Hi there guys. This is my first topic on TL. I don't play BW or SC2 but I've been following Korean leagues for more than a year so far I like it very much. I'm not much of a writer but I just thought you might be interested in some Proleague statistic I prepared.
As all of you probably know the changes in Proleague format affected how teams play and their results as well, obviously favoring teams with deep lineups rather than teams with sole strong aces.
I've been looking for a handy way to evaluate the balance of each particular team, so here's my method. A little explanation on the team balance coefficient that I used. I analyzed all players on team rosters that appeared on stage at least once during the series (even if they didn't manage to win a single match) based on the number of games they won. In two words this means: 1 win = +1 point, appearance with no wins = 0 points but this zero result affects the team's stats. Then I formed the arrays for each team consisting of the points scored by the players. The formula for my team balance coefficient is as following:
So if all players in a team have won equal number of games each the team balance coefficient is 100% (the team is perfectly balanced). On the other hand, if all wins were brought to the team by a single player the coefficient will be 0% (a team of one player). The more is the coefficient, the more balanced is the team (the smoother is the bar graph representing wins count and the more even are the slices on the pie chart).
Now, the least balanced team in today's Proleague is of course: + Show Spoiler +
I would be glad to see your comments on this. If you liked my analysis I could do it after future rounds. The more data, the more accurate are the stats.
Cool topic, although I didn't need a formula to know which team is the most imbalanced. lol Currently team balance does not directly correspond to the team's overall performance but will that change as the season progresses?
I appreciate the effort you put I think your formula is not relevant
We are actually interested in the depth of the team, which team has at least 3 decent players, including an ace.
A team with 3 players with a very high winrate (including ace matches) and 3 players who suck is a lot better than a team with 7 players who suck equally. But your formula would say the contrary.
STX got a very good score with your formula because they used 11 different players, and have an overall Diff = 0
And SKT got a very mediocre score just because Bisu raped the shit out of everyone and made the win / loss record unbalanced ? I mean do you realize that if Bisu had lost two games, the team would have got a better score with a worse ace player ? It doesn't make any sense. A balanced team is a team with a good ace player (it doesn't mean a good ace player makes a team good, Oz anyone ?)
Oh and my examples have nothing to do with my T1 icon I promise, they were just appropriate examples of the flaws of your formula
And very nice graphics btw !!
edit : typo and yay 1337 posts
edit 2 : wait, a team who would have lost the 4 matches 0-4 using the same 4 players would have got a 100% right ? When we use the word balanced for a team it's related to its depth. According to your formula, a team composed of 4 awful players would be the deepest team of the proleague ? wtf.
Those are some ridiculously good looking graphs. However I don't think you have enough data yet, I would love to see what these would look like after a full round has been completed. Good work anyway!
On October 27 2010 19:20 endy wrote: I mean do you realize that if Bisu had lost two games, the team would have got a better score with a worse ace player ? It doesn't make any sense. A balanced team is a team with a good ace player (it doesn't mean a good ace player makes a team good, Oz anyone ?)
You're right, but I must emphasize that my balance rating isn't supposed to be related to team's performance. Theoretically it's possible to have a team with 10-20% balance ratio and it will rape any other team. The coefficient is just depicting the distribution of efforts between the players.
On October 27 2010 19:20 endy wrote: edit 2 : wait, a team who would have lost the 4 matches 0-4 using the same 4 players would have got a 100% right ? When we use the word balanced for a team it's related to its depth. According to your formula, a team composed of 4 awful players would be the deepest team of the proleague ? wtf.
Haha, luckily my formula doesn't work with zero wins (it results in division by zero). Edit: But I think I should consider adding some correction so that the formula will also depict the absolute strengths of the players.
On October 27 2010 19:20 endy wrote: I mean do you realize that if Bisu had lost two games, the team would have got a better score with a worse ace player ? It doesn't make any sense. A balanced team is a team with a good ace player (it doesn't mean a good ace player makes a team good, Oz anyone ?)
You're right, but I must emphasize that my balance rating isn't supposed to be related to team's performance. Theoretically it's possible to have a team with 10-20% balance ratio and it will rape any other team. The coefficient is just depicting the distribution of efforts between the players.
Yes, but if that was your purpose, we need a lot more data to obtain relevant results. Leta and Really are making a lot of efforts, believe me.
Also, read my edit2, which is an even bigger flaw in your formula. edit : you saw my edit, I know it doesnt work with 0 wins, but that's mathematically the results you obtain (using the limits, amirite ?)
I'm not saying this just because I harshly criticized your results, I sincerely find your graphics awesome. I'd love to see you designing the Proleague Round 1recaps (no offense at the current TL graphics team :D)
Is it possible to fuse this with team ''strenght'' to find out which teams have the highest amount of good-to-very-good playesrs. thus more likely to win in best of seven format?
The chart is really nice, though I think the depth of the team is more important than the balance, I still appreciate the efforts you made. BTW, I have translated your thread on www.plu.cn: http://sc.plu.cn/note/2010-10-27/22276.html
On October 27 2010 20:43 Hsanrb wrote: The Hwaseung OZ pie chart adds up to 112.5%, other than that its a great read. Though these stats probably won't work for Winners League.
Fixed this and one other mistake in KHAN graph no one noticed. Thank you.
On October 27 2010 20:47 vizhi_j wrote: The chart is really nice, though I think the depth of the team is more important than the balance, I still appreciate the efforts you made. BTW, I have translated your thread on www.plu.cn: http://sc.plu.cn/note/2010-10-27/22276.html
Thanks again! It is really an awesome work!
Thank you too! Please fix OZ and KHAN charts, I fixed some mistakes in them.
On October 27 2010 20:43 Hsanrb wrote: The Hwaseung OZ pie chart adds up to 112.5%, other than that its a great read. Though these stats probably won't work for Winners League.
Fixed this and one other mistake in KHAN graph no one noticed. Thank you.
On October 27 2010 20:47 vizhi_j wrote: The chart is really nice, though I think the depth of the team is more important than the balance, I still appreciate the efforts you made. BTW, I have translated your thread on www.plu.cn: http://sc.plu.cn/note/2010-10-27/22276.html
Thanks again! It is really an awesome work!
Thank you too! Please fix OZ and KHAN charts, I fixed some mistakes in them.
Well, you overestimate me, in fact, I know nothing about kinda maths like the formula, I did learn it since I was enrolled into university. All I know about maths is plus and minus....
On October 27 2010 21:08 vizhi_j wrote: Well, you overestimate me, in fact, I know nothing about kinda maths like the formula, I did learn it since I was enrolled into university. All I know about maths is plus and minus....
Just replace the pictures you uploaded with the new pictures from my OP post. :D
On October 27 2010 20:43 Hsanrb wrote: The Hwaseung OZ pie chart adds up to 112.5%, other than that its a great read. Though these stats probably won't work for Winners League.
Fixed this and one other mistake in KHAN graph no one noticed. Thank you.
On October 27 2010 20:47 vizhi_j wrote: The chart is really nice, though I think the depth of the team is more important than the balance, I still appreciate the efforts you made. BTW, I have translated your thread on www.plu.cn: http://sc.plu.cn/note/2010-10-27/22276.html
Thanks again! It is really an awesome work!
Thank you too! Please fix OZ and KHAN charts, I fixed some mistakes in them.
Well, you overestimate me, in fact, I know nothing about kinda maths like the formula, I did learn it since I was enrolled into university. All I know about maths is plus and minus....
i have been led to believe that all chinese people can do calculus at birth. fucking self-hating liberal bs.
i'd like to see this go on later.. i still think it's a bit too early to see which team ist he most balanced... like fantasy is bound to get better right???
On October 27 2010 20:43 Hsanrb wrote: The Hwaseung OZ pie chart adds up to 112.5%, other than that its a great read. Though these stats probably won't work for Winners League.
Fixed this and one other mistake in KHAN graph no one noticed. Thank you.
On October 27 2010 20:47 vizhi_j wrote: The chart is really nice, though I think the depth of the team is more important than the balance, I still appreciate the efforts you made. BTW, I have translated your thread on www.plu.cn: http://sc.plu.cn/note/2010-10-27/22276.html
Thanks again! It is really an awesome work!
Thank you too! Please fix OZ and KHAN charts, I fixed some mistakes in them.
Well, you overestimate me, in fact, I know nothing about kinda maths like the formula, I did learn it since I was enrolled into university. All I know about maths is plus and minus....
i have been led to believe that all chinese people can do calculus at birth. fucking self-hating liberal bs.
Perhaps it is the truth, but it is just for few genius. Most of us are common, but our textbook is much difficult than yours. SO, there is a rumor comes.... and what's more, due to the exam system, many of us are limped about arts or science. that's sad.
On October 27 2010 21:08 vizhi_j wrote: Well, you overestimate me, in fact, I know nothing about kinda maths like the formula, I did learn it since I was enrolled into university. All I know about maths is plus and minus....
Just replace the pictures you uploaded with the new pictures from my OP post. :D
On October 27 2010 19:20 endy wrote:...but that's mathematically the results you obtain (using the limits, amirite ?)
No, I didn't use limits there. Stdev stands for standard deviation:
I know what's the standard dev lol. I meant if you use the limits where the amount of games won is very close to zero, in the case all the players won, you obtain a 100% score
hey thats a really cool statistic. you should keep it updated throughout the year so the numbers become more significant. later on, you may have to remove some players to keep it realistic (i.e. players with only 1 or 2 games played after round 5 shouldn't be counted).
in regards to some of the other posts, the formula doesn't say which team is BETTER, just which team is more BALANCED in getting wins throughout the roster. There really is no perfect statistic, even W-L (Perfectman is NOT a better player than Light or Fantasy). But used in conjunction with other statistics, such as number of appearances, ace match records, opponents records, etc., this statistic adds to the discussion.
This is not a very good data since there hasn't been many games played yet. I'm betting few teams will rise quickly and few will fall after couple more games.
Interesting, and this also made me realize some funny things. (calm -2, turn +3, ) and others lol
You should talk to the people who do the proleague reports because they do stats sometimes and it seems like you could help, they are very interesting.
This is really sick. I think you should do it at the end/in the middle of each round, with the one at the end being a complete one for the whole season (compiling rounds), and with the in the middle being for that round only. It's a lot of work but that seems like the most rewarding to me. Also, I feel like though team balance is really important, neglecting ace is foolish. SKT beat KT and another team in ace with 2 different players - this adds a new concept of depth. I think that there should be some sort of measure for ace victories. Thanks for this!
On October 27 2010 19:00 Ryo wrote: Cool topic, although I didn't need a formula to know which team is the most imbalanced. lol Currently team balance does not directly correspond to the team's overall performance but will that change as the season progresses?
To elaborate on this, it'd be nice to have a separate statistic for how accurately your chart predicts wins, and how it is is relevant to current standings... I think that'd be the best qualifier for you work (:
This is cool, but it doesn't really say all that much at this point; I'd be interested in seeing the results later, when there's more data to work with.
This is interesting stuff. I think the "balance" that is trying to be calculated here is how evenly are the players pulling their weight for the team. Now it doesn't necessarily reflect that some players are better than others in the grand scheme but rather it reflects how they are doing now this particular round (i.e. Fantasy slumping. So much internal rage.) These numbers will be dependent on whether players dominate or fall into slumps and the length of time these periods occur. So this isn't exactly a predictor for things to come but it's interesting work nonetheless.
Coolio graphs. Tho I don't think this is that relevant without including several other variables. It would be cool to see some analysis regarding which team is best suited to win in the quarter finals, or which 4 players are most likely to win the MSL/OSL based on their records and the upcoming match-ups.
It's a cool graph, although there are flaws. MBC doesn't seem no where near as balanced when you consider that it's literally only those six (Sea/Light/Hyun/Saint/Jaehoon/Tyson) that actually ever play.
Still, good work. These graphs are definitely interesting reads
wow, you know at the beginning of the season, the order of balanced teams was nothing like the order of the teams that were leading in score, but now it is pretty damn close. Apparently with the bo7 format, if there is not more than a couple players pulling their weight, then the more balanced team actually is better. Although, there is 2 rounds of winners league, so we'll see how the stats look after that..