|
On July 03 2012 18:34 keglu wrote:Show nested quote +On July 03 2012 18:18 Assirra wrote:On July 03 2012 18:09 keglu wrote:On July 03 2012 17:49 Stiluz wrote: OP's own analysis about qualifiers is methodologically extremely flawed. If you want to count qualifiers to compare winrates with previous months, you have to go back and include tournament qualifying rounds for every single month. As another post mentioned, you can't cherry pick when to include and discuss winrates from qualifiers, you have to be consistant about where your data comes from. Or we can start now so for once we can have singifcant amount of data to jugde anything (500 games instead of 100). Who cares about previous data, you can't draw any conlusion on winrates changing over 10-20% each month. Also TSL, OSL qualifiers werent played before, MLG im not sure. And neither can you change the rules of something in the middle of series, your whole series is broken that way. Who knows what would have happened if the code A qualifiers for instance always got counted. It's kinda silly that people want to include qualifiers that were never included just to proof their point. Again i dont care about series, i care about current state of the game and having reliable data. Till know i never took Korean graph seriously because of sample size so let it be. We have chance to have reliable graph in future if we add all this data. It does not matter what you care about...a graph without consistency is not a graph at all. What you want is raw stats, not a graph.
|
On July 03 2012 18:40 Assirra wrote:Show nested quote +On July 03 2012 18:34 keglu wrote:On July 03 2012 18:18 Assirra wrote:On July 03 2012 18:09 keglu wrote:On July 03 2012 17:49 Stiluz wrote: OP's own analysis about qualifiers is methodologically extremely flawed. If you want to count qualifiers to compare winrates with previous months, you have to go back and include tournament qualifying rounds for every single month. As another post mentioned, you can't cherry pick when to include and discuss winrates from qualifiers, you have to be consistant about where your data comes from. Or we can start now so for once we can have singifcant amount of data to jugde anything (500 games instead of 100). Who cares about previous data, you can't draw any conlusion on winrates changing over 10-20% each month. Also TSL, OSL qualifiers werent played before, MLG im not sure. And neither can you change the rules of something in the middle of series, your whole series is broken that way. Who knows what would have happened if the code A qualifiers for instance always got counted. It's kinda silly that people want to include qualifiers that were never included just to proof their point. Again i dont care about series, i care about current state of the game and having reliable data. Till know i never took Korean graph seriously because of sample size so let it be. We have chance to have reliable graph in future if we add all this data. It does not matter what you care about...a graph without consistency is not a graph at all. What you want is raw stats, not a graph.
Ok lets have consistent graph showing nothing, let it be. Also to clarify: its ok to add new tournamet - proleague to graph but not ok to add new TSL/OSL qualifiers with better players and much more games?
|
Just to pull a protoss:
And still Hero won. Just shows that toss is even more imba!
|
On July 03 2012 18:43 keglu wrote:Show nested quote +On July 03 2012 18:40 Assirra wrote:On July 03 2012 18:34 keglu wrote:On July 03 2012 18:18 Assirra wrote:On July 03 2012 18:09 keglu wrote:On July 03 2012 17:49 Stiluz wrote: OP's own analysis about qualifiers is methodologically extremely flawed. If you want to count qualifiers to compare winrates with previous months, you have to go back and include tournament qualifying rounds for every single month. As another post mentioned, you can't cherry pick when to include and discuss winrates from qualifiers, you have to be consistant about where your data comes from. Or we can start now so for once we can have singifcant amount of data to jugde anything (500 games instead of 100). Who cares about previous data, you can't draw any conlusion on winrates changing over 10-20% each month. Also TSL, OSL qualifiers werent played before, MLG im not sure. And neither can you change the rules of something in the middle of series, your whole series is broken that way. Who knows what would have happened if the code A qualifiers for instance always got counted. It's kinda silly that people want to include qualifiers that were never included just to proof their point. Again i dont care about series, i care about current state of the game and having reliable data. Till know i never took Korean graph seriously because of sample size so let it be. We have chance to have reliable graph in future if we add all this data. It does not matter what you care about...a graph without consistency is not a graph at all. What you want is raw stats, not a graph. Ok lets have consistent graph showing nothing, let it be. Also to clarify: its ok to add new tournamet - proleague to graph but not ok to add new TSL/OSL qualifiers with better players and much more games? Well if that is the case why not just remove international graphics completely? I mean, its clearly all about the better players so. You cannot adjust the rules as you see fit, cause then this whole graph would be worthless. This is about tournaments, not qualifiers, doesn't matter how good the qualifiers are.
|
On July 03 2012 17:22 ABear wrote: I do feel like Zerg probably has an advantage over every race ATM(larva inject needs to be nerfed), but if you are going to include code A qualfiiers/TSL qualifiers/OSL qualfiers, and you want to remove teamless/amateur koreans from these results, then you have to do this every single month for every MU and not just the months that match your desired result.
I don't think you want to do that, since GSL code S has been unnaturally favorable statistics wise for terrans over the course of SC2s life, and if you went back and edited in all high level qualifiers and tournaments that didn't have random ladder players in it, it would devalue the code S stats and you'd more than likely see terran drop a couple of percentages every month and months that might have seemed imbalanced now seem a lot more balanced.
I agree that all high level relevant tournamnets should be included in the data. It's more of a logistics issue at this stage. The graphs only use TLPD data.
|
On July 03 2012 19:18 Assirra wrote:Show nested quote +On July 03 2012 18:43 keglu wrote:On July 03 2012 18:40 Assirra wrote:On July 03 2012 18:34 keglu wrote:On July 03 2012 18:18 Assirra wrote:On July 03 2012 18:09 keglu wrote:On July 03 2012 17:49 Stiluz wrote: OP's own analysis about qualifiers is methodologically extremely flawed. If you want to count qualifiers to compare winrates with previous months, you have to go back and include tournament qualifying rounds for every single month. As another post mentioned, you can't cherry pick when to include and discuss winrates from qualifiers, you have to be consistant about where your data comes from. Or we can start now so for once we can have singifcant amount of data to jugde anything (500 games instead of 100). Who cares about previous data, you can't draw any conlusion on winrates changing over 10-20% each month. Also TSL, OSL qualifiers werent played before, MLG im not sure. And neither can you change the rules of something in the middle of series, your whole series is broken that way. Who knows what would have happened if the code A qualifiers for instance always got counted. It's kinda silly that people want to include qualifiers that were never included just to proof their point. Again i dont care about series, i care about current state of the game and having reliable data. Till know i never took Korean graph seriously because of sample size so let it be. We have chance to have reliable graph in future if we add all this data. It does not matter what you care about...a graph without consistency is not a graph at all. What you want is raw stats, not a graph. Ok lets have consistent graph showing nothing, let it be. Also to clarify: its ok to add new tournamet - proleague to graph but not ok to add new TSL/OSL qualifiers with better players and much more games? Well if that is the case why not just remove international graphics completely? I mean, its clearly all about the better players so. You cannot adjust the rules as you see fit, cause then this whole graph would be worthless. This is about tournaments, not qualifiers, doesn't matter how good the qualifiers are.
Why remove international graph? Its only one you can draw conclusion from. I didnt know that tournament/qulifier rule existed. Code A is qulifier to Code S so it seems like there is not such a rule btw.
|
On July 03 2012 19:18 Assirra wrote:Show nested quote +On July 03 2012 18:43 keglu wrote:On July 03 2012 18:40 Assirra wrote:On July 03 2012 18:34 keglu wrote:On July 03 2012 18:18 Assirra wrote:On July 03 2012 18:09 keglu wrote:On July 03 2012 17:49 Stiluz wrote: OP's own analysis about qualifiers is methodologically extremely flawed. If you want to count qualifiers to compare winrates with previous months, you have to go back and include tournament qualifying rounds for every single month. As another post mentioned, you can't cherry pick when to include and discuss winrates from qualifiers, you have to be consistant about where your data comes from. Or we can start now so for once we can have singifcant amount of data to jugde anything (500 games instead of 100). Who cares about previous data, you can't draw any conlusion on winrates changing over 10-20% each month. Also TSL, OSL qualifiers werent played before, MLG im not sure. And neither can you change the rules of something in the middle of series, your whole series is broken that way. Who knows what would have happened if the code A qualifiers for instance always got counted. It's kinda silly that people want to include qualifiers that were never included just to proof their point. Again i dont care about series, i care about current state of the game and having reliable data. Till know i never took Korean graph seriously because of sample size so let it be. We have chance to have reliable graph in future if we add all this data. It does not matter what you care about...a graph without consistency is not a graph at all. What you want is raw stats, not a graph. Ok lets have consistent graph showing nothing, let it be. Also to clarify: its ok to add new tournamet - proleague to graph but not ok to add new TSL/OSL qualifiers with better players and much more games? Well if that is the case why not just remove international graphics completely? I mean, its clearly all about the better players so. You cannot adjust the rules as you see fit, cause then this whole graph would be worthless. This is about tournaments, not qualifiers, doesn't matter how good the qualifiers are.
Those qualifiers are tournaments.
|
On July 03 2012 20:59 MrSalamandra wrote:Show nested quote +On July 03 2012 19:18 Assirra wrote:On July 03 2012 18:43 keglu wrote:On July 03 2012 18:40 Assirra wrote:On July 03 2012 18:34 keglu wrote:On July 03 2012 18:18 Assirra wrote:On July 03 2012 18:09 keglu wrote:On July 03 2012 17:49 Stiluz wrote: OP's own analysis about qualifiers is methodologically extremely flawed. If you want to count qualifiers to compare winrates with previous months, you have to go back and include tournament qualifying rounds for every single month. As another post mentioned, you can't cherry pick when to include and discuss winrates from qualifiers, you have to be consistant about where your data comes from. Or we can start now so for once we can have singifcant amount of data to jugde anything (500 games instead of 100). Who cares about previous data, you can't draw any conlusion on winrates changing over 10-20% each month. Also TSL, OSL qualifiers werent played before, MLG im not sure. And neither can you change the rules of something in the middle of series, your whole series is broken that way. Who knows what would have happened if the code A qualifiers for instance always got counted. It's kinda silly that people want to include qualifiers that were never included just to proof their point. Again i dont care about series, i care about current state of the game and having reliable data. Till know i never took Korean graph seriously because of sample size so let it be. We have chance to have reliable graph in future if we add all this data. It does not matter what you care about...a graph without consistency is not a graph at all. What you want is raw stats, not a graph. Ok lets have consistent graph showing nothing, let it be. Also to clarify: its ok to add new tournamet - proleague to graph but not ok to add new TSL/OSL qualifiers with better players and much more games? Well if that is the case why not just remove international graphics completely? I mean, its clearly all about the better players so. You cannot adjust the rules as you see fit, cause then this whole graph would be worthless. This is about tournaments, not qualifiers, doesn't matter how good the qualifiers are. Those qualifiers are tournaments. Then why were they never accounted before but suddenly people are screaming they should? a qualifier is not a tournament, it is a qualifier to get in the tournament.
|
On July 03 2012 20:59 MrSalamandra wrote:Show nested quote +On July 03 2012 19:18 Assirra wrote:On July 03 2012 18:43 keglu wrote:On July 03 2012 18:40 Assirra wrote:On July 03 2012 18:34 keglu wrote:On July 03 2012 18:18 Assirra wrote:On July 03 2012 18:09 keglu wrote:On July 03 2012 17:49 Stiluz wrote: OP's own analysis about qualifiers is methodologically extremely flawed. If you want to count qualifiers to compare winrates with previous months, you have to go back and include tournament qualifying rounds for every single month. As another post mentioned, you can't cherry pick when to include and discuss winrates from qualifiers, you have to be consistant about where your data comes from. Or we can start now so for once we can have singifcant amount of data to jugde anything (500 games instead of 100). Who cares about previous data, you can't draw any conlusion on winrates changing over 10-20% each month. Also TSL, OSL qualifiers werent played before, MLG im not sure. And neither can you change the rules of something in the middle of series, your whole series is broken that way. Who knows what would have happened if the code A qualifiers for instance always got counted. It's kinda silly that people want to include qualifiers that were never included just to proof their point. Again i dont care about series, i care about current state of the game and having reliable data. Till know i never took Korean graph seriously because of sample size so let it be. We have chance to have reliable graph in future if we add all this data. It does not matter what you care about...a graph without consistency is not a graph at all. What you want is raw stats, not a graph. Ok lets have consistent graph showing nothing, let it be. Also to clarify: its ok to add new tournamet - proleague to graph but not ok to add new TSL/OSL qualifiers with better players and much more games? Well if that is the case why not just remove international graphics completely? I mean, its clearly all about the better players so. You cannot adjust the rules as you see fit, cause then this whole graph would be worthless. This is about tournaments, not qualifiers, doesn't matter how good the qualifiers are. Those qualifiers are tournaments.
yes, a tournament to a much larger tournament where every player is a super star. The qualifiers is weeding out the bad players from the good, if we haven't counted qualifiers in the past, why should we now? The qualifiers are like College football, everyone is pretty good but there is only 1 or 2 people on each football team who could be NFL potential, but in the major tournament, everyone is a super star. Qualifiers shouldn't be counted because there is too large of a skill gap, a qualifier could consist of bronze to GM players for all we know. Make your own graph if you feel so bad about it, instead of letting other people do everything for you.
|
On July 03 2012 21:11 Assirra wrote:Show nested quote +On July 03 2012 20:59 MrSalamandra wrote:On July 03 2012 19:18 Assirra wrote:On July 03 2012 18:43 keglu wrote:On July 03 2012 18:40 Assirra wrote:On July 03 2012 18:34 keglu wrote:On July 03 2012 18:18 Assirra wrote:On July 03 2012 18:09 keglu wrote:On July 03 2012 17:49 Stiluz wrote: OP's own analysis about qualifiers is methodologically extremely flawed. If you want to count qualifiers to compare winrates with previous months, you have to go back and include tournament qualifying rounds for every single month. As another post mentioned, you can't cherry pick when to include and discuss winrates from qualifiers, you have to be consistant about where your data comes from. Or we can start now so for once we can have singifcant amount of data to jugde anything (500 games instead of 100). Who cares about previous data, you can't draw any conlusion on winrates changing over 10-20% each month. Also TSL, OSL qualifiers werent played before, MLG im not sure. And neither can you change the rules of something in the middle of series, your whole series is broken that way. Who knows what would have happened if the code A qualifiers for instance always got counted. It's kinda silly that people want to include qualifiers that were never included just to proof their point. Again i dont care about series, i care about current state of the game and having reliable data. Till know i never took Korean graph seriously because of sample size so let it be. We have chance to have reliable graph in future if we add all this data. It does not matter what you care about...a graph without consistency is not a graph at all. What you want is raw stats, not a graph. Ok lets have consistent graph showing nothing, let it be. Also to clarify: its ok to add new tournamet - proleague to graph but not ok to add new TSL/OSL qualifiers with better players and much more games? Well if that is the case why not just remove international graphics completely? I mean, its clearly all about the better players so. You cannot adjust the rules as you see fit, cause then this whole graph would be worthless. This is about tournaments, not qualifiers, doesn't matter how good the qualifiers are. Those qualifiers are tournaments. Then why were they never accounted before but suddenly people are screaming they should? a qualifier is not a tournament, it is a qualifier to get in the tournament.
Because it's relevant data and that the sample size is too small otherwise. These games have a big competitive meaning, so it's not like the players are fucking around like in ladder and the competition is way harder that actual tournaments that are already included in the graph.
Those games should have been accounted from the start. Following your logic we shouldnt even use code A games since it's only a qualifier for Code S.
|
On July 03 2012 21:11 Assirra wrote:Show nested quote +On July 03 2012 20:59 MrSalamandra wrote:On July 03 2012 19:18 Assirra wrote:On July 03 2012 18:43 keglu wrote:On July 03 2012 18:40 Assirra wrote:On July 03 2012 18:34 keglu wrote:On July 03 2012 18:18 Assirra wrote:On July 03 2012 18:09 keglu wrote:On July 03 2012 17:49 Stiluz wrote: OP's own analysis about qualifiers is methodologically extremely flawed. If you want to count qualifiers to compare winrates with previous months, you have to go back and include tournament qualifying rounds for every single month. As another post mentioned, you can't cherry pick when to include and discuss winrates from qualifiers, you have to be consistant about where your data comes from. Or we can start now so for once we can have singifcant amount of data to jugde anything (500 games instead of 100). Who cares about previous data, you can't draw any conlusion on winrates changing over 10-20% each month. Also TSL, OSL qualifiers werent played before, MLG im not sure. And neither can you change the rules of something in the middle of series, your whole series is broken that way. Who knows what would have happened if the code A qualifiers for instance always got counted. It's kinda silly that people want to include qualifiers that were never included just to proof their point. Again i dont care about series, i care about current state of the game and having reliable data. Till know i never took Korean graph seriously because of sample size so let it be. We have chance to have reliable graph in future if we add all this data. It does not matter what you care about...a graph without consistency is not a graph at all. What you want is raw stats, not a graph. Ok lets have consistent graph showing nothing, let it be. Also to clarify: its ok to add new tournamet - proleague to graph but not ok to add new TSL/OSL qualifiers with better players and much more games? Well if that is the case why not just remove international graphics completely? I mean, its clearly all about the better players so. You cannot adjust the rules as you see fit, cause then this whole graph would be worthless. This is about tournaments, not qualifiers, doesn't matter how good the qualifiers are. Those qualifiers are tournaments. Then why were they never accounted before but suddenly people are screaming they should? a qualifier is not a tournament, it is a qualifier to get in the tournament.
They were never accounted before because TLPD isn't perfect and what it does and doesn't include is a bit weird. It lists the OSL qualifier, but not the Code A qualifer, for example.
A qualifier is a tournament to get into another tournament.
http://en.wikipedia.org/wiki/Tournament
Really, do you think that the Code A qualifier isn't a tournament?
The Korean sample size is already pretty low, we don't need to be pointlessly omitting high level games because it's "just a qualifier".
|
On July 03 2012 21:14 emc wrote:Show nested quote +On July 03 2012 20:59 MrSalamandra wrote:On July 03 2012 19:18 Assirra wrote:On July 03 2012 18:43 keglu wrote:On July 03 2012 18:40 Assirra wrote:On July 03 2012 18:34 keglu wrote:On July 03 2012 18:18 Assirra wrote:On July 03 2012 18:09 keglu wrote:On July 03 2012 17:49 Stiluz wrote: OP's own analysis about qualifiers is methodologically extremely flawed. If you want to count qualifiers to compare winrates with previous months, you have to go back and include tournament qualifying rounds for every single month. As another post mentioned, you can't cherry pick when to include and discuss winrates from qualifiers, you have to be consistant about where your data comes from. Or we can start now so for once we can have singifcant amount of data to jugde anything (500 games instead of 100). Who cares about previous data, you can't draw any conlusion on winrates changing over 10-20% each month. Also TSL, OSL qualifiers werent played before, MLG im not sure. And neither can you change the rules of something in the middle of series, your whole series is broken that way. Who knows what would have happened if the code A qualifiers for instance always got counted. It's kinda silly that people want to include qualifiers that were never included just to proof their point. Again i dont care about series, i care about current state of the game and having reliable data. Till know i never took Korean graph seriously because of sample size so let it be. We have chance to have reliable graph in future if we add all this data. It does not matter what you care about...a graph without consistency is not a graph at all. What you want is raw stats, not a graph. Ok lets have consistent graph showing nothing, let it be. Also to clarify: its ok to add new tournamet - proleague to graph but not ok to add new TSL/OSL qualifiers with better players and much more games? Well if that is the case why not just remove international graphics completely? I mean, its clearly all about the better players so. You cannot adjust the rules as you see fit, cause then this whole graph would be worthless. This is about tournaments, not qualifiers, doesn't matter how good the qualifiers are. Those qualifiers are tournaments. yes, a tournament to a much larger tournament where every player is a super star. The qualifiers is weeding out the bad players from the good, if we haven't counted qualifiers in the past, why should we now? The qualifiers are like College football, everyone is pretty good but there is only 1 or 2 people on each football team who could be NFL potential, but in the major tournament, everyone is a super star. Qualifiers shouldn't be counted because there is too large of a skill gap, a qualifier could consist of bronze to GM players for all we know. Make your own graph if you feel so bad about it, instead of letting other people do everything for you.
Its like completly you have no idea what are you talking about. Korean weekly and proleague are tournaments with only superstars and TSL/ OSL/MLG qaulifiers are full of bad players. Wow...
Just do your research before posting please.
|
Any attempt to try to adjust the dataset when one doesn't find the expected result is outright attrocious. People do this over and over again in these threads. Of course it is wrong to include TSL4 qualifier results when the TLPD data shows a near 50/50 win ratio to try to get "better" results. It is completely wrong to use statistics in that way. So just stop this discussion already.
The TLPD statistics for june in Korea is flawed. The dataset for june is way too small to be statistically significant when the winratios are close to 50/50.
So both sides are wrong. In fact, there should be no "sides" in this discussion. You can't prove any statistical significant "imbalance" OR "balance" from these numbers. The only thing you can say is that terran won 50.9% = 58 of 114 games and that zerg won 49.1%=56 out of 114 games.
If we step out of the constraint that we have to use TLPD data only and add the data suggested in the OP we get a sample size of 335 games. As it turns out, 335 games is a way better sample size to work with. Now if, and that is a big IF, we assume that those 335 games can be expected to display the current state of TvZ in a complete way and then analyze the numbers - what happens then ?
I take the liberty to link back to one of my previous posts in this thread where I did just that.
But bare in mind that the assumptions that win rates will tell anything about balance is very questionable.
|
On July 03 2012 21:31 keglu wrote:Show nested quote +On July 03 2012 21:14 emc wrote:On July 03 2012 20:59 MrSalamandra wrote:On July 03 2012 19:18 Assirra wrote:On July 03 2012 18:43 keglu wrote:On July 03 2012 18:40 Assirra wrote:On July 03 2012 18:34 keglu wrote:On July 03 2012 18:18 Assirra wrote:On July 03 2012 18:09 keglu wrote:On July 03 2012 17:49 Stiluz wrote: OP's own analysis about qualifiers is methodologically extremely flawed. If you want to count qualifiers to compare winrates with previous months, you have to go back and include tournament qualifying rounds for every single month. As another post mentioned, you can't cherry pick when to include and discuss winrates from qualifiers, you have to be consistant about where your data comes from. Or we can start now so for once we can have singifcant amount of data to jugde anything (500 games instead of 100). Who cares about previous data, you can't draw any conlusion on winrates changing over 10-20% each month. Also TSL, OSL qualifiers werent played before, MLG im not sure. And neither can you change the rules of something in the middle of series, your whole series is broken that way. Who knows what would have happened if the code A qualifiers for instance always got counted. It's kinda silly that people want to include qualifiers that were never included just to proof their point. Again i dont care about series, i care about current state of the game and having reliable data. Till know i never took Korean graph seriously because of sample size so let it be. We have chance to have reliable graph in future if we add all this data. It does not matter what you care about...a graph without consistency is not a graph at all. What you want is raw stats, not a graph. Ok lets have consistent graph showing nothing, let it be. Also to clarify: its ok to add new tournamet - proleague to graph but not ok to add new TSL/OSL qualifiers with better players and much more games? Well if that is the case why not just remove international graphics completely? I mean, its clearly all about the better players so. You cannot adjust the rules as you see fit, cause then this whole graph would be worthless. This is about tournaments, not qualifiers, doesn't matter how good the qualifiers are. Those qualifiers are tournaments. yes, a tournament to a much larger tournament where every player is a super star. The qualifiers is weeding out the bad players from the good, if we haven't counted qualifiers in the past, why should we now? The qualifiers are like College football, everyone is pretty good but there is only 1 or 2 people on each football team who could be NFL potential, but in the major tournament, everyone is a super star. Qualifiers shouldn't be counted because there is too large of a skill gap, a qualifier could consist of bronze to GM players for all we know. Make your own graph if you feel so bad about it, instead of letting other people do everything for you. Its like completly you have no idea what are you talking about. Korean weekly and proleague are tournaments with only superstars and TSL/ OSL/MLG qaulifiers are full of bad players. Wow... Just do your research before posting please.
virtually anyone can sign up for Code A qualifiers, surely you must understand that there is a much bigger skill gap in a qualifier compared to an actual tournament? If we're looking for unbiased balanced, you have to take players who are closely equal in skill, qualifiers are the furthest from equality because the best players roll over everyone and make it to the real deal.
|
On July 03 2012 23:17 one-one-one wrote:Any attempt to try to adjust the dataset when one doesn't find the expected result is outright attrocious. People do this over and over again in these threads. Of course it is wrong to include TSL4 qualifier results when the TLPD data shows a near 50/50 win ratio to try to get "better" results. It is completely wrong to use statistics in that way. So just stop this discussion already. The TLPD statistics for june in Korea is flawed. The dataset for june is way too small to be statistically significant when the winratios are close to 50/50. So both sides are wrong. In fact, there should be no "sides" in this discussion. You can't prove any statistical significant "imbalance" OR "balance" from these numbers. The only thing you can say is that terran won 50.9% = 58 of 114 games and that zerg won 49.1%=56 out of 114 games. If we step out of the constraint that we have to use TLPD data only and add the data suggested in the OP we get a sample size of 335 games. As it turns out, 335 games is a way better sample size to work with. Now if, and that is a big IF, we assume that those 335 games can be expected to display the current state of TvZ in a complete way and then analyze the numbers - what happens then ? I take the liberty to link back to one of my previous posts in this thread where I did just that. But bare in mind that the assumptions that win rates will tell anything about balance is very questionable. Nah, we shouldn't care about previous graphs, bigger sample with the same players playing in a similar system is always better, code A qualifiers is one of the hardest tournaments in the world(if anything code A and S are the tournaments that fall into a completely different category given how much preparation for a specific opponent is important), period. I didn't even feel like arguing about this because it's downright asinine to even think about denying this bigger sample.
|
On July 03 2012 23:22 emc wrote:Show nested quote +On July 03 2012 21:31 keglu wrote:On July 03 2012 21:14 emc wrote:On July 03 2012 20:59 MrSalamandra wrote:On July 03 2012 19:18 Assirra wrote:On July 03 2012 18:43 keglu wrote:On July 03 2012 18:40 Assirra wrote:On July 03 2012 18:34 keglu wrote:On July 03 2012 18:18 Assirra wrote:On July 03 2012 18:09 keglu wrote: [quote]
Or we can start now so for once we can have singifcant amount of data to jugde anything (500 games instead of 100). Who cares about previous data, you can't draw any conlusion on winrates changing over 10-20% each month. Also TSL, OSL qualifiers werent played before, MLG im not sure.
And neither can you change the rules of something in the middle of series, your whole series is broken that way. Who knows what would have happened if the code A qualifiers for instance always got counted. It's kinda silly that people want to include qualifiers that were never included just to proof their point. Again i dont care about series, i care about current state of the game and having reliable data. Till know i never took Korean graph seriously because of sample size so let it be. We have chance to have reliable graph in future if we add all this data. It does not matter what you care about...a graph without consistency is not a graph at all. What you want is raw stats, not a graph. Ok lets have consistent graph showing nothing, let it be. Also to clarify: its ok to add new tournamet - proleague to graph but not ok to add new TSL/OSL qualifiers with better players and much more games? Well if that is the case why not just remove international graphics completely? I mean, its clearly all about the better players so. You cannot adjust the rules as you see fit, cause then this whole graph would be worthless. This is about tournaments, not qualifiers, doesn't matter how good the qualifiers are. Those qualifiers are tournaments. yes, a tournament to a much larger tournament where every player is a super star. The qualifiers is weeding out the bad players from the good, if we haven't counted qualifiers in the past, why should we now? The qualifiers are like College football, everyone is pretty good but there is only 1 or 2 people on each football team who could be NFL potential, but in the major tournament, everyone is a super star. Qualifiers shouldn't be counted because there is too large of a skill gap, a qualifier could consist of bronze to GM players for all we know. Make your own graph if you feel so bad about it, instead of letting other people do everything for you. Its like completly you have no idea what are you talking about. Korean weekly and proleague are tournaments with only superstars and TSL/ OSL/MLG qaulifiers are full of bad players. Wow... Just do your research before posting please. virtually anyone can sign up for Code A qualifiers, surely you must understand that there is a much bigger skill gap in a qualifier compared to an actual tournament? If we're looking for unbiased balanced, you have to take players who are closely equal in skill, qualifiers are the furthest from equality because the best players roll over everyone and make it to the real deal. Stop spreading bullshit, you need to be at least in code B, i.e stronger than all of the foreign only tournaments.
|
On July 03 2012 23:29 IshinShishi wrote:Show nested quote +On July 03 2012 23:22 emc wrote:On July 03 2012 21:31 keglu wrote:On July 03 2012 21:14 emc wrote:On July 03 2012 20:59 MrSalamandra wrote:On July 03 2012 19:18 Assirra wrote:On July 03 2012 18:43 keglu wrote:On July 03 2012 18:40 Assirra wrote:On July 03 2012 18:34 keglu wrote:On July 03 2012 18:18 Assirra wrote: [quote] And neither can you change the rules of something in the middle of series, your whole series is broken that way. Who knows what would have happened if the code A qualifiers for instance always got counted. It's kinda silly that people want to include qualifiers that were never included just to proof their point. Again i dont care about series, i care about current state of the game and having reliable data. Till know i never took Korean graph seriously because of sample size so let it be. We have chance to have reliable graph in future if we add all this data. It does not matter what you care about...a graph without consistency is not a graph at all. What you want is raw stats, not a graph. Ok lets have consistent graph showing nothing, let it be. Also to clarify: its ok to add new tournamet - proleague to graph but not ok to add new TSL/OSL qualifiers with better players and much more games? Well if that is the case why not just remove international graphics completely? I mean, its clearly all about the better players so. You cannot adjust the rules as you see fit, cause then this whole graph would be worthless. This is about tournaments, not qualifiers, doesn't matter how good the qualifiers are. Those qualifiers are tournaments. yes, a tournament to a much larger tournament where every player is a super star. The qualifiers is weeding out the bad players from the good, if we haven't counted qualifiers in the past, why should we now? The qualifiers are like College football, everyone is pretty good but there is only 1 or 2 people on each football team who could be NFL potential, but in the major tournament, everyone is a super star. Qualifiers shouldn't be counted because there is too large of a skill gap, a qualifier could consist of bronze to GM players for all we know. Make your own graph if you feel so bad about it, instead of letting other people do everything for you. Its like completly you have no idea what are you talking about. Korean weekly and proleague are tournaments with only superstars and TSL/ OSL/MLG qaulifiers are full of bad players. Wow... Just do your research before posting please. virtually anyone can sign up for Code A qualifiers, surely you must understand that there is a much bigger skill gap in a qualifier compared to an actual tournament? If we're looking for unbiased balanced, you have to take players who are closely equal in skill, qualifiers are the furthest from equality because the best players roll over everyone and make it to the real deal. Stop spreading bullshit, you need to be at least in code B, i.e stronger than all of the foreign only tournaments.
Isn't Code B a ladder rank ? If it is then probably anyone that is gm in EU/NA can get into code B.
|
On July 03 2012 23:22 emc wrote:Show nested quote +On July 03 2012 21:31 keglu wrote:On July 03 2012 21:14 emc wrote:On July 03 2012 20:59 MrSalamandra wrote:On July 03 2012 19:18 Assirra wrote:On July 03 2012 18:43 keglu wrote:On July 03 2012 18:40 Assirra wrote:On July 03 2012 18:34 keglu wrote:On July 03 2012 18:18 Assirra wrote:On July 03 2012 18:09 keglu wrote: [quote]
Or we can start now so for once we can have singifcant amount of data to jugde anything (500 games instead of 100). Who cares about previous data, you can't draw any conlusion on winrates changing over 10-20% each month. Also TSL, OSL qualifiers werent played before, MLG im not sure.
And neither can you change the rules of something in the middle of series, your whole series is broken that way. Who knows what would have happened if the code A qualifiers for instance always got counted. It's kinda silly that people want to include qualifiers that were never included just to proof their point. Again i dont care about series, i care about current state of the game and having reliable data. Till know i never took Korean graph seriously because of sample size so let it be. We have chance to have reliable graph in future if we add all this data. It does not matter what you care about...a graph without consistency is not a graph at all. What you want is raw stats, not a graph. Ok lets have consistent graph showing nothing, let it be. Also to clarify: its ok to add new tournamet - proleague to graph but not ok to add new TSL/OSL qualifiers with better players and much more games? Well if that is the case why not just remove international graphics completely? I mean, its clearly all about the better players so. You cannot adjust the rules as you see fit, cause then this whole graph would be worthless. This is about tournaments, not qualifiers, doesn't matter how good the qualifiers are. Those qualifiers are tournaments. yes, a tournament to a much larger tournament where every player is a super star. The qualifiers is weeding out the bad players from the good, if we haven't counted qualifiers in the past, why should we now? The qualifiers are like College football, everyone is pretty good but there is only 1 or 2 people on each football team who could be NFL potential, but in the major tournament, everyone is a super star. Qualifiers shouldn't be counted because there is too large of a skill gap, a qualifier could consist of bronze to GM players for all we know. Make your own graph if you feel so bad about it, instead of letting other people do everything for you. Its like completly you have no idea what are you talking about. Korean weekly and proleague are tournaments with only superstars and TSL/ OSL/MLG qaulifiers are full of bad players. Wow... Just do your research before posting please. virtually anyone can sign up for Code A qualifiers, surely you must understand that there is a much bigger skill gap in a qualifier compared to an actual tournament? If we're looking for unbiased balanced, you have to take players who are closely equal in skill, qualifiers are the furthest from equality because the best players roll over everyone and make it to the real deal.
Point is to have more data to reduce random elements like palyers skill. Right know we have Korean graph with 20% of games actual played in Korea. At this point its better drop Korean graph altogether to reduce confusion.
|
On July 03 2012 23:29 IshinShishi wrote:Show nested quote +On July 03 2012 23:22 emc wrote:On July 03 2012 21:31 keglu wrote:On July 03 2012 21:14 emc wrote:On July 03 2012 20:59 MrSalamandra wrote:On July 03 2012 19:18 Assirra wrote:On July 03 2012 18:43 keglu wrote:On July 03 2012 18:40 Assirra wrote:On July 03 2012 18:34 keglu wrote:On July 03 2012 18:18 Assirra wrote: [quote] And neither can you change the rules of something in the middle of series, your whole series is broken that way. Who knows what would have happened if the code A qualifiers for instance always got counted. It's kinda silly that people want to include qualifiers that were never included just to proof their point. Again i dont care about series, i care about current state of the game and having reliable data. Till know i never took Korean graph seriously because of sample size so let it be. We have chance to have reliable graph in future if we add all this data. It does not matter what you care about...a graph without consistency is not a graph at all. What you want is raw stats, not a graph. Ok lets have consistent graph showing nothing, let it be. Also to clarify: its ok to add new tournamet - proleague to graph but not ok to add new TSL/OSL qualifiers with better players and much more games? Well if that is the case why not just remove international graphics completely? I mean, its clearly all about the better players so. You cannot adjust the rules as you see fit, cause then this whole graph would be worthless. This is about tournaments, not qualifiers, doesn't matter how good the qualifiers are. Those qualifiers are tournaments. yes, a tournament to a much larger tournament where every player is a super star. The qualifiers is weeding out the bad players from the good, if we haven't counted qualifiers in the past, why should we now? The qualifiers are like College football, everyone is pretty good but there is only 1 or 2 people on each football team who could be NFL potential, but in the major tournament, everyone is a super star. Qualifiers shouldn't be counted because there is too large of a skill gap, a qualifier could consist of bronze to GM players for all we know. Make your own graph if you feel so bad about it, instead of letting other people do everything for you. Its like completly you have no idea what are you talking about. Korean weekly and proleague are tournaments with only superstars and TSL/ OSL/MLG qaulifiers are full of bad players. Wow... Just do your research before posting please. virtually anyone can sign up for Code A qualifiers, surely you must understand that there is a much bigger skill gap in a qualifier compared to an actual tournament? If we're looking for unbiased balanced, you have to take players who are closely equal in skill, qualifiers are the furthest from equality because the best players roll over everyone and make it to the real deal. Stop spreading bullshit, you need to be at least in code B, i.e stronger than all of the foreign only tournaments.
Participation Requirements:
- Must be at least 12 years old - Need to have a SC2 Battle.net account (cannot use your family's or friends' accounts) [Does not have to be a Korean Battle.net account] - Identification (Passport, ID, driver's license)
These are the requirements to participate in the 2012 GSL Season 3 Code A Qualifiers. Source: GomTV
|
On July 03 2012 23:39 Jebediah wrote:Show nested quote +On July 03 2012 23:29 IshinShishi wrote:On July 03 2012 23:22 emc wrote:On July 03 2012 21:31 keglu wrote:On July 03 2012 21:14 emc wrote:On July 03 2012 20:59 MrSalamandra wrote:On July 03 2012 19:18 Assirra wrote:On July 03 2012 18:43 keglu wrote:On July 03 2012 18:40 Assirra wrote:On July 03 2012 18:34 keglu wrote: [quote]
Again i dont care about series, i care about current state of the game and having reliable data. Till know i never took Korean graph seriously because of sample size so let it be. We have chance to have reliable graph in future if we add all this data.
It does not matter what you care about...a graph without consistency is not a graph at all. What you want is raw stats, not a graph. Ok lets have consistent graph showing nothing, let it be. Also to clarify: its ok to add new tournamet - proleague to graph but not ok to add new TSL/OSL qualifiers with better players and much more games? Well if that is the case why not just remove international graphics completely? I mean, its clearly all about the better players so. You cannot adjust the rules as you see fit, cause then this whole graph would be worthless. This is about tournaments, not qualifiers, doesn't matter how good the qualifiers are. Those qualifiers are tournaments. yes, a tournament to a much larger tournament where every player is a super star. The qualifiers is weeding out the bad players from the good, if we haven't counted qualifiers in the past, why should we now? The qualifiers are like College football, everyone is pretty good but there is only 1 or 2 people on each football team who could be NFL potential, but in the major tournament, everyone is a super star. Qualifiers shouldn't be counted because there is too large of a skill gap, a qualifier could consist of bronze to GM players for all we know. Make your own graph if you feel so bad about it, instead of letting other people do everything for you. Its like completly you have no idea what are you talking about. Korean weekly and proleague are tournaments with only superstars and TSL/ OSL/MLG qaulifiers are full of bad players. Wow... Just do your research before posting please. virtually anyone can sign up for Code A qualifiers, surely you must understand that there is a much bigger skill gap in a qualifier compared to an actual tournament? If we're looking for unbiased balanced, you have to take players who are closely equal in skill, qualifiers are the furthest from equality because the best players roll over everyone and make it to the real deal. Stop spreading bullshit, you need to be at least in code B, i.e stronger than all of the foreign only tournaments. Participation Requirements: - Must be at least 12 years old - Need to have a SC2 Battle.net account (cannot use your family's or friends' accounts) [Does not have to be a Korean Battle.net account] - Identification (Passport, ID, driver's license) These are the requirements to participate in the 2012 GSL Season 3 Code A Qualifiers. Source: GomTV ?Is that supposed to mean anything?I'm sure that the tournament would be full of platinum and diamond players if those were the only requirements, not full of players in sponsored teams that don't always qualify, this is completely meaningless as it does not represent the reality by any means, never did.
|
|
|
|