|
On January 09 2015 19:46 Grumbels wrote:Show nested quote +On January 09 2015 19:03 mCon.Hephaistas wrote:
Ofcourse the winrates will look decent when you have less zergs in tournaments atm.
Better zergs will play vs lesser players of other races since more of them qualified. I'm not sure if that rule holds up in practice. You could think that if zerg is weak vs terran and that only strong zerg players survive, that they're still not favored vs top players of other races because of the (assumed) imbalance, so the win rates won't necessarily be 50/50.
Ur right man, in fact i guess Zerg is just slightly too strong in ZvT then according to those winrates :')
|
On January 09 2015 19:03 mCon.Hephaistas wrote:
Ofcourse the winrates will look decent when you have less zergs in tournaments atm.
Better zergs will play vs lesser players of other races since more of them qualified.
This definitely affects the statistics to some degree, the question is how much, and which races? If ZvT is a huge problem but ZvP is balanced (which is what a "non-Terran favored" map pool is meant to address), ZvP should be considerably higher than ZvT, because the same top Zergs that are on even ground against the average Terrans should be eating crappy balanced Protosses for breakfast. Instead, a cursory look reveals the following stats for ZvP:
Proleague: 4-5
IEM SJ: 10-11
IEM SJ Korea: 9-15
IEM SJ Asia: 9-11
Hot6ix: 1-5
SSL qualifier: 6-3
For a ZvP ratio of 39-51 or 43.3%. This is actually way, way worse than ZvT across those events, which is approximately 52%.
|
I'd just like to reiterate the common wisdom that still holds:
a) We're too close to LotV for any changes to occur anyway. And the game will change so much that this data is all but moot.
b) The numbers are so incredibly small, they effectively tell us nothing. The 39-51 ZvP winrate, i.e., 43% would shift 2% by a single 2-0 victory by a Z over T. This should make it clear that the data is way too volatile to tell us anything about balance at the moment. In fact, it takes only 12 victories, i.e., four BO3 sweeps for perfect balance. This could easily occur over a single qualifying tournament.
Of course, looking at the numbers is better than saying RANDOM RACE OP without any evidence, but still, we're not at a point where we can say anything sensible about balance.
|
The thrend that we have is, the more competitive the tournament setting the less zergs we have proportionally, and the worst Zerg winrates become, lets see if GSL/SSL keep that up.
I wonder if Life sucked more if Zerg wouldn't have gotten a buff.
|
On January 09 2015 20:44 pure.Wasted wrote:Show nested quote +On January 09 2015 19:03 mCon.Hephaistas wrote:
Ofcourse the winrates will look decent when you have less zergs in tournaments atm.
Better zergs will play vs lesser players of other races since more of them qualified. This definitely affects the statistics to some degree, the question is how much, and which races? If ZvT is a huge problem but ZvP is balanced (which is what a "non-Terran favored" map pool is meant to address), ZvP should be considerably higher than ZvT, because the same top Zergs that are on even ground against the average Terrans should be eating crappy balanced Protosses for breakfast. Instead, a cursory look reveals the following stats for ZvP: Proleague: 4-5 IEM SJ: 10-11 IEM SJ Korea: 9-15 IEM SJ Asia: 9-11 Hot6ix: 1-5 SSL qualifier: 6-3 For a ZvP ratio of 39-51 or 43.3%. This is actually way, way worse than ZvT across those events, which is approximately 52%.
Many maps are a problem in ZvP. Blizzard never said that Zerg was struggling due to Terran dominating them too hard as far as I recall. They said that in general Zerg was doing slightly worse at the toplevel and they want to fix this with maps.
Imo, in both matchups the maps are a bit of a problem for Zerg. The only good Zerg map is Merry Go Round vs Protoss, while both P and T have multiple strong and very strong maps up on Zerg. This isn't too much of a problem when there are vetoes and it is Bo3, but if the maps are given (like it often happens in qualifiers) or people have to play semi-finals/finals in Bo5s and Bo7s, they are eventually forced to play the sub-45% maps and then it's a little tough.
|
On January 09 2015 20:56 Ghanburighan wrote:
I'd just like to reiterate the common wisdom that still holds:
a) We're too close to LotV for any changes to occur anyway. And the game will change so much that this data is all but moot.
b) The numbers are so incredibly small, they effectively tell us nothing. The 39-51 ZvP winrate, i.e., 43% would shift 2% by a single 2-0 victory by a Z over T. This should make it clear that the data is way too volatile to tell us anything about balance at the moment. In fact, it takes only 12 victories, i.e., four BO3 sweeps for perfect balance. This could easily occur over a single qualifying tournament.
Of course, looking at the numbers is better than saying RANDOM RACE OP without any evidence, but still, we're not at a point where we can say anything sensible about balance.
You're definitely right about the ZvP sample size, 90 games isn't much. ZvT is a different story, over the three counts I've done since the WM patch, I've tallied more than 250 games. So I think we're getting to a place where we can start having intelligent conversations about balance.
|
I think that TooDming-San,Mc-Sakya,Zest-PiG,Stats-Sakya,Hyun-Soul and Solar-Soul from the Asian qualifiers (IEM SJ) shouldn't be considered cause the difference between the players is huge (4 series won by toss and 2 by zerg)
That means that 8 games won by protoss and 4 by zerg are not from top tier korean vs top tier korean.
That changes the win rates from 39-51 to 35-43
|
On January 09 2015 21:33 Kuchikikun wrote:
I think that TooDming-San,Mc-Sakya,Zest-PiG,Stats-Sakya,Hyun-Soul and Solar-Soul from the Asian qualifiers (IEM SJ) shouldn't be considered cause the difference between the players is huge (4 series won by toss and 2 by zerg)
That means that 8 games won by protoss and 4 by zerg are not from top tier korean vs top tier korean.
That changes the win rates from 39-51 to 35-43
I don't count any non-Korean vs Korean games in any of my calculations, except the very few times where I specifically note otherwise! So your wish was granted by default.
|
On January 09 2015 20:16 mCon.Hephaistas wrote:Show nested quote +On January 09 2015 19:46 Grumbels wrote:On January 09 2015 19:03 mCon.Hephaistas wrote:
Ofcourse the winrates will look decent when you have less zergs in tournaments atm.
Better zergs will play vs lesser players of other races since more of them qualified. I'm not sure if that rule holds up in practice. You could think that if zerg is weak vs terran and that only strong zerg players survive, that they're still not favored vs top players of other races because of the (assumed) imbalance, so the win rates won't necessarily be 50/50. Ur right man, in fact i guess Zerg is just slightly too strong in ZvT then according to those winrates :')
I can't tell anymore when people are being sarcastic. 
Just tell me if this is wrong:
Scenario 1:
- a tournament with 10 T players vs 10 equally skilled Z players will have an expected win rate of 50%
Scenario 2:
- let's say you add some imbalance so that all zerg players lose, like, 5 spots in the ranking
- now you have a tournament with 10 top T players vs 5 top Z players. The zerg players are going to match up fairly vs the terran players ranked #6-#10, but they're going to lose against the terran players ranked #1-#5
- therefore the expected win rate should be more than 50% in favor of terran
I see this argument a lot: win rates will level off to 50% because only good zerg players qualify. It could be true, but I somewhat doubt it. I think it's true that it will become closer to 50% if you're looking at an entire ladder because only at the absolute highest and lowest level will there be unfair match-ups. However, one thing to keep in mind is that players that win games will on average play more games because they get deeper into the tournaments. Therefore stats from the pro scene are biased to resemble this more than ladder and so maybe you should expect >50% win rates assuming imbalance exists.
There's more to win rates though. There are legacy spots for tournaments based on old balance that affect win rates and there's also the fact that balance differs per skill level, so maybe zerg is fine at the top level but weak at the level below that (as an example).
Anyway, I don't know if my reasoning is sound, but at least I think people should stop just saying that win rates will tend towards 50/50 without adding some caution. :/
|
On January 09 2015 21:39 pure.Wasted wrote:Show nested quote +On January 09 2015 21:33 Kuchikikun wrote:
I think that TooDming-San,Mc-Sakya,Zest-PiG,Stats-Sakya,Hyun-Soul and Solar-Soul from the Asian qualifiers (IEM SJ) shouldn't be considered cause the difference between the players is huge (4 series won by toss and 2 by zerg)
That means that 8 games won by protoss and 4 by zerg are not from top tier korean vs top tier korean.
That changes the win rates from 39-51 to 35-43
I don't count any non-Korean vs Korean games in any of my calculations, except the very few times where I specifically note otherwise! So your wish was granted by default.
I think that's wrong cause if the diparity of skill is that huge the race doesn't matter
There's also AK versus Dear in the KR qualifiers so the winrate between players that are more of less at the same level of skill should be 35-41
|
On January 09 2015 21:45 Kuchikikun wrote:Show nested quote +On January 09 2015 21:39 pure.Wasted wrote:On January 09 2015 21:33 Kuchikikun wrote:
I think that TooDming-San,Mc-Sakya,Zest-PiG,Stats-Sakya,Hyun-Soul and Solar-Soul from the Asian qualifiers (IEM SJ) shouldn't be considered cause the difference between the players is huge (4 series won by toss and 2 by zerg)
That means that 8 games won by protoss and 4 by zerg are not from top tier korean vs top tier korean.
That changes the win rates from 39-51 to 35-43
I don't count any non-Korean vs Korean games in any of my calculations, except the very few times where I specifically note otherwise! So your wish was granted by default. I think that's wrong cause if the diparity of skill is that huge the race doesn't matter There's also AK versus Dear in the KR qualifiers so the winrate between players that are more of less at the same level of skill should be 35-41
I think you misunderstood me. I don't count games where the skill gap is obviously huge. The determining factor I use is whether the player is Korean or not. It's not a flawless method, but I can't think of one that would be less biased. So all my numbers are Korean vs Korean games only.
If I'm the one misunderstanding you, then uh... please rephrase.
|
On January 09 2015 22:04 pure.Wasted wrote:Show nested quote +On January 09 2015 21:45 Kuchikikun wrote:On January 09 2015 21:39 pure.Wasted wrote:On January 09 2015 21:33 Kuchikikun wrote:
I think that TooDming-San,Mc-Sakya,Zest-PiG,Stats-Sakya,Hyun-Soul and Solar-Soul from the Asian qualifiers (IEM SJ) shouldn't be considered cause the difference between the players is huge (4 series won by toss and 2 by zerg)
That means that 8 games won by protoss and 4 by zerg are not from top tier korean vs top tier korean.
That changes the win rates from 39-51 to 35-43
I don't count any non-Korean vs Korean games in any of my calculations, except the very few times where I specifically note otherwise! So your wish was granted by default. I think that's wrong cause if the diparity of skill is that huge the race doesn't matter There's also AK versus Dear in the KR qualifiers so the winrate between players that are more of less at the same level of skill should be 35-41 I think you misunderstood me. I don't count games where the skill gap is obviously huge. The determining factor I use is whether the player is Korean or not. It's not a flawless method, but I can't think of one that would be less biased. So all my numbers are Korean vs Korean games only. If I'm the one misunderstanding you, then uh... please rephrase.
He's saying the evidence is contrary to his belief and we should burn the infidels.
At least that's what it looks like to me.
|
On January 09 2015 21:43 Grumbels wrote:Show nested quote +On January 09 2015 20:16 mCon.Hephaistas wrote:On January 09 2015 19:46 Grumbels wrote:On January 09 2015 19:03 mCon.Hephaistas wrote:
Ofcourse the winrates will look decent when you have less zergs in tournaments atm.
Better zergs will play vs lesser players of other races since more of them qualified. I'm not sure if that rule holds up in practice. You could think that if zerg is weak vs terran and that only strong zerg players survive, that they're still not favored vs top players of other races because of the (assumed) imbalance, so the win rates won't necessarily be 50/50. Ur right man, in fact i guess Zerg is just slightly too strong in ZvT then according to those winrates :') I can't tell anymore when people are being sarcastic. Just tell me if this is wrong: Scenario 1: - a tournament with 10 T players vs 10 equally skilled Z players will have an expected win rate of 50% Scenario 2: - let's say you add some imbalance so that all zerg players lose, like, 5 spots in the ranking - now you have a tournament with 10 top T players vs 5 top Z players. The zerg players are going to match up fairly vs the terran players ranked #6-#10, but they're going to lose against the terran players ranked #1-#5 - therefore the expected win rate should be more than 50% in favor of terran I see this argument a lot: win rates will level off to 50% because only good zerg players qualify. It could be true, but I somewhat doubt it. I think it's true that it will become closer to 50% if you're looking at an entire ladder because only at the absolute highest and lowest level will there be unfair match-ups. However, one thing to keep in mind is that players that win games will on average play more games because they get deeper into the tournaments. Therefore stats from the pro scene are biased to resemble this more than ladder and so maybe you should expect >50% win rates assuming imbalance exists. There's more to win rates though. There are legacy spots for tournaments based on old balance that affect win rates and there's also the fact that balance differs per skill level, so maybe zerg is fine at the top level but weak at the level below that (as an example). Anyway, I don't know if my reasoning is sound, but at least I think people should stop just saying that win rates will tend towards 50/50 without adding some caution. :/
No but it's quite obvious Zerg is in a tough spot, I'm also quite sure Terran winrates weren't bad before they got mine/hellbat/thor buffed.
Anyway let's have a look at blizzcon. 3 out of 4 Zergs lost to a terran in the first round.
But because Life on his own beat a some Terrans the winrates still looked good, but that doesn't mean it's balanced at all.
|
If you're gonna use just recent data to gauge balance, and completely ignore the last WCS season (which terran absolutely wrecked everyone from ro8 on), you should wait until GSL & SSL progress further, qualifiers are an awful way to gauge balance, unless you not only filter for koreans vs koreans only but also for Aligulac rating or Code S participation in the past 6months or so.
|
On January 10 2015 05:06 sibs wrote:
If you're gonna use just recent data to gauge balance, and completely ignore the last WCS season (which terran absolutely wrecked everyone from ro8 on), you should wait until GSL & SSL progress further, qualifiers are an awful way to gauge balance, unless you not only filter for koreans vs koreans only but also for Aligulac rating or Code S participation in the past 6months or so.
If you filter for recent Code S participation you bias the results in favor of the matches of players that used to do well six months ago with different balance. For instance, let's say you end up looking at a small group of terran players that did well in 2014, then if the current balance is at 50% but in the past terran was disfavored, then maybe you'll get >50% stats indicating that terran is overpowered. So I think the results become tricky to interpret.
|
I think you misunderstood me. I don't count games where the skill gap is obviously huge. The determining factor I use is whether the player is Korean or not. It's not a flawless method, but I can't think of one that would be less biased. So all my numbers are Korean vs Korean games only.
If I'm the one misunderstanding you, then uh... please rephrase.
Sorry,I was the one that misunderstood!
|
|
|
Nice read. Especially agree on that part:
Blizzard can focus on small things and become too indecisive as a result - 150+ official balance updates over the course of Starcraft II’s history is not a staggering amount. When you take that 150+ and realize how many of those changes have either been reversed or patched over, it gets a lot smaller. The Thor and Void Ray alone account for 20 balance updates, most of which have been patched over and over. This type of indecisiveness has lead Blizzard to focus on aspects of the game that aren’t as important to most players, and over the years that has had an impact on the community. When everyone is asking Blizzard for tools to deal with something or to fix a certain aspect of a race and there’s a balance update that changes spine crawler AI, it’s hard not to feel like the balance team is disjointed from the sentiment of the community.
There have been way too few patches for my liking. And the ones they do are often not the important ones. And come late. Especially in 2013 they should have patched much more, since HotS was young, yet, patterns were already being explored. (e.g. swarm hosts, blink allins, SCV pulls to name a few ones that have been around and more than annoying up to that day)
|
On January 10 2015 08:58 Big J wrote:Nice read. Especially agree on that part: Blizzard can focus on small things and become too indecisive as a result - 150+ official balance updates over the course of Starcraft II’s history is not a staggering amount. When you take that 150+ and realize how many of those changes have either been reversed or patched over, it gets a lot smaller. The Thor and Void Ray alone account for 20 balance updates, most of which have been patched over and over. This type of indecisiveness has lead Blizzard to focus on aspects of the game that aren’t as important to most players, and over the years that has had an impact on the community. When everyone is asking Blizzard for tools to deal with something or to fix a certain aspect of a race and there’s a balance update that changes spine crawler AI, it’s hard not to feel like the balance team is disjointed from the sentiment of the community.There have been way too few patches for my liking. And the ones they do are often not the important ones. And come late. Especially in 2013 they should have patched much more, since HotS was young, yet, patterns were already being explored. (e.g. swarm hosts, blink allins, SCV pulls to name a few ones that have been around and more than annoying up to that day)
Meanwhile there is another Starbow patch out with like 30 changes. :p
Anyway, I thought the article raised some interesting questions. So, you can see that at some point in 2011 Blizzard stopped balancing the game and the author correlates this to starting work on Heart of the Swarm. I don't know if this is true, but it seems like a good hypothesis because it would explain why they took so long with Heart of the Swarm, since they were still focused on Wings of Liberty in late 2011.
I also think that if you are focused on a game that this will be reflected in having more changes. I don't buy this idea that Blizzard was taking a cautious approach because they felt that the metagame was developing nicely or whatever, and that's why they were hesitant in patching too much, especially considering how broken the game still was in 2012. So extrapolating this to the low amount of Heart of the Swarm patching, this gives you the idea that the team is not focused on Heart of the Swarm and instead is working on Heroes / LotV.
It's a bit of a rough metric, equating patch changes with number of Blizzard employees working on Starcraft, but maybe it's a useful guide?
|
On January 11 2015 01:28 Grumbels wrote:Show nested quote +On January 10 2015 08:58 Big J wrote:Nice read. Especially agree on that part: Blizzard can focus on small things and become too indecisive as a result - 150+ official balance updates over the course of Starcraft II’s history is not a staggering amount. When you take that 150+ and realize how many of those changes have either been reversed or patched over, it gets a lot smaller. The Thor and Void Ray alone account for 20 balance updates, most of which have been patched over and over. This type of indecisiveness has lead Blizzard to focus on aspects of the game that aren’t as important to most players, and over the years that has had an impact on the community. When everyone is asking Blizzard for tools to deal with something or to fix a certain aspect of a race and there’s a balance update that changes spine crawler AI, it’s hard not to feel like the balance team is disjointed from the sentiment of the community.There have been way too few patches for my liking. And the ones they do are often not the important ones. And come late. Especially in 2013 they should have patched much more, since HotS was young, yet, patterns were already being explored. (e.g. swarm hosts, blink allins, SCV pulls to name a few ones that have been around and more than annoying up to that day) Meanwhile there is another Starbow patch out with like 30 changes. :p Anyway, I thought the article raised some interesting questions. So, you can see that at some point in 2011 Blizzard stopped balancing the game and the author correlates this to starting work on Heart of the Swarm. I don't know if this is true, but it seems like a good hypothesis because it would explain why they took so long with Heart of the Swarm, since they were still focused on Wings of Liberty in late 2011. I also think that if you are focused on a game that this will be reflected in having more changes. I don't buy this idea that Blizzard was taking a cautious approach because they felt that the metagame was developing nicely or whatever, and that's why they were hesitant in patching too much, especially considering how broken the game still was in 2012. So extrapolating this to the low amount of Heart of the Swarm patching, this gives you the idea that the team is not focused on Heart of the Swarm and instead is working on Heroes / LotV. It's a bit of a rough metric, equating patch changes with number of Blizzard employees working on Starcraft, but maybe it's a useful guide?
In contrary, I do actually believe blizzard when they say they have changed their balance approach to be more cautious. It makes sense, since they got so much shit in 2010-11 for the heavier patching. And anytime they even just release their thoughts on the tiniest changes it feels like 50% of the comments are "just let the players figure it out on their own".
I wouldn't dismiss the other theory either, because why change units when you are already planning on solving the problems with expansion content.
However I think the heavy community pressure is the main reason why it often feels like they are hiding in the darkest corners. Basically any of their comments these days gets so much bullshit that I find it somewhat brave that they are communicating at all.
|
|
|
|
|
|
EPT
