|
On December 03 2012 06:17 Emzeeshady wrote:Show nested quote +On December 03 2012 06:10 Godwrath wrote:On December 03 2012 06:04 Emzeeshady wrote:On December 03 2012 05:08 Jebediah wrote:
The stats might not be spot on, because Bomber vs Polt distracted me while counting :D
I looked through the Nov premier tournaments on Liquipedia and took the map wins, not the match wins. Mirrors weren't counted.
Code S:
TvZ: 24-39
PvZ: 10-17
TvP: 16-14
Code A:
TvZ: 4-8
PvZ: 5-3
TvP: 21-12
ESWC:
TvZ: 7-8
PvZ: 5-3
TvP: 21-12
MLG:
TvZ: 26-28
PvZ: 38-36
TvP: 26-18
BWC:
TvZ: 5-5
PvZ: 41-28
TvP: 5-8
IEM:
TvZ: 14-14
PvZ: 29-30
TvP: 14-8
DHW:
TvZ: 11-7
PvZ; 33-26
TvP: 17-13
WCG:
TvZ 3-2
PvZ: 33-26
TvP: 6-10
IPL5:
TvZ: 29-49
PvZ: 31-41
TvP: 26-21
Overall:
TvZ: 123-160 -> 43,5% / 56,5%
PvZ; 246-224 -> 51,3% / 48,7%
TvP: 137-111 -> 55,2% / 44,8%
TvX: 531 Games, 260 Wins -> 49%
PvX: 708 Games, 347 Wins -> 49%
ZvX: 748 Games, 384 Wins -> 51,9%
Again, this might not be completely accurate. But even if I missed a couple of games, it shouldn't make much of a difference.
I KNEW IT Protoss players are full of shit! Yes, because BWC which is carrying those stats had a lot of big korean Z. Because they couldn't qualify -_- not to mention protoss is favoured at many other of these tournaments as well. Just admit it. Toss players just like to whine for no reason
Interesting. Tell me again how is the GSL %winrate at PvZ, which is the most stacked format out there.
|
On December 02 2012 16:05 Cascade wrote:
Not saying that sc2 is balance or inbalanced, but this is a completely pointless thread. Like a lot of "statistics" threads in sc2.
1. Include errors
snip..
TL:DR error analysis pl0x.
Ok, this post makes my brain bleed. While I don't find the OP particularly interesting or important, the poster calculated the actual win rates for this tournament. He took all the games that were played and just divided two numbers. There is no error.
|
On December 02 2012 07:17 Zenbrez wrote:Show nested quote +On December 02 2012 07:11 corpuscle wrote:On December 02 2012 07:07 Jormundr wrote:On December 02 2012 07:04 corpuscle wrote:
Um, seems to me like you're factoring in mirror matchups in the "overall" score, unless I'm missing something big... I don't really see how a 69% vT and 61% vP can average out to 52%, for example That's the percentage of games won by the race out of total number of games played, not an average win %. The numbers still don't make any sense, though. If Terran is winning 31% of their vZ, 55% of their vP, and 50% of their vT (obviously), how does it end up as 24% overall? If I won 9-1 tvz, I have a 90% winrate.
If I won 1-4 tvp, I have a 20% winrate.
I have a 66.67% (10/15) total winrate, even though the average of the 2 is 55% (110/2). That's not how you calculate an average in a situation like that. You've played twice as many TvZs, so you use a weighted average, i.e. (2*90 + 20)/3 = 66.67 You're telling me that doing 10/15 does not calculate your total average win percentage..? Total wins / total games = total win rate. Don't need to weight anything. There is more than 1 way of doing simple calculations. Show nested quote +Aside from being interesting using data like this from a single tournament is pointless. Sample size way to small. Sample size? He used the population. There can not physically be a bigger "sample".
lol what a great way to misuse those two words.
Yes, it is the population of all the games of the tournament. Therefore his results are 100% true within the games of this tournament.
This tournament, however, is only a very small sample of all games played within recent times, hence the other poster pointed out that it was the sample was too small to say anything about the entire population of games played.
No one is surprised that these stats are completely true within this tournament, the interesting question to follow is whether they have any external validity for the balance situation of SC2. You're acting as if he is questioning the obvious.
|
|
|
On December 03 2012 06:32 Emzeeshady wrote:Show nested quote +On December 03 2012 06:26 Godwrath wrote:On December 03 2012 06:17 Emzeeshady wrote:On December 03 2012 06:10 Godwrath wrote:On December 03 2012 06:04 Emzeeshady wrote:On December 03 2012 05:08 Jebediah wrote:
The stats might not be spot on, because Bomber vs Polt distracted me while counting :D
I looked through the Nov premier tournaments on Liquipedia and took the map wins, not the match wins. Mirrors weren't counted.
Code S:
TvZ: 24-39
PvZ: 10-17
TvP: 16-14
Code A:
TvZ: 4-8
PvZ: 5-3
TvP: 21-12
ESWC:
TvZ: 7-8
PvZ: 5-3
TvP: 21-12
MLG:
TvZ: 26-28
PvZ: 38-36
TvP: 26-18
BWC:
TvZ: 5-5
PvZ: 41-28
TvP: 5-8
IEM:
TvZ: 14-14
PvZ: 29-30
TvP: 14-8
DHW:
TvZ: 11-7
PvZ; 33-26
TvP: 17-13
WCG:
TvZ 3-2
PvZ: 33-26
TvP: 6-10
IPL5:
TvZ: 29-49
PvZ: 31-41
TvP: 26-21
Overall:
TvZ: 123-160 -> 43,5% / 56,5%
PvZ; 246-224 -> 51,3% / 48,7%
TvP: 137-111 -> 55,2% / 44,8%
TvX: 531 Games, 260 Wins -> 49%
PvX: 708 Games, 347 Wins -> 49%
ZvX: 748 Games, 384 Wins -> 51,9%
Again, this might not be completely accurate. But even if I missed a couple of games, it shouldn't make much of a difference.
I KNEW IT Protoss players are full of shit! Yes, because BWC which is carrying those stats had a lot of big korean Z. Because they couldn't qualify -_- not to mention protoss is favoured at many other of these tournaments as well. Just admit it. Toss players just like to whine for no reason Interesting. Tell me again how is the GSL %winrate at PvZ, which is the most stacked format out there. since when does one tournament mean anything conclusive?
Since they have somewhat the best skill/race distribution. BWC ? WCG ? For all we can see IPL5 and GSL are the only tourneys stacked enough for this to happen.
No one is surprised that these stats are completely true within this tournament, the interesting question to follow is whether they have any external validity for the balance situation of SC2.
The game never should be balanced by stats, not even big ass stats. It should be balanced around how the game is played out. Nothing else.
|
On December 03 2012 06:37 Godwrath wrote:Show nested quote +On December 03 2012 06:32 Emzeeshady wrote:On December 03 2012 06:26 Godwrath wrote:On December 03 2012 06:17 Emzeeshady wrote:On December 03 2012 06:10 Godwrath wrote:On December 03 2012 06:04 Emzeeshady wrote:On December 03 2012 05:08 Jebediah wrote:
The stats might not be spot on, because Bomber vs Polt distracted me while counting :D
I looked through the Nov premier tournaments on Liquipedia and took the map wins, not the match wins. Mirrors weren't counted.
Code S:
TvZ: 24-39
PvZ: 10-17
TvP: 16-14
Code A:
TvZ: 4-8
PvZ: 5-3
TvP: 21-12
ESWC:
TvZ: 7-8
PvZ: 5-3
TvP: 21-12
MLG:
TvZ: 26-28
PvZ: 38-36
TvP: 26-18
BWC:
TvZ: 5-5
PvZ: 41-28
TvP: 5-8
IEM:
TvZ: 14-14
PvZ: 29-30
TvP: 14-8
DHW:
TvZ: 11-7
PvZ; 33-26
TvP: 17-13
WCG:
TvZ 3-2
PvZ: 33-26
TvP: 6-10
IPL5:
TvZ: 29-49
PvZ: 31-41
TvP: 26-21
Overall:
TvZ: 123-160 -> 43,5% / 56,5%
PvZ; 246-224 -> 51,3% / 48,7%
TvP: 137-111 -> 55,2% / 44,8%
TvX: 531 Games, 260 Wins -> 49%
PvX: 708 Games, 347 Wins -> 49%
ZvX: 748 Games, 384 Wins -> 51,9%
Again, this might not be completely accurate. But even if I missed a couple of games, it shouldn't make much of a difference.
I KNEW IT Protoss players are full of shit! Yes, because BWC which is carrying those stats had a lot of big korean Z. Because they couldn't qualify -_- not to mention protoss is favoured at many other of these tournaments as well. Just admit it. Toss players just like to whine for no reason Interesting. Tell me again how is the GSL %winrate at PvZ, which is the most stacked format out there. since when does one tournament mean anything conclusive? Since they have somewhat the best skill/race distribution. BWC ? WCG ? For all we can see IPL5 and GSL are the only tourneys stacked enough for this to happen. Show nested quote +No one is surprised that these stats are completely true within this tournament, the interesting question to follow is whether they have any external validity for the balance situation of SC2. The game never should be balanced by stats, not even big ass stats. It should be balanced around how the game is played out. Nothing else.
I strongly disagree. Stats are quantitative, "played out" is qualitative. One of these do not have viewer bias. While looking at how things play out is important, it is extremely unscientific to ignore the stats.
|
|
|
On December 03 2012 06:41 achan1058 wrote:Show nested quote +On December 03 2012 06:37 Godwrath wrote:On December 03 2012 06:32 Emzeeshady wrote:On December 03 2012 06:26 Godwrath wrote:On December 03 2012 06:17 Emzeeshady wrote:On December 03 2012 06:10 Godwrath wrote:On December 03 2012 06:04 Emzeeshady wrote:On December 03 2012 05:08 Jebediah wrote:
The stats might not be spot on, because Bomber vs Polt distracted me while counting :D
I looked through the Nov premier tournaments on Liquipedia and took the map wins, not the match wins. Mirrors weren't counted.
Code S:
TvZ: 24-39
PvZ: 10-17
TvP: 16-14
Code A:
TvZ: 4-8
PvZ: 5-3
TvP: 21-12
ESWC:
TvZ: 7-8
PvZ: 5-3
TvP: 21-12
MLG:
TvZ: 26-28
PvZ: 38-36
TvP: 26-18
BWC:
TvZ: 5-5
PvZ: 41-28
TvP: 5-8
IEM:
TvZ: 14-14
PvZ: 29-30
TvP: 14-8
DHW:
TvZ: 11-7
PvZ; 33-26
TvP: 17-13
WCG:
TvZ 3-2
PvZ: 33-26
TvP: 6-10
IPL5:
TvZ: 29-49
PvZ: 31-41
TvP: 26-21
Overall:
TvZ: 123-160 -> 43,5% / 56,5%
PvZ; 246-224 -> 51,3% / 48,7%
TvP: 137-111 -> 55,2% / 44,8%
TvX: 531 Games, 260 Wins -> 49%
PvX: 708 Games, 347 Wins -> 49%
ZvX: 748 Games, 384 Wins -> 51,9%
Again, this might not be completely accurate. But even if I missed a couple of games, it shouldn't make much of a difference.
I KNEW IT Protoss players are full of shit! Yes, because BWC which is carrying those stats had a lot of big korean Z. Because they couldn't qualify -_- not to mention protoss is favoured at many other of these tournaments as well. Just admit it. Toss players just like to whine for no reason Interesting. Tell me again how is the GSL %winrate at PvZ, which is the most stacked format out there. since when does one tournament mean anything conclusive? Since they have somewhat the best skill/race distribution. BWC ? WCG ? For all we can see IPL5 and GSL are the only tourneys stacked enough for this to happen. No one is surprised that these stats are completely true within this tournament, the interesting question to follow is whether they have any external validity for the balance situation of SC2. The game never should be balanced by stats, not even big ass stats. It should be balanced around how the game is played out. Nothing else. I strongly disagree. Stats are quantitative, "played out" is qualitative. One of these do not have viewer bias. While looking at how things play out is important, it is extremely unscientific to ignore the stats.
Anybody with braincells will disagree with you playing and watching PvZ. A match up entirely balanced by stats is awful for the game. If the "bias" is about being fun, i don't see anything wrong with it since it's a game, not the final balance of a company.
|
On December 03 2012 06:52 Godwrath wrote:Show nested quote +On December 03 2012 06:41 achan1058 wrote:On December 03 2012 06:37 Godwrath wrote:On December 03 2012 06:32 Emzeeshady wrote:On December 03 2012 06:26 Godwrath wrote:On December 03 2012 06:17 Emzeeshady wrote:On December 03 2012 06:10 Godwrath wrote:On December 03 2012 06:04 Emzeeshady wrote:On December 03 2012 05:08 Jebediah wrote:
The stats might not be spot on, because Bomber vs Polt distracted me while counting :D
I looked through the Nov premier tournaments on Liquipedia and took the map wins, not the match wins. Mirrors weren't counted.
Code S:
TvZ: 24-39
PvZ: 10-17
TvP: 16-14
Code A:
TvZ: 4-8
PvZ: 5-3
TvP: 21-12
ESWC:
TvZ: 7-8
PvZ: 5-3
TvP: 21-12
MLG:
TvZ: 26-28
PvZ: 38-36
TvP: 26-18
BWC:
TvZ: 5-5
PvZ: 41-28
TvP: 5-8
IEM:
TvZ: 14-14
PvZ: 29-30
TvP: 14-8
DHW:
TvZ: 11-7
PvZ; 33-26
TvP: 17-13
WCG:
TvZ 3-2
PvZ: 33-26
TvP: 6-10
IPL5:
TvZ: 29-49
PvZ: 31-41
TvP: 26-21
Overall:
TvZ: 123-160 -> 43,5% / 56,5%
PvZ; 246-224 -> 51,3% / 48,7%
TvP: 137-111 -> 55,2% / 44,8%
TvX: 531 Games, 260 Wins -> 49%
PvX: 708 Games, 347 Wins -> 49%
ZvX: 748 Games, 384 Wins -> 51,9%
Again, this might not be completely accurate. But even if I missed a couple of games, it shouldn't make much of a difference.
I KNEW IT Protoss players are full of shit! Yes, because BWC which is carrying those stats had a lot of big korean Z. Because they couldn't qualify -_- not to mention protoss is favoured at many other of these tournaments as well. Just admit it. Toss players just like to whine for no reason Interesting. Tell me again how is the GSL %winrate at PvZ, which is the most stacked format out there. since when does one tournament mean anything conclusive? Since they have somewhat the best skill/race distribution. BWC ? WCG ? For all we can see IPL5 and GSL are the only tourneys stacked enough for this to happen. No one is surprised that these stats are completely true within this tournament, the interesting question to follow is whether they have any external validity for the balance situation of SC2. The game never should be balanced by stats, not even big ass stats. It should be balanced around how the game is played out. Nothing else. I strongly disagree. Stats are quantitative, "played out" is qualitative. One of these do not have viewer bias. While looking at how things play out is important, it is extremely unscientific to ignore the stats. Anybody with braincells will disagree with you playing and watching PvZ. A match up entirely balanced by stats is awful for the game. If the "bias" is about being fun, i don't see anything wrong with it since it's a game, not the final balance of a company.
I never said entirely balanced by stats. I said stats must take a significant factor. Besides, whether a match up is boring or not is not remotely the same as balance.
|
Sample size is too small to conclude anything.
Fun stats regardless. 
|
On December 03 2012 06:58 achan1058 wrote:Show nested quote +On December 03 2012 06:52 Godwrath wrote:On December 03 2012 06:41 achan1058 wrote:On December 03 2012 06:37 Godwrath wrote:On December 03 2012 06:32 Emzeeshady wrote:On December 03 2012 06:26 Godwrath wrote:On December 03 2012 06:17 Emzeeshady wrote:On December 03 2012 06:10 Godwrath wrote:On December 03 2012 06:04 Emzeeshady wrote:On December 03 2012 05:08 Jebediah wrote:
The stats might not be spot on, because Bomber vs Polt distracted me while counting :D
I looked through the Nov premier tournaments on Liquipedia and took the map wins, not the match wins. Mirrors weren't counted.
Code S:
TvZ: 24-39
PvZ: 10-17
TvP: 16-14
Code A:
TvZ: 4-8
PvZ: 5-3
TvP: 21-12
ESWC:
TvZ: 7-8
PvZ: 5-3
TvP: 21-12
MLG:
TvZ: 26-28
PvZ: 38-36
TvP: 26-18
BWC:
TvZ: 5-5
PvZ: 41-28
TvP: 5-8
IEM:
TvZ: 14-14
PvZ: 29-30
TvP: 14-8
DHW:
TvZ: 11-7
PvZ; 33-26
TvP: 17-13
WCG:
TvZ 3-2
PvZ: 33-26
TvP: 6-10
IPL5:
TvZ: 29-49
PvZ: 31-41
TvP: 26-21
Overall:
TvZ: 123-160 -> 43,5% / 56,5%
PvZ; 246-224 -> 51,3% / 48,7%
TvP: 137-111 -> 55,2% / 44,8%
TvX: 531 Games, 260 Wins -> 49%
PvX: 708 Games, 347 Wins -> 49%
ZvX: 748 Games, 384 Wins -> 51,9%
Again, this might not be completely accurate. But even if I missed a couple of games, it shouldn't make much of a difference.
I KNEW IT Protoss players are full of shit! Yes, because BWC which is carrying those stats had a lot of big korean Z. Because they couldn't qualify -_- not to mention protoss is favoured at many other of these tournaments as well. Just admit it. Toss players just like to whine for no reason Interesting. Tell me again how is the GSL %winrate at PvZ, which is the most stacked format out there. since when does one tournament mean anything conclusive? Since they have somewhat the best skill/race distribution. BWC ? WCG ? For all we can see IPL5 and GSL are the only tourneys stacked enough for this to happen. No one is surprised that these stats are completely true within this tournament, the interesting question to follow is whether they have any external validity for the balance situation of SC2. The game never should be balanced by stats, not even big ass stats. It should be balanced around how the game is played out. Nothing else. I strongly disagree. Stats are quantitative, "played out" is qualitative. One of these do not have viewer bias. While looking at how things play out is important, it is extremely unscientific to ignore the stats. Anybody with braincells will disagree with you playing and watching PvZ. A match up entirely balanced by stats is awful for the game. If the "bias" is about being fun, i don't see anything wrong with it since it's a game, not the final balance of a company. I never said entirely balanced by stats. I said stats must take a significant factor. Besides, whether a match up is boring or not is not remotely the same as balance.
I'm a big fan of stats too
I would like to further add to what you said to satisfy the previous poster that you could also just make more specific stats that would cover how the game is played out better.
A simple example would be to calculate winrates for games at the 10-20 min mark and then one for 20 and beyond.
You could also make stats for when a particular united is fielded, how many bases a player has relative to the other, openings and so on and so. Do toss mostly win with timings or by dragging the game out...We all have an idea about that but having the actual stats would be so helpful
It's just most of the time that this data is not readily available and would require alot of work to obtain, but I really do feel that they could aid us alot in the judgement of the true balance issue!
It's so much easier to support your claims with stats. In the field of research, while qualitative studies are acknowledge as a scientific method, it is still at the bottom of the food chain when it comes to making actual claims about how things truly work.
|
Remember when you calculate the total win percentage, you take the total games won divided by the total games, not the average of the two percentages.
With that you get this:
Average results (WB 1-5 + LB 1-7)
Zerg wins%:
- vT: 68,75%
- vP: 58,62%
Terran wins%:
- vZ: 31,25%
- vP: 55,00%
Protoss wins%:
- vZ: 41,38%
- vT: 45,00%
|
Poor Terrans... Protoss seem to be louder but look at those stats..
|
I'm pretty sure the terran skill cap is way higher than the zergs, but even with a near-perfect terran play, the zerg still wins because the race is simply better and easier. That's bad. Leenock does perform consistently at foreign tournaments, he's frequently getting 1st/finals/etc. but is it because of his skill or just balance? It's very hard to tell, but I can safely tell you infestors aren't hard to use.
And the problems with interpreting the statistics is that, obviously it's a small sample. Another thing is, in SC2, it's much more random than any other game. Consistency is barely even there. In WC3, you KNEW who the top players were, and they ALWAYS got to the finals. That was SORT of the case back when MVP and Nestea were winning GSL/other big events, but even then, you'd see them lose to people all the time in other events. SC2 contains a lot more elements of randomness. Obviously, scouting and corner-cutting plays a large part of this, but another thing may just be that players do not perform that well at live events. Chess doesn't contain nearly the same amount of randomness as SC2 does, as you can pretty much see what they are doing always and make the best decisions.
We all hear about how players play the best at home or when practicing, so doesn't that mean the BEST games are during those times? Do live events even MEAN anything if they do not put out the best performances???? The reason we have "live" events in sports is because there is no other way to do it, but with SC2, there isn't even LAN, so what is even the point??????? Just to imitate real sports? If zerg is more forgiving or easier, maybe it's just that the live event environment causes both players to play less than superb and since zerg doesn't punish mistakes as much (-cough- infestors are too flexible -cough-), then the zergs just end up winning.
tl;dr..... Skill ceiling with zerg is bad, randomness in professional games is ALWAYS bad, and live events can produce poor games.
|
On December 03 2012 06:31 budar wrote:Show nested quote +On December 02 2012 16:05 Cascade wrote:
Not saying that sc2 is balance or inbalanced, but this is a completely pointless thread. Like a lot of "statistics" threads in sc2.
1. Include errors
snip..
TL:DR error analysis pl0x. Ok, this post makes my brain bleed. While I don't find the OP particularly interesting or important, the poster calculated the actual win rates for this tournament. He took all the games that were played and just divided two numbers. There is no error.
Do you understand the concept of error in statistics?
You see, we're trying to ascertain whether or not the matchups are balanced, that would be the population parameter. These games are merely a sample of all possible games that can be played, giving us a sample statistic.
We can now use the sample statistic to estimate the population parameter.
So if ZvT in IPL 5 is 70% or something, that is the sample statistic. Now, to see whether the matchup is balanced or not, we have to include error. Let's just say error is 15%. Thus, the actual matchup balance (disregarding bias, incorrect sampling, other variables such as player skill, etc.) is between 55% and 85%. That is what error means.
|
|
|
On December 03 2012 07:24 Entirety wrote:Show nested quote +On December 03 2012 06:31 budar wrote:On December 02 2012 16:05 Cascade wrote:
Not saying that sc2 is balance or inbalanced, but this is a completely pointless thread. Like a lot of "statistics" threads in sc2.
1. Include errors
snip..
TL:DR error analysis pl0x. Ok, this post makes my brain bleed. While I don't find the OP particularly interesting or important, the poster calculated the actual win rates for this tournament. He took all the games that were played and just divided two numbers. There is no error. Do you understand the concept of error in statistics? You see, we're trying to ascertain whether or not the matchups are balanced, that would be the population parameter. These games are merely a sample of all possible games that can be played, giving us a sample statistic. We can now use the sample statistic to estimate the population parameter. So if ZvT in IPL 5 is 70% or something, that is the sample statistic. Now, to see whether the matchup is balanced or not, we have to include error. Let's just say error is 15%. Thus, the actual matchup balance (disregarding bias, incorrect sampling, other variables such as player skill, etc.) is between 55% and 85%. That is what error means.
Why would you need error though, I'm kinda confused on that. Isn't error usually used when you want to extrapolate a sample statistic onto a larger unknown population?
|
|
|
On December 03 2012 07:35 SupLilSon wrote:Show nested quote +On December 03 2012 07:24 Entirety wrote:On December 03 2012 06:31 budar wrote:On December 02 2012 16:05 Cascade wrote:
Not saying that sc2 is balance or inbalanced, but this is a completely pointless thread. Like a lot of "statistics" threads in sc2.
1. Include errors
snip..
TL:DR error analysis pl0x. Ok, this post makes my brain bleed. While I don't find the OP particularly interesting or important, the poster calculated the actual win rates for this tournament. He took all the games that were played and just divided two numbers. There is no error. Do you understand the concept of error in statistics? You see, we're trying to ascertain whether or not the matchups are balanced, that would be the population parameter. These games are merely a sample of all possible games that can be played, giving us a sample statistic. We can now use the sample statistic to estimate the population parameter. So if ZvT in IPL 5 is 70% or something, that is the sample statistic. Now, to see whether the matchup is balanced or not, we have to include error. Let's just say error is 15%. Thus, the actual matchup balance (disregarding bias, incorrect sampling, other variables such as player skill, etc.) is between 55% and 85%. That is what error means. Why would you need error though, I'm kinda confused on that. Isn't error usually used when you want to extrapolate a sample statistic onto a larger unknown population?
Yes, exactly! I don't think anyone really cares about the IPL 5 statistics themselves, all they show is that Zerg did well at this particular tournament.
The larger unknown population we're interested in is the population of all possible games, the matchup itself... so we extrapolate IPL 5 TvZ statistics into the parameter of the TvZ matchup as a whole.
IPL 5 ZvT = 70%? No problem.
ZvT = 70%? HUGE problem.
So, if we say something like Zergs win 70% of their games versus Terran, just look at IPL 5! That would be flat-out wrong because we didn't include the error... Now, with error, we can say something along the lines of "we are 95% confident that Zergs win between 55%-85% of their games against Terran based upon the data from IPL 5" (yes I'm making the numbers up, but we can calculate the actual numbers, I'm just too lazy to do so right now)
|
On December 02 2012 07:07 n0ise wrote:
glad to see it's all close to 50%
I see what you did there.
|
|
|
|
|
|
EPT
