|
On December 02 2012 08:31 EtherealDeath wrote:
Overall is a nearly worthless stat. Reason - it is a positive feedback loop if one race is slightly imba and/or happens to win more in the first round or two. Then more of the race distribution will be of that race, and overall counts mirrors - thus that race's overall win% will go up.
It's actually a "negative" feedback loop. If Zerg wins more games, there will be more ZvZs, which, by nature, are going to always be 50% winrate overall. If a race is winning non-mirrors more than 50%, mirrors being included brings the overall winrate down, not up.
also kinda annoying that you did all this precise calculation, but then you used commas as opposed to periods for your decimal places in your percentages??? wtf?
just saying... :D
Because Europe is backwards and weird (they use commas instead of decimal points and vice versa, iirc)
|
On December 02 2012 08:31 EtherealDeath wrote:
Overall is a nearly worthless stat. Reason - it is a positive feedback loop if one race is slightly imba and/or happens to win more in the first round or two. Then more of the race distribution will be of that race, and overall counts mirrors - thus that race's overall win% will go up.
Overall stats might be non significant but surely other stats are not (affected by race distribution)?
|
South Africa4316 Posts
On December 02 2012 07:17 Zenbrez wrote:Show nested quote +Aside from being interesting using data like this from a single tournament is pointless. Sample size way to small. Sample size? He used the population. There can not physically be a bigger "sample".
As you say, the sample used is the entire population of games at IPL, so we can, based on the sample, conclusively say that Zerg has done better than Terran on average. However, when people say the sample size is too small, they mean it's too small to draw conclusions regarding the overall balance (or perhaps just the overall pro-level balance) of Starcraft 2. In this case, the tournament is only a small sample of all the games played at a professional level, and it's probably too small to draw any strong conclusions from. Combining the results from the IPL5 with those from other more recent tournaments does provide fairly strong evidence that Zerg is too strong at the moment, but then again, not many people are denying that anymore.
Regarding overall winrates, it's expected to be 33%/33%/33% if there wre equal numbers of all races. However, since there weren't an equal number of players from each race, this is not the case. A more useful statistic would be to look at how significantly each race has over or under-performed its expected rate.
|
Three sets of games won't even make a difference compared to the hundreds that are already in the OP.
i agree
|
On December 02 2012 08:43 corpuscle wrote:Show nested quote +also kinda annoying that you did all this precise calculation, but then you used commas as opposed to periods for your decimal places in your percentages??? wtf?
just saying... :D Because Europe is awesome and right (they use commas instead of decimal points and vice versa, iirc, because that's the proper way to do it)
Fixed that for you.
|
well at least we were all wrong about tvp being an imbalanced matchup, its lk 50-50 across the board lol
|
Wow zergs players got really good in few weeks!!
|
On December 02 2012 07:59 Swagasaurus wrote:
OP you're seriously so incomprehensibly stupid i don't know where to begin.
The tournament isn't even over yet but you stressed that you were looking for the "Overall" winrates? Is this a really elaborate joke or something? That's going to skew the results immensely, THERE ARE STILL GAMES THAT HAVE TO BE PLAYED.
I wish people would actually spend time playing the game instead of coming up with the most retarded ways to justify their balance complaints.
User was warned for this post
This is why the op should be read in full. That said op thank you for the stats, think we could all just take a second to breath and let P and T adjust and watchthe stats even out overtime.
The overall I feel makes alit of sense to include simply because the other stats are seperated before hand. As stated by others please no more balance stuff please.
|
I'm not surprised at these statistics. It's this point it's laughable at how blatantly overpowered Zerg has been and still is consistently for such a long period of time. How can anyone play Zerg and publicly admit it?
After the latest "update" from Blizzard I refused to watch SC2 streams anymore, but as a hardcore fan I'm still concerned about the future of the game. Blizzard needs a new balance team.
|
On December 02 2012 08:40 urbaNo wrote:
the results are pretty one sided. i wonder if blizzard will exclude this piece of info so they can keep saying that zerg does not have an advantage globally or whatever.
also kinda annoying that you did all this precise calculation, but then you used commas as opposed to periods for your decimal places in your percentages??? wtf?
just saying... :D
Hah, I rarely pay attention to details unfortunately (believe me, I am not proud of it :D ).
Backing to topic, I don't believe if my results tell what is op and what is up. For example: "Overall" obviously favors zerg, but what explanation stands behind? I used all games played as denominator. At IPL5 the most players are playing zerg, thus generate the most games and (eventually) wins. Liquid'NonY explained it better than I could. In case if you need further explanation - his post is somewhere in 3rd page 
|
Well done, I think it's good we get more evidence for this.
|
On December 02 2012 08:53 Darrkhan wrote:
Wow zergs players got really good in few weeks!!
Oh yeah I guess all the zergs have just been practicing a lot more and are just better overall the game is completely balanced though #Blizzardlogic
|
On December 02 2012 08:43 corpuscle wrote:Show nested quote +On December 02 2012 08:31 EtherealDeath wrote:
Overall is a nearly worthless stat. Reason - it is a positive feedback loop if one race is slightly imba and/or happens to win more in the first round or two. Then more of the race distribution will be of that race, and overall counts mirrors - thus that race's overall win% will go up. It's actually a "negative" feedback loop. If Zerg wins more games, there will be more ZvZs, which, by nature, are going to always be 50% winrate overall. If a race is winning non-mirrors more than 50%, mirrors being included brings the overall winrate down, not up. Show nested quote +also kinda annoying that you did all this precise calculation, but then you used commas as opposed to periods for your decimal places in your percentages??? wtf?
just saying... :D Because Europe is backwards and weird (they use commas instead of decimal points and vice versa, iirc)
No you don't get it. His overall is actually not overall win % out of all a race's games. It's (# wins won by X race) / (#total number of games played - every game in the tourney regardless of race matchup). So actually a mirror counts as 100%, which drives the number up. Thus it's a positive feedback loop. Notice how the overall % numbers add up to 100%.
|
On December 02 2012 09:17 EtherealDeath wrote:Show nested quote +On December 02 2012 08:43 corpuscle wrote:On December 02 2012 08:31 EtherealDeath wrote:
Overall is a nearly worthless stat. Reason - it is a positive feedback loop if one race is slightly imba and/or happens to win more in the first round or two. Then more of the race distribution will be of that race, and overall counts mirrors - thus that race's overall win% will go up. It's actually a "negative" feedback loop. If Zerg wins more games, there will be more ZvZs, which, by nature, are going to always be 50% winrate overall. If a race is winning non-mirrors more than 50%, mirrors being included brings the overall winrate down, not up. also kinda annoying that you did all this precise calculation, but then you used commas as opposed to periods for your decimal places in your percentages??? wtf?
just saying... :D Because Europe is backwards and weird (they use commas instead of decimal points and vice versa, iirc) No you don't get it. His overall is actually not overall win % out of all a race's games. It's (# wins won by X race) / (#total number of games played - every game in the tourney regardless of race matchup). So actually a mirror oucnts as 100%, which drives the number up. Thus it's a positive feedback loop. Notice how the overall % numbers add up to 100%.
So the stats are still wrong in the opposite direction.
|
On December 02 2012 09:20 Solarsail wrote:Show nested quote +On December 02 2012 09:17 EtherealDeath wrote:On December 02 2012 08:43 corpuscle wrote:On December 02 2012 08:31 EtherealDeath wrote:
Overall is a nearly worthless stat. Reason - it is a positive feedback loop if one race is slightly imba and/or happens to win more in the first round or two. Then more of the race distribution will be of that race, and overall counts mirrors - thus that race's overall win% will go up. It's actually a "negative" feedback loop. If Zerg wins more games, there will be more ZvZs, which, by nature, are going to always be 50% winrate overall. If a race is winning non-mirrors more than 50%, mirrors being included brings the overall winrate down, not up. also kinda annoying that you did all this precise calculation, but then you used commas as opposed to periods for your decimal places in your percentages??? wtf?
just saying... :D Because Europe is backwards and weird (they use commas instead of decimal points and vice versa, iirc) No you don't get it. His overall is actually not overall win % out of all a race's games. It's (# wins won by X race) / (#total number of games played - every game in the tourney regardless of race matchup). So actually a mirror oucnts as 100%, which drives the number up. Thus it's a positive feedback loop. Notice how the overall % numbers add up to 100%. So the stats are still wrong in the opposite direction.
What do you mean they are still wrong in the opposite direction? His stats are correct.
edit - Though I'm not sure what his source is, as his % wins don't match up with the "source" in the OP.
|
On December 02 2012 09:17 EtherealDeath wrote:Show nested quote +On December 02 2012 08:43 corpuscle wrote:On December 02 2012 08:31 EtherealDeath wrote:
Overall is a nearly worthless stat. Reason - it is a positive feedback loop if one race is slightly imba and/or happens to win more in the first round or two. Then more of the race distribution will be of that race, and overall counts mirrors - thus that race's overall win% will go up. It's actually a "negative" feedback loop. If Zerg wins more games, there will be more ZvZs, which, by nature, are going to always be 50% winrate overall. If a race is winning non-mirrors more than 50%, mirrors being included brings the overall winrate down, not up. also kinda annoying that you did all this precise calculation, but then you used commas as opposed to periods for your decimal places in your percentages??? wtf?
just saying... :D Because Europe is backwards and weird (they use commas instead of decimal points and vice versa, iirc) No you don't get it. His overall is actually not overall win % out of all a race's games. It's (# wins won by X race) / (#total number of games played - every game in the tourney regardless of race matchup). So actually a mirror counts as 100%, which drives the number up. Thus it's a positive feedback loop. Notice how the overall % numbers add up to 100%.
Ah, yeah, okay, I see what you're saying. I thought you meant winrates in a more general sense, not the specific (and really odd) way that the OP did it.
|
OP should delete the misleading stats.
This guy is right:
On December 02 2012 07:20 corpuscle wrote:
You don't include mirrors because it makes the numbers misleading. Whichever race plays the most mirrors (in this case, Zerg) is going to have their number somewhat arbitrarily skewed towards 50%.
If, let's say, Zerg was 15-5 vT, 15-5 vP, and 30-30 vZ, they'd have a 60% overall winrate, which doesn't sound awful, but covers up the fact that they're actually winning 75% of the time in non-mirrors.
|
On December 02 2012 09:31 corpuscle wrote:Show nested quote +On December 02 2012 09:17 EtherealDeath wrote:On December 02 2012 08:43 corpuscle wrote:On December 02 2012 08:31 EtherealDeath wrote:
Overall is a nearly worthless stat. Reason - it is a positive feedback loop if one race is slightly imba and/or happens to win more in the first round or two. Then more of the race distribution will be of that race, and overall counts mirrors - thus that race's overall win% will go up. It's actually a "negative" feedback loop. If Zerg wins more games, there will be more ZvZs, which, by nature, are going to always be 50% winrate overall. If a race is winning non-mirrors more than 50%, mirrors being included brings the overall winrate down, not up. also kinda annoying that you did all this precise calculation, but then you used commas as opposed to periods for your decimal places in your percentages??? wtf?
just saying... :D Because Europe is backwards and weird (they use commas instead of decimal points and vice versa, iirc) No you don't get it. His overall is actually not overall win % out of all a race's games. It's (# wins won by X race) / (#total number of games played - every game in the tourney regardless of race matchup). So actually a mirror counts as 100%, which drives the number up. Thus it's a positive feedback loop. Notice how the overall % numbers add up to 100%. Ah, yeah, okay, I see what you're saying. I thought you meant winrates in a more general sense, not the specific (and really odd) way that the OP did it.
Yea his overall is a bit odd, but it is what it is xD
|
On December 02 2012 07:09 Zenbrez wrote:Show nested quote +On December 02 2012 07:07 Viperbird wrote:On December 02 2012 07:04 corpuscle wrote:
I don't really see how a 69% vT and 61% vP can average out to 52%, for example I don't understand either... If I won 9-1 tvz, I have a 90% winrate. If I won 1-4 tvp, I have a 20% winrate.
I have a 66.67% (10/15) total winrate, even though the average of the 2 is 55% (110/2).
Wrong math broski. Both of the winrate above were above 52% so there is no way the average can be less when both the numbers that make it up are over 52%
|
It would be nice to see the amount of games played next to each stat.
|
|
|
|
|
|
EPT
