|
How are the toss winrates actually going DOWN?
|
Everybody should notice that october has only 810 games (which is an all time low). This is a lot less games than the average (2200) of games.
|
On November 07 2011 05:55 doko100 wrote:Show nested quote +On November 07 2011 05:49 InvalidID wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. I agree that 52-48% is pretty balanced, and 54-46 is not bad, but your analogy is not great. The graphs show the standard deviation, which demonstrates the expected bounds of random variation. They include a large number of players of differing skills. If you can make the assumption that the skill distribution is equal among the races at the highest level, then they are valid. The graphs simply ignore individual skill, these numbers will never be valid as long as you have a majority of koreans who come to foreign events be terran and as long as they record games of semi-'pros' and SEA players against Koreans. The graphs don't show any imbalances, but they also don't prove anything because I personally was shocked when I checked the games from last month, there were players playing koreans that I've never even heard of before, SEA or really bad NA/EU semi-pros who lost to players like MKP or MMA. the fact that these matches are actually taken into account completely de-legitimizes this graph. random sea player X losing to MKP has nothing to do with imbalance. Show nested quote +On November 07 2011 05:53 Akhee wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. you re wrong, 56-43 is a very imbalanced scenario It's borderline, but then again, these stats ignore individual skills. If I send 100 korean zergs and terrans to a tournament full of NA/EU protosses and the koreans win 90% of their matches does that really prove anything? And this is partially what happened.
Plenty of Koreans lost to NA and EU semi pros this month (DRG and TOP to Gatored), several other sets in the MLG open tourney. Again, as long as the the comparative skill level of the players playing the different races is balanced, the stats will be valid. It doesn't effect it unless, like in your example, all of the Koreans are zergs. They have been pretty evenly balanced in race from the tourneys I have seen.
|
On November 07 2011 05:59 Lore-Fighting wrote:Show nested quote +On November 07 2011 05:55 doko100 wrote:On November 07 2011 05:49 InvalidID wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. I agree that 52-48% is pretty balanced, and 54-46 is not bad, but your analogy is not great. The graphs show the standard deviation, which demonstrates the expected bounds of random variation. They include a large number of players of differing skills. If you can make the assumption that the skill distribution is equal among the races at the highest level, then they are valid. The graphs simply ignore individual skill, these numbers will never be valid as long as you have a majority of koreans who come to foreign events be terran and as long as they record games of semi-'pros' and SEA players against Koreans. The graphs don't show any imbalances, but they also don't prove anything because I personally was shocked when I checked the games from last month, there were players playing koreans that I've never even heard of before, SEA or really bad NA/EU semi-pros who lost to players like MKP or MMA. the fact that these matches are actually taken into account completely de-legitimizes this graph. random sea player X losing to MKP has nothing to do with imbalance. On November 07 2011 05:53 Akhee wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. you re wrong, 56-43 is a very imbalanced scenario It's borderline, but then again, these stats ignore individual skills. If I send 100 korean zergs and terrans to a tournament full of NA/EU protosses and the koreans win 90% of their matches does that really prove anything? And this is partially what happened. Wonder why the korean s aren't playing protoss.....
they are,there are quite alot of them in code a and due to the new format there will be quite alot of them in code s in the future.
people need to be more patient and realize that these graphs are pointless
On November 07 2011 06:01 InvalidID wrote:Show nested quote +On November 07 2011 05:55 doko100 wrote:On November 07 2011 05:49 InvalidID wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. I agree that 52-48% is pretty balanced, and 54-46 is not bad, but your analogy is not great. The graphs show the standard deviation, which demonstrates the expected bounds of random variation. They include a large number of players of differing skills. If you can make the assumption that the skill distribution is equal among the races at the highest level, then they are valid. The graphs simply ignore individual skill, these numbers will never be valid as long as you have a majority of koreans who come to foreign events be terran and as long as they record games of semi-'pros' and SEA players against Koreans. The graphs don't show any imbalances, but they also don't prove anything because I personally was shocked when I checked the games from last month, there were players playing koreans that I've never even heard of before, SEA or really bad NA/EU semi-pros who lost to players like MKP or MMA. the fact that these matches are actually taken into account completely de-legitimizes this graph. random sea player X losing to MKP has nothing to do with imbalance. On November 07 2011 05:53 Akhee wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. you re wrong, 56-43 is a very imbalanced scenario It's borderline, but then again, these stats ignore individual skills. If I send 100 korean zergs and terrans to a tournament full of NA/EU protosses and the koreans win 90% of their matches does that really prove anything? And this is partially what happened. Plenty of Koreans lost to NA and EU semi pros this month (DRG and TOP to Gatored), several other sets in the MLG open tourney. Again, as long as the the comparative skill level of the players playing the different races is balanced, the stats will be valid. It doesn't effect it unless, like in your example, all of the Koreans are zergs. They have been pretty evenly balanced in race from the tourneys I have seen.
so you pick 2 games where koreans lost but ignore the more than 100 games that are recorded in this where koreans beat foreigners? are you serious?
|
sry double post mod pls delete
|
On November 07 2011 06:04 doko100 wrote:Show nested quote +On November 07 2011 06:01 InvalidID wrote:On November 07 2011 05:55 doko100 wrote:On November 07 2011 05:49 InvalidID wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. I agree that 52-48% is pretty balanced, and 54-46 is not bad, but your analogy is not great. The graphs show the standard deviation, which demonstrates the expected bounds of random variation. They include a large number of players of differing skills. If you can make the assumption that the skill distribution is equal among the races at the highest level, then they are valid. The graphs simply ignore individual skill, these numbers will never be valid as long as you have a majority of koreans who come to foreign events be terran and as long as they record games of semi-'pros' and SEA players against Koreans. The graphs don't show any imbalances, but they also don't prove anything because I personally was shocked when I checked the games from last month, there were players playing koreans that I've never even heard of before, SEA or really bad NA/EU semi-pros who lost to players like MKP or MMA. the fact that these matches are actually taken into account completely de-legitimizes this graph. random sea player X losing to MKP has nothing to do with imbalance. On November 07 2011 05:53 Akhee wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. you re wrong, 56-43 is a very imbalanced scenario It's borderline, but then again, these stats ignore individual skills. If I send 100 korean zergs and terrans to a tournament full of NA/EU protosses and the koreans win 90% of their matches does that really prove anything? And this is partially what happened. Plenty of Koreans lost to NA and EU semi pros this month (DRG and TOP to Gatored), several other sets in the MLG open tourney. Again, as long as the the comparative skill level of the players playing the different races is balanced, the stats will be valid. It doesn't effect it unless, like in your example, all of the Koreans are zergs. They have been pretty evenly balanced in race from the tourneys I have seen. so you pick 2 games where koreans lost but ignore the more than 100 games that are recorded in this where koreans beat foreigners? are you serious?
You completely missed my point: it doesn't matter if the Koreans beat the foreigners, if the Koreans are balanced in their race. The situation would only be a problem if Koreans of only one race were coming over.
|
Those graphs are major bullshit, since they only go form 40-60%, which makes the differences seem much more harsh then they actually are. This is just motivation for stupid people to continue whining.
|
On November 07 2011 05:44 doko100 wrote:Show nested quote +On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill.
I agree that the previous commentors analogy is incorrect. If you flip a coin 100 times, you may have 54-46 heads vs. tails. But if you flip that coin (assuming is evenly weighted) a thousand times, you will not get 540-460. If you do, the coin is imbalanced.
|
On November 07 2011 06:02 doko100 wrote:Show nested quote +On November 07 2011 05:59 Lore-Fighting wrote:On November 07 2011 05:55 doko100 wrote:On November 07 2011 05:49 InvalidID wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. I agree that 52-48% is pretty balanced, and 54-46 is not bad, but your analogy is not great. The graphs show the standard deviation, which demonstrates the expected bounds of random variation. They include a large number of players of differing skills. If you can make the assumption that the skill distribution is equal among the races at the highest level, then they are valid. The graphs simply ignore individual skill, these numbers will never be valid as long as you have a majority of koreans who come to foreign events be terran and as long as they record games of semi-'pros' and SEA players against Koreans. The graphs don't show any imbalances, but they also don't prove anything because I personally was shocked when I checked the games from last month, there were players playing koreans that I've never even heard of before, SEA or really bad NA/EU semi-pros who lost to players like MKP or MMA. the fact that these matches are actually taken into account completely de-legitimizes this graph. random sea player X losing to MKP has nothing to do with imbalance. On November 07 2011 05:53 Akhee wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. you re wrong, 56-43 is a very imbalanced scenario It's borderline, but then again, these stats ignore individual skills. If I send 100 korean zergs and terrans to a tournament full of NA/EU protosses and the koreans win 90% of their matches does that really prove anything? And this is partially what happened. Wonder why the korean s aren't playing protoss..... they are,there are quite alot of them in code a and due to the new format there will be quite alot of them in code s in the future. people need to be more patient and realize that these graphs are pointless Show nested quote +On November 07 2011 06:01 InvalidID wrote:On November 07 2011 05:55 doko100 wrote:On November 07 2011 05:49 InvalidID wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. I agree that 52-48% is pretty balanced, and 54-46 is not bad, but your analogy is not great. The graphs show the standard deviation, which demonstrates the expected bounds of random variation. They include a large number of players of differing skills. If you can make the assumption that the skill distribution is equal among the races at the highest level, then they are valid. The graphs simply ignore individual skill, these numbers will never be valid as long as you have a majority of koreans who come to foreign events be terran and as long as they record games of semi-'pros' and SEA players against Koreans. The graphs don't show any imbalances, but they also don't prove anything because I personally was shocked when I checked the games from last month, there were players playing koreans that I've never even heard of before, SEA or really bad NA/EU semi-pros who lost to players like MKP or MMA. the fact that these matches are actually taken into account completely de-legitimizes this graph. random sea player X losing to MKP has nothing to do with imbalance. On November 07 2011 05:53 Akhee wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. you re wrong, 56-43 is a very imbalanced scenario It's borderline, but then again, these stats ignore individual skills. If I send 100 korean zergs and terrans to a tournament full of NA/EU protosses and the koreans win 90% of their matches does that really prove anything? And this is partially what happened. Plenty of Koreans lost to NA and EU semi pros this month (DRG and TOP to Gatored), several other sets in the MLG open tourney. Again, as long as the the comparative skill level of the players playing the different races is balanced, the stats will be valid. It doesn't effect it unless, like in your example, all of the Koreans are zergs. They have been pretty evenly balanced in race from the tourneys I have seen. so you pick 2 games where koreans lost but ignore the more than 100 games that are recorded in this where koreans beat foreigners? are you serious?
omg doko, just stop, you re completely wrong.
protoss in korea is doing bad too, believe me, even worse than foreigners, look that http://i.imgur.com/w8nXZ.png, you think its balance? and graphs REALLY say something, stop discussing it doesn't, omg...
|
On November 07 2011 06:05 Grapefruit wrote:
Those graphs are major bullshit, since they only go form 40-60%, which makes the differences seem much more harsh then they actually are. This is just motivation for stupid people to continue whining.
I appreciate how you have read this thread so carefully. Everyone else is obviously too stupid to notice how the graph is displayed. What would we do without you?
All protoss players are obviously too stupid to practice hard or try to win also? I mean a 40% win ratio is something everyone should love and enjoy?
Go on. Do tell more.
|
On November 07 2011 06:05 Grapefruit wrote:
Those graphs are major bullshit, since they only go form 40-60%, which makes the differences seem much more harsh then they actually are. This is just motivation for stupid people to continue whining.
The graphs are perfectly fine. They make it actually mean something when you look at it and actually read the sides, instead of derping around. Also if you didn't know, 40-60 is fine, because if one race is winning 60% against another, their win ratio is 1.5:1 which is already an absolutely massive difference.
|
On November 07 2011 06:02 doko100 wrote:Show nested quote +On November 07 2011 05:59 Lore-Fighting wrote:On November 07 2011 05:55 doko100 wrote:On November 07 2011 05:49 InvalidID wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. I agree that 52-48% is pretty balanced, and 54-46 is not bad, but your analogy is not great. The graphs show the standard deviation, which demonstrates the expected bounds of random variation. They include a large number of players of differing skills. If you can make the assumption that the skill distribution is equal among the races at the highest level, then they are valid. The graphs simply ignore individual skill, these numbers will never be valid as long as you have a majority of koreans who come to foreign events be terran and as long as they record games of semi-'pros' and SEA players against Koreans. The graphs don't show any imbalances, but they also don't prove anything because I personally was shocked when I checked the games from last month, there were players playing koreans that I've never even heard of before, SEA or really bad NA/EU semi-pros who lost to players like MKP or MMA. the fact that these matches are actually taken into account completely de-legitimizes this graph. random sea player X losing to MKP has nothing to do with imbalance. On November 07 2011 05:53 Akhee wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. you re wrong, 56-43 is a very imbalanced scenario It's borderline, but then again, these stats ignore individual skills. If I send 100 korean zergs and terrans to a tournament full of NA/EU protosses and the koreans win 90% of their matches does that really prove anything? And this is partially what happened. Wonder why the korean s aren't playing protoss..... they are,there are quite alot of them in code a and due to the new format there will be quite alot of them in code s in the future. people need to be more patient and realize that these graphs are pointless I'm sorry, your right. Statistics are pointless. Balance should be based on opinion. Show nested quote +On November 07 2011 06:01 InvalidID wrote:On November 07 2011 05:55 doko100 wrote:On November 07 2011 05:49 InvalidID wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. I agree that 52-48% is pretty balanced, and 54-46 is not bad, but your analogy is not great. The graphs show the standard deviation, which demonstrates the expected bounds of random variation. They include a large number of players of differing skills. If you can make the assumption that the skill distribution is equal among the races at the highest level, then they are valid. The graphs simply ignore individual skill, these numbers will never be valid as long as you have a majority of koreans who come to foreign events be terran and as long as they record games of semi-'pros' and SEA players against Koreans. The graphs don't show any imbalances, but they also don't prove anything because I personally was shocked when I checked the games from last month, there were players playing koreans that I've never even heard of before, SEA or really bad NA/EU semi-pros who lost to players like MKP or MMA. the fact that these matches are actually taken into account completely de-legitimizes this graph. random sea player X losing to MKP has nothing to do with imbalance. On November 07 2011 05:53 Akhee wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. you re wrong, 56-43 is a very imbalanced scenario It's borderline, but then again, these stats ignore individual skills. If I send 100 korean zergs and terrans to a tournament full of NA/EU protosses and the koreans win 90% of their matches does that really prove anything? And this is partially what happened. Plenty of Koreans lost to NA and EU semi pros this month (DRG and TOP to Gatored), several other sets in the MLG open tourney. Again, as long as the the comparative skill level of the players playing the different races is balanced, the stats will be valid. It doesn't effect it unless, like in your example, all of the Koreans are zergs. They have been pretty evenly balanced in race from the tourneys I have seen. so you pick 2 games where koreans lost but ignore the more than 100 games that are recorded in this where koreans beat foreigners? are you serious?
|
On November 07 2011 06:06 Lore-Fighting wrote:Show nested quote +On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. I agree that the previous commentors analogy is incorrect. If you flip a coin 100 times, you may have 54-46 heads vs. tails. But if you flip that coin (assuming is evenly weighted) a thousand times, you will not get 540-460. If you do, the coin is imbalanced.
That's just not true. It's random and a terrible analogy. In theory you could get 1000-0 and it doesn't mean it's imbalanced.... it's just random. you have a 50:50 chance but in theory you could always get one side and it doesnt say anything about balance. sc2 is skill based, coin flipping really isn't.
P.S: and to the guy who said that all 3 korean races are evenly distributed in foreign tournaments, they are not there are alot more terrans coming over, thus giving terran automatically a high win rate in every tournament they participate in.
|
On November 07 2011 06:15 doko100 wrote:Show nested quote +On November 07 2011 06:06 Lore-Fighting wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. I agree that the previous commentors analogy is incorrect. If you flip a coin 100 times, you may have 54-46 heads vs. tails. But if you flip that coin (assuming is evenly weighted) a thousand times, you will not get 540-460. If you do, the coin is imbalanced. That's just not true. It's random and a terrible analogy. In theory you could get 1000-0 and it doesn't mean it's imbalanced.... it's just random. you have a 50:50 chance but in theory you could always get one side and it doesnt say anything about balance. sc2 is skill based, coin flipping really isn't. P.S: and to the guy who said that all 3 korean races are evenly distributed in foreign tournaments, they are not there are alot more terrans coming over, thus giving terran automatically a high win rate in every tournament they participate in.
Ah so this is why Terran has been 5% higher in winrate on average the last 12 months. You solved the problem I would message Blizzard asap!
|
On November 07 2011 06:08 JonnyLaw wrote:Show nested quote +On November 07 2011 06:05 Grapefruit wrote:
Those graphs are major bullshit, since they only go form 40-60%, which makes the differences seem much more harsh then they actually are. This is just motivation for stupid people to continue whining.
I appreciate how you have read this thread so carefully. Everyone else is obviously too stupid to notice how the graph is displayed. What would we do without you? All protoss players are obviously too stupid to practice hard or try to win also? I mean a 40% win ratio is something everyone should love and enjoy? Go on. Do tell more.
but remember last months' outcry about how terran is so much ahead of zerg, and now after updates of few hundred games winrates between T and Z for sept is actually extremely close?
They're basically arguing a few percentage differences that falls within the standard deviation. You will be surprised at how few people can perceive graphs and scales. Business do the same thing all the time to trick their consumers, and it works
also 5% higher is really nothing. Do you know in Chess white have a 55% winrates to black's 45%? I don't think many people would argue that game is imbalance
|
On November 07 2011 06:15 doko100 wrote:Show nested quote +On November 07 2011 06:06 Lore-Fighting wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. I agree that the previous commentors analogy is incorrect. If you flip a coin 100 times, you may have 54-46 heads vs. tails. But if you flip that coin (assuming is evenly weighted) a thousand times, you will not get 540-460. If you do, the coin is imbalanced. That's just not true. It's random and a terrible analogy. In theory you could get 1000-0 and it doesn't mean it's imbalanced.... it's just random. you have a 50:50 chance but in theory you could always get one side and it doesnt say anything about balance. sc2 is skill based, coin flipping really isn't. P.S: and to the guy who said that all 3 korean races are evenly distributed in foreign tournaments, they are not there are alot more terrans coming over, thus giving terran automatically a high win rate in every tournament they participate in.
You just seem to try discredit every single statistical reference in this thread, as well as break up any reasonable discussion over the interpretation of statistics. Why are you in a thread that's meant to be for discussing about those numbers in the first place?
Protoss just keeps trending down, and that's with a long period of not rising above 50% in all matchups too. That is a huge cause of concern for anybody who actually cares about interpreting the numbers.
|
On November 07 2011 06:11 Lore-Fighting wrote:Show nested quote +On November 07 2011 06:02 doko100 wrote:On November 07 2011 05:59 Lore-Fighting wrote:On November 07 2011 05:55 doko100 wrote:On November 07 2011 05:49 InvalidID wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. I agree that 52-48% is pretty balanced, and 54-46 is not bad, but your analogy is not great. The graphs show the standard deviation, which demonstrates the expected bounds of random variation. They include a large number of players of differing skills. If you can make the assumption that the skill distribution is equal among the races at the highest level, then they are valid. The graphs simply ignore individual skill, these numbers will never be valid as long as you have a majority of koreans who come to foreign events be terran and as long as they record games of semi-'pros' and SEA players against Koreans. The graphs don't show any imbalances, but they also don't prove anything because I personally was shocked when I checked the games from last month, there were players playing koreans that I've never even heard of before, SEA or really bad NA/EU semi-pros who lost to players like MKP or MMA. the fact that these matches are actually taken into account completely de-legitimizes this graph. random sea player X losing to MKP has nothing to do with imbalance. On November 07 2011 05:53 Akhee wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. you re wrong, 56-43 is a very imbalanced scenario It's borderline, but then again, these stats ignore individual skills. If I send 100 korean zergs and terrans to a tournament full of NA/EU protosses and the koreans win 90% of their matches does that really prove anything? And this is partially what happened. Wonder why the korean s aren't playing protoss..... they are,there are quite alot of them in code a and due to the new format there will be quite alot of them in code s in the future. people need to be more patient and realize that these graphs are pointless I'm sorry, your right. Statistics are pointless. Balance should be based on opinion. On November 07 2011 06:01 InvalidID wrote:On November 07 2011 05:55 doko100 wrote:On November 07 2011 05:49 InvalidID wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. I agree that 52-48% is pretty balanced, and 54-46 is not bad, but your analogy is not great. The graphs show the standard deviation, which demonstrates the expected bounds of random variation. They include a large number of players of differing skills. If you can make the assumption that the skill distribution is equal among the races at the highest level, then they are valid. The graphs simply ignore individual skill, these numbers will never be valid as long as you have a majority of koreans who come to foreign events be terran and as long as they record games of semi-'pros' and SEA players against Koreans. The graphs don't show any imbalances, but they also don't prove anything because I personally was shocked when I checked the games from last month, there were players playing koreans that I've never even heard of before, SEA or really bad NA/EU semi-pros who lost to players like MKP or MMA. the fact that these matches are actually taken into account completely de-legitimizes this graph. random sea player X losing to MKP has nothing to do with imbalance. On November 07 2011 05:53 Akhee wrote:On November 07 2011 05:44 doko100 wrote:On November 07 2011 05:39 InvalidID wrote:On November 07 2011 05:18 doko100 wrote:Look at the overall win rate. In january terran has a 53,9% win rate but the curve is somewhere at around 55% and in july terran has a 53.8% win rate but the curve is at 52%. It makes this graph useless because when you have a first look just the line that is drawn makes it look alot more imbalanced than it is.. explain this to me please? On November 07 2011 05:16 Hetz wrote:
That top graph...why is the blue line always above the others? yeah it's bullshit, when zerg had a higher overall win rate than terran in june the line still showed an almost 3% advantage for terran. like I said soooo misleading The graph explains exactly what the trend line represents(moving average), and provides its scale precisely. The creator cannot help if you do not understand the information it provides. It is not misleading in any way. The scale is pretty normal(slightly larger then the biggest deviation from 50%), and provides its numbers. The 3 month moving average smooths out the random fluctuations to give you a better idea of the trend, hence the "trend line". I do understand the graphs very well, the problem is that most other people interpret this as a proof for major imbalance when there really is none. 52%-48% is not imbalanced, that's about as balanced as it gets, neither is 46%-54% that is super balanced aswell. Even if I play 100 games against an equally skilled player on my level there is no way we both would win 50 games each. it would be something like 54-46. Out of 100 games that is only 4 games 1 player won more than the other one did, that is NOTHING and doesnt prove anything. especially since the stats completely ignore individual skill. you re wrong, 56-43 is a very imbalanced scenario It's borderline, but then again, these stats ignore individual skills. If I send 100 korean zergs and terrans to a tournament full of NA/EU protosses and the koreans win 90% of their matches does that really prove anything? And this is partially what happened. Plenty of Koreans lost to NA and EU semi pros this month (DRG and TOP to Gatored), several other sets in the MLG open tourney. Again, as long as the the comparative skill level of the players playing the different races is balanced, the stats will be valid. It doesn't effect it unless, like in your example, all of the Koreans are zergs. They have been pretty evenly balanced in race from the tourneys I have seen. so you pick 2 games where koreans lost but ignore the more than 100 games that are recorded in this where koreans beat foreigners? are you serious?
When 2 players of the same skill face each other guess who wins ? The one who plays the imbalanced race. So no, skill is not the answer to imbalance. The answer of imbalance is balance and only blizzard can take care of that, the players can just deal with it and try their hardest.
|
The y-axis scale REALLY exadurates everything. It's like the graph you would show on fox news to prove a point or something, while technicaly the information there isn't false, it's manipulated in a way that makes it seem very different from how it would actually look.
|
On November 07 2011 06:10 Amui wrote:Show nested quote +On November 07 2011 06:05 Grapefruit wrote:
Those graphs are major bullshit, since they only go form 40-60%, which makes the differences seem much more harsh then they actually are. This is just motivation for stupid people to continue whining.
The graphs are perfectly fine. They make it actually mean something when you look at it and actually read the sides, instead of derping around. Also if you didn't know, 40-60 is fine, because if one race is winning 60% against another, their win ratio is 1.5:1 which is already an absolutely massive difference.
I have no idea what you're talking about.
I didn't even talk about the win percentages, you derp.
The problem is that the graphs visually suggest that Terran and Zerg a winning about twice as often as Protoss, when the actual difference is only ~7-8%. This could easily be solved by showing graphs from 1-100% instead of 40-60%.
|
On November 07 2011 06:19 Logros wrote:
Ah so this is why Terran has been 5% higher in winrate on average the last 12 months. You solved the problem I would message Blizzard asap!
I wouldn't bother replying to him. His post history says everything you need to know, ranging from "protoss players are just being retards as usual" (here) to "protoss players...... /facepalm" (here).
He clearly either wants attention or is just heavily biased and doesn't want his race to look biased. Not to mention that what he wrote most recently is just plain wrong - if you flip a coin 1000 times and it lands on it's head 1000/1000 of those times, there is a (1/2)^1000 chance of that happening. Sure, it is chance, but they possibility is so small it is far more likely to be weighted. This is why statistics are used.
On November 07 2011 06:20 iky43210 wrote:
also 5% higher is really nothing. Do you know in Chess white have a 55% winrates to black's 45%? I don't think many people would argue that game is imbalance
I'd be very interested to know where you got that piece of information. Until there is a source, I call bollocks. How would you get every chess game, even in the last hundred years, to get a winrate from it?
|
|
|
|
|
|
EPT
