of them were to Flash. | ||
|
Holgerius
Sweden16951 Posts
 | ||
| ||
