|
On July 01 2013 20:05 Ghanburighan wrote:
As I posted in the DB thread, I'll repeat here:
A pretty low number of games. Around 300 per MU. I would have expected more fluctuation due to the numbers.
The Korean sub 100 MU's show's greater fluctuation. But still remains roughly balanced.
Nice to see all the MU's are nicely balanced, and that's roughly the feel you get from watching tournaments too. We've had a whole lot of P, T and Z win tournaments recently.
Thanks to ChaosTerran.
Since when is 300 per MU a low number of games... or do you just mean relative to other months?
|
On July 02 2013 13:51 AxionSteel wrote:Show nested quote +On July 02 2013 13:46 BronzeKnee wrote:On July 02 2013 13:40 Thrillz wrote:On July 02 2013 13:33 BronzeKnee wrote:On July 02 2013 02:59 SlixSC wrote:On July 02 2013 02:57 BronzeKnee wrote:On July 02 2013 02:55 SlixSC wrote:On July 02 2013 02:52 BronzeKnee wrote:On July 02 2013 02:49 SlixSC wrote:On July 02 2013 02:43 BronzeKnee wrote:
Exactly as MasterOfPuppets said, you don't know what blackmail means.
This statement I made, is not blackmail:
More interesting is the fact that a guy named "ChaosTerran" came up with the statistics saying that Terran has a losing winrate in both mirrors... conflict of interests?
Yeah I'm sorry I misused the word blackmail. It's actually defamation, you are basically implying that the person has skewed the statistics out of personal interest. Before making such accusations though shouldn't you be presenting at least some evidence? And I won't reply to MasterOfPuppets anylonger, because his argument simply doesn't hold water. "They are at least master and thus relevant to pro level statistics". I'm sorry but I am high master and my level of play is absolutely not relevant to the pro level play, they are world's apart. "At least master" is such a terrible parameter to use to decide wether or not a game is relevant because pro level (even low GM) and masters are worlds apart. The question mark is there for a reason. If the digging that Plansix did is correct, then he indeed does have a Terran bias. That is incredibly important, because he "filtered" the games. He filtered out qualifiers for tournaments, but used the tournament statistics. That is really important. If Terran is truly OP, then they'll dominate the qualifiers and will let in a lot of "bad " Terran players. Then those bad Terrans will lose to good Protoss and Zerg, and there we have the reason why the statistics are the way they are. Actually, the parameters used were the same as in every other month. In the first two months they showed Terran as the most winning race, this month they didn't. SAME PARAMETERS 100%. But you missed my point. If Terran is truly OP, then they'll dominate the qualifiers and will let in a lot of "bad " Terran players. Then those bad Terrans will lose to good Protoss and Zerg, and there we have the reason why the statistics are the way they are.
That could completely explain the statistics. I'll go through and add in the qualifier data later. Time to go to work. Ok? But that's exactly what would happen every month, regardless of what the person compiling these statistics does? Not only these statistics, the same logic could be applied to aligulac's win rates too. So then I ask, why even bother in the first place? (and the effect could be even amplified in aligulac's statistics seeing as they include all kinds of low level matches and qualifiers?) Ah, we reached a consensus. Why should they bother? Both ChaosTerran and Aligulac have a filtering process that may actually render their data invalid. Looking purely at winrates is silly to be honest. Which is why I suggested this:
What would really be a worthwhile exercise (and I'll probably contact the Aligulac guy about this later) would be to use the Aligulac rating of each player, and then compare winrates by race. Thus if a heavily favored Protoss player like Dear defeated QXC, it wouldn't necessarily show that Protoss was imbalanced. However if QXC (who is a big underdog) defeated Dear, it might indicate imbalance, especially if Terran underdogs were defeating Protoss favorites on a regular basis when they normally did not.
And TheBB delivered. http://aligulac.com/reports/And now we can see that when I selected Protoss I made a massive error. That doesn't say Protoss is horribly bad right now at all..........not to mention there are a few problems with that method, just like the previous ones. The method is far better than the old one, because it somewhat correct for player skill, while the old one does not. Perfect? No. But better. Anyway, I like your edit from Protoss isn't bad at all to Protoss isn't horribly bad right now. Protoss has been massively underperforming, meaning that high level Protoss players are losing more to lower level Terrans and Zergs. Don't see how that is relevant to you making a massive error picking protoss, though. omg imba terran stealing my ladder points =( As long as it's balanced in Korea, then I'm somewhat satisfied. Would still encourage a slight nerf to hellbats though.
Over the history of SC2, Protoss has clearly done the poorest according to that chart, and it means that a Protoss player more often than not loses to Terran and Zerg players of equal skill. That is annoying to me, because I like to win.
But we all knew this didn't we... this chart just confirms what all the win rates showed. And the results.
|
On July 02 2013 13:54 BronzeKnee wrote:Show nested quote +On July 02 2013 13:51 Thrillz wrote:On July 02 2013 13:46 BronzeKnee wrote:On July 02 2013 13:40 Thrillz wrote:On July 02 2013 13:33 BronzeKnee wrote:On July 02 2013 02:59 SlixSC wrote:On July 02 2013 02:57 BronzeKnee wrote:On July 02 2013 02:55 SlixSC wrote:On July 02 2013 02:52 BronzeKnee wrote:On July 02 2013 02:49 SlixSC wrote:
[quote]
Yeah I'm sorry I misused the word blackmail. It's actually defamation, you are basically implying that the person has skewed the statistics out of personal interest. Before making such accusations though shouldn't you be presenting at least some evidence?
And I won't reply to MasterOfPuppets anylonger, because his argument simply doesn't hold water. "They are at least master and thus relevant to pro level statistics". I'm sorry but I am high master and my level of play is absolutely not relevant to the pro level play, they are world's apart. "At least master" is such a terrible parameter to use to decide wether or not a game is relevant because pro level (even low GM) and masters are worlds apart. The question mark is there for a reason. If the digging that Plansix did is correct, then he indeed does have a Terran bias. That is incredibly important, because he "filtered" the games. He filtered out qualifiers for tournaments, but used the tournament statistics. That is really important. If Terran is truly OP, then they'll dominate the qualifiers and will let in a lot of "bad " Terran players. Then those bad Terrans will lose to good Protoss and Zerg, and there we have the reason why the statistics are the way they are. Actually, the parameters used were the same as in every other month. In the first two months they showed Terran as the most winning race, this month they didn't. SAME PARAMETERS 100%. But you missed my point. If Terran is truly OP, then they'll dominate the qualifiers and will let in a lot of "bad " Terran players. Then those bad Terrans will lose to good Protoss and Zerg, and there we have the reason why the statistics are the way they are.
That could completely explain the statistics. I'll go through and add in the qualifier data later. Time to go to work. Ok? But that's exactly what would happen every month, regardless of what the person compiling these statistics does? Not only these statistics, the same logic could be applied to aligulac's win rates too. So then I ask, why even bother in the first place? (and the effect could be even amplified in aligulac's statistics seeing as they include all kinds of low level matches and qualifiers?) Ah, we reached a consensus. Why should they bother? Both ChaosTerran and Aligulac have a filtering process that may actually render their data invalid. Looking purely at winrates is silly to be honest. Which is why I suggested this:
What would really be a worthwhile exercise (and I'll probably contact the Aligulac guy about this later) would be to use the Aligulac rating of each player, and then compare winrates by race. Thus if a heavily favored Protoss player like Dear defeated QXC, it wouldn't necessarily show that Protoss was imbalanced. However if QXC (who is a big underdog) defeated Dear, it might indicate imbalance, especially if Terran underdogs were defeating Protoss favorites on a regular basis when they normally did not.
And TheBB delivered. http://aligulac.com/reports/And now we can see that when I selected Protoss I made a massive error. That doesn't say Protoss is horribly bad right now at all..........not to mention there are a few problems with that method, just like the previous ones. The method is far better than the old one, because it somewhat correct for player skill, while the old one does not. Perfect? No. But better. Anyway, I like your edit from Protoss isn't bad at all to Protoss isn't horribly bad right now. Protoss has been massively underperforming in SC2 compared to the other races, and right now high level Protoss players are losing more to lower level players from other races, than high level Terrans or Zergs do. But evidence show the contrary. Sorry but -11 says nothing to me, and aligulac has admitted that it isn't perfect, (see forgg nearly being #1, Soulkey not even top #10. Not jumping on the hysteria bandwagon, but nice try anyways. You literally just posted nothing. And I just said Aligulac wasn't perfect either. Why are you telling me it isn't?
Then why are you drawing a conclusion when you only have a few points from one method on a hypothesis? I personally think it's pretty good right now. Also, evidence is that winrates are uniform and not enough to point to imbalance, both methods from aligulac and liquipedia winrates show nothing to positively conclude imbalance. So what's your evidence that show the opposite? Certainly not a -11 rating?
|
On July 02 2013 13:58 Thrillz wrote:Show nested quote +On July 02 2013 13:54 BronzeKnee wrote:On July 02 2013 13:51 Thrillz wrote:On July 02 2013 13:46 BronzeKnee wrote:On July 02 2013 13:40 Thrillz wrote:On July 02 2013 13:33 BronzeKnee wrote:On July 02 2013 02:59 SlixSC wrote:On July 02 2013 02:57 BronzeKnee wrote:On July 02 2013 02:55 SlixSC wrote:On July 02 2013 02:52 BronzeKnee wrote:
[quote]
The question mark is there for a reason. If the digging that Plansix did is correct, then he indeed does have a Terran bias.
That is incredibly important, because he "filtered" the games. He filtered out qualifiers for tournaments, but used the tournament statistics. That is really important. If Terran is truly OP, then they'll dominate the qualifiers and will let in a lot of "bad " Terran players. Then those bad Terrans will lose to good Protoss and Zerg, and there we have the reason why the statistics are the way they are.
Actually, the parameters used were the same as in every other month. In the first two months they showed Terran as the most winning race, this month they didn't. SAME PARAMETERS 100%. But you missed my point. If Terran is truly OP, then they'll dominate the qualifiers and will let in a lot of "bad " Terran players. Then those bad Terrans will lose to good Protoss and Zerg, and there we have the reason why the statistics are the way they are.
That could completely explain the statistics. I'll go through and add in the qualifier data later. Time to go to work. Ok? But that's exactly what would happen every month, regardless of what the person compiling these statistics does? Not only these statistics, the same logic could be applied to aligulac's win rates too. So then I ask, why even bother in the first place? (and the effect could be even amplified in aligulac's statistics seeing as they include all kinds of low level matches and qualifiers?) Ah, we reached a consensus. Why should they bother? Both ChaosTerran and Aligulac have a filtering process that may actually render their data invalid. Looking purely at winrates is silly to be honest. Which is why I suggested this:
What would really be a worthwhile exercise (and I'll probably contact the Aligulac guy about this later) would be to use the Aligulac rating of each player, and then compare winrates by race. Thus if a heavily favored Protoss player like Dear defeated QXC, it wouldn't necessarily show that Protoss was imbalanced. However if QXC (who is a big underdog) defeated Dear, it might indicate imbalance, especially if Terran underdogs were defeating Protoss favorites on a regular basis when they normally did not.
And TheBB delivered. http://aligulac.com/reports/And now we can see that when I selected Protoss I made a massive error. That doesn't say Protoss is horribly bad right now at all..........not to mention there are a few problems with that method, just like the previous ones. The method is far better than the old one, because it somewhat correct for player skill, while the old one does not. Perfect? No. But better. Anyway, I like your edit from Protoss isn't bad at all to Protoss isn't horribly bad right now. Protoss has been massively underperforming in SC2 compared to the other races, and right now high level Protoss players are losing more to lower level players from other races, than high level Terrans or Zergs do. But evidence show the contrary. Sorry but -11 says nothing to me, and aligulac has admitted that it isn't perfect, (see forgg nearly being #1, Soulkey not even top #10. Not jumping on the hysteria bandwagon, but nice try anyways. You literally just posted nothing. And I just said Aligulac wasn't perfect either. Why are you telling me it isn't? Then why are you drawing a conclusion when you only have a few points from one method on a hypothesis? I personally think it's pretty good right now.
Pretty good relative to the past? Sure. Balanced... well you know the answer to that.
-22 is how much Protoss players are underperforming.
-11 is simply the difference between Protoss and Zerg and how much they underperform compared to predicted performance. But that is a useless calculation. So use -22.
|
On July 02 2013 13:33 BronzeKnee wrote:Show nested quote +On July 02 2013 02:59 SlixSC wrote:On July 02 2013 02:57 BronzeKnee wrote:On July 02 2013 02:55 SlixSC wrote:On July 02 2013 02:52 BronzeKnee wrote:On July 02 2013 02:49 SlixSC wrote:On July 02 2013 02:43 BronzeKnee wrote:
Exactly as MasterOfPuppets said, you don't know what blackmail means.
This statement I made, is not blackmail:
More interesting is the fact that a guy named "ChaosTerran" came up with the statistics saying that Terran has a losing winrate in both mirrors... conflict of interests?
Yeah I'm sorry I misused the word blackmail. It's actually defamation, you are basically implying that the person has skewed the statistics out of personal interest. Before making such accusations though shouldn't you be presenting at least some evidence? And I won't reply to MasterOfPuppets anylonger, because his argument simply doesn't hold water. "They are at least master and thus relevant to pro level statistics". I'm sorry but I am high master and my level of play is absolutely not relevant to the pro level play, they are world's apart. "At least master" is such a terrible parameter to use to decide wether or not a game is relevant because pro level (even low GM) and masters are worlds apart. The question mark is there for a reason. If the digging that Plansix did is correct, then he indeed does have a Terran bias. That is incredibly important, because he "filtered" the games. He filtered out qualifiers for tournaments, but used the tournament statistics. That is really important. If Terran is truly OP, then they'll dominate the qualifiers and will let in a lot of "bad " Terran players. Then those bad Terrans will lose to good Protoss and Zerg, and there we have the reason why the statistics are the way they are. Actually, the parameters used were the same as in every other month. In the first two months they showed Terran as the most winning race, this month they didn't. SAME PARAMETERS 100%. But you missed my point. If Terran is truly OP, then they'll dominate the qualifiers and will let in a lot of "bad " Terran players. Then those bad Terrans will lose to good Protoss and Zerg, and there we have the reason why the statistics are the way they are.
That could completely explain the statistics. I'll go through and add in the qualifier data later. Time to go to work. Ok? But that's exactly what would happen every month, regardless of what the person compiling these statistics does? Not only these statistics, the same logic could be applied to aligulac's win rates too. So then I ask, why even bother in the first place? (and the effect could be even amplified in aligulac's statistics seeing as they include all kinds of low level matches and qualifiers?) Ah, we reached a consensus. Why should they bother? Both ChaosTerran and Aligulac have a filtering process that may actually render their data invalid. Let's go back briefly to GSL October 2011, Code S. This was the height of GomTvT, where the final four were all Terrans. But if we go back, we see that the winrate looks okay, and if we look at the games in Code S, we see that in the 19 matches between Terran and Zerg, Terran won 10 and Zerg won 9. Looks great right? But 20 of the players in Code S were Terran, while only 7 were Zerg. And those 7 Zergs were an elite group (for the time). Nestea, Leenock, Losira, DRG, July, Coca and Zenio (Zenio was knocked out 0-2 in the group stage too, winning nothing). So we have elite Zergs taking on lesser Terrans and winning, which keeps the win rate close. But when the elite Zergs hit the elite Terrans, they were swiftly knocked out. In fact, of the 5 who made the round of 16, only two advanced, and one was against a Protoss opponent. None of them made the round of 4. So where were all the lesser Zergs, who's elimination would show the imbalance in the winrate? They didn't even qualify! And that is why filtering qualifiers can lead to problems! So looking purely at winrates is silly to be honest, and this example shows how filtering can skew results. Which is why I suggested this: Show nested quote +
What would really be a worthwhile exercise (and I'll probably contact the Aligulac guy about this later) would be to use the Aligulac rating of each player, and then compare winrates by race. Thus if a heavily favored Protoss player like Dear defeated QXC, it wouldn't necessarily show that Protoss was imbalanced. However if QXC (who is a big underdog) defeated Dear, it might indicate imbalance, especially if Terran underdogs were defeating Protoss favorites on a regular basis when they normally did not.
And TheBB delivered. http://aligulac.com/reports/And now we can see that when I selected Protoss I made a massive error.
This bottom graph looks almost identical like top one...
Also this not contain data from ladder so i does not affect you. Also you can change the race if you see your decision as "massive error"
|
On July 02 2013 14:06 BronzeKnee wrote:Show nested quote +On July 02 2013 13:58 Thrillz wrote:On July 02 2013 13:54 BronzeKnee wrote:On July 02 2013 13:51 Thrillz wrote:On July 02 2013 13:46 BronzeKnee wrote:On July 02 2013 13:40 Thrillz wrote:On July 02 2013 13:33 BronzeKnee wrote:On July 02 2013 02:59 SlixSC wrote:On July 02 2013 02:57 BronzeKnee wrote:On July 02 2013 02:55 SlixSC wrote:
[quote]
Actually, the parameters used were the same as in every other month. In the first two months they showed Terran as the most winning race, this month they didn't. SAME PARAMETERS 100%.
But you missed my point. If Terran is truly OP, then they'll dominate the qualifiers and will let in a lot of "bad " Terran players. Then those bad Terrans will lose to good Protoss and Zerg, and there we have the reason why the statistics are the way they are.
That could completely explain the statistics. I'll go through and add in the qualifier data later. Time to go to work. Ok? But that's exactly what would happen every month, regardless of what the person compiling these statistics does? Not only these statistics, the same logic could be applied to aligulac's win rates too. So then I ask, why even bother in the first place? (and the effect could be even amplified in aligulac's statistics seeing as they include all kinds of low level matches and qualifiers?) Ah, we reached a consensus. Why should they bother? Both ChaosTerran and Aligulac have a filtering process that may actually render their data invalid. Looking purely at winrates is silly to be honest. Which is why I suggested this:
What would really be a worthwhile exercise (and I'll probably contact the Aligulac guy about this later) would be to use the Aligulac rating of each player, and then compare winrates by race. Thus if a heavily favored Protoss player like Dear defeated QXC, it wouldn't necessarily show that Protoss was imbalanced. However if QXC (who is a big underdog) defeated Dear, it might indicate imbalance, especially if Terran underdogs were defeating Protoss favorites on a regular basis when they normally did not.
And TheBB delivered. http://aligulac.com/reports/And now we can see that when I selected Protoss I made a massive error. That doesn't say Protoss is horribly bad right now at all..........not to mention there are a few problems with that method, just like the previous ones. The method is far better than the old one, because it somewhat correct for player skill, while the old one does not. Perfect? No. But better. Anyway, I like your edit from Protoss isn't bad at all to Protoss isn't horribly bad right now. Protoss has been massively underperforming in SC2 compared to the other races, and right now high level Protoss players are losing more to lower level players from other races, than high level Terrans or Zergs do. But evidence show the contrary. Sorry but -11 says nothing to me, and aligulac has admitted that it isn't perfect, (see forgg nearly being #1, Soulkey not even top #10. Not jumping on the hysteria bandwagon, but nice try anyways. You literally just posted nothing. And I just said Aligulac wasn't perfect either. Why are you telling me it isn't? Then why are you drawing a conclusion when you only have a few points from one method on a hypothesis? I personally think it's pretty good right now. Pretty good relative to the past? Sure. Balanced... well you know the answer to that. -11 is the difference between how much worse Protoss players perform compared to Zergs. -48 is the difference between how much worse Protoss players perform compared to Terran. But -22 is how much Protoss players are underperforming.
That's it? You can see my skepticism here, one method from an admittedly imperfect method. Even still those numbers don't conclude imba, not to mention I'd rather wait as they seem to be trending upwards. You have not met enough criteria to conclude imba, and thank god Blizzard isn't looking to make big chances, other than those to tweak gameplay.
EDIT: Also just read : "However, as ratings catch up to the performances of the players, this chart will tend toward equilibrium, even if balance never changes." That certainly brings a new light when interpreting the ratings.
|
On July 02 2013 14:15 Thrillz wrote:Show nested quote +On July 02 2013 14:06 BronzeKnee wrote:On July 02 2013 13:58 Thrillz wrote:On July 02 2013 13:54 BronzeKnee wrote:On July 02 2013 13:51 Thrillz wrote:On July 02 2013 13:46 BronzeKnee wrote:On July 02 2013 13:40 Thrillz wrote:On July 02 2013 13:33 BronzeKnee wrote:On July 02 2013 02:59 SlixSC wrote:On July 02 2013 02:57 BronzeKnee wrote:
[quote]
But you missed my point.
If Terran is truly OP, then they'll dominate the qualifiers and will let in a lot of "bad " Terran players. Then those bad Terrans will lose to good Protoss and Zerg, and there we have the reason why the statistics are the way they are.
That could completely explain the statistics. I'll go through and add in the qualifier data later.
Time to go to work. Ok? But that's exactly what would happen every month, regardless of what the person compiling these statistics does? Not only these statistics, the same logic could be applied to aligulac's win rates too. So then I ask, why even bother in the first place? (and the effect could be even amplified in aligulac's statistics seeing as they include all kinds of low level matches and qualifiers?) Ah, we reached a consensus. Why should they bother? Both ChaosTerran and Aligulac have a filtering process that may actually render their data invalid. Looking purely at winrates is silly to be honest. Which is why I suggested this:
What would really be a worthwhile exercise (and I'll probably contact the Aligulac guy about this later) would be to use the Aligulac rating of each player, and then compare winrates by race. Thus if a heavily favored Protoss player like Dear defeated QXC, it wouldn't necessarily show that Protoss was imbalanced. However if QXC (who is a big underdog) defeated Dear, it might indicate imbalance, especially if Terran underdogs were defeating Protoss favorites on a regular basis when they normally did not.
And TheBB delivered. http://aligulac.com/reports/And now we can see that when I selected Protoss I made a massive error. That doesn't say Protoss is horribly bad right now at all..........not to mention there are a few problems with that method, just like the previous ones. The method is far better than the old one, because it somewhat correct for player skill, while the old one does not. Perfect? No. But better. Anyway, I like your edit from Protoss isn't bad at all to Protoss isn't horribly bad right now. Protoss has been massively underperforming in SC2 compared to the other races, and right now high level Protoss players are losing more to lower level players from other races, than high level Terrans or Zergs do. But evidence show the contrary. Sorry but -11 says nothing to me, and aligulac has admitted that it isn't perfect, (see forgg nearly being #1, Soulkey not even top #10. Not jumping on the hysteria bandwagon, but nice try anyways. You literally just posted nothing. And I just said Aligulac wasn't perfect either. Why are you telling me it isn't? Then why are you drawing a conclusion when you only have a few points from one method on a hypothesis? I personally think it's pretty good right now. Pretty good relative to the past? Sure. Balanced... well you know the answer to that. -11 is the difference between how much worse Protoss players perform compared to Zergs. -48 is the difference between how much worse Protoss players perform compared to Terran. But -22 is how much Protoss players are underperforming. That's it? You can see my skepticism here, one method from an admittedly imperfect method. Even still those numbers don't conclude imba, not to mention I'd rather wait as they seem to be trending upwards. You have not met enough criteria to conclude imba, and thank god Blizzard isn't looking to make big chances, other than those to tweak gameplay. EDIT: Also just read : "However, as ratings catch up to the performances of the players, this chart will tend toward equilibrium, even if balance never changes."
Reminds me of the arguments against global warming...
"Well there is all this imperfect science that we can poke holes in, and were going to ignore what is actually happening, and conclude nothing is happening. Excellent work everyone!"
I mean, there is just no evidence at all that Protoss hasn't performed as well as the other races...
![[image loading]](http://img690.imageshack.us/img690/2237/b2y1.png)
Anyways, what does the edit have to do with anything? The chart will tend toward equilibrium only if balance, player skill, maps and the metagame never change. In other words if everyone plays exactly how they are playing now, ratings will catch up, so it will no longer be an upset for Sjow to defeat Life, because Sjow will repeat that performance over and over, and he will climb, while Life will fall for the same reason. Then the underdog (Sjow), who skews the rating when he wins, is now the favorite and the ratings will no longer be skewed. If that happens, then either Sjow is actually better and deserves the rating, or there is in imbalance in the game.
Either way the chart does tend toward equilibrium if that happens.
|
On July 02 2013 14:13 keglu wrote:Show nested quote +On July 02 2013 13:33 BronzeKnee wrote:On July 02 2013 02:59 SlixSC wrote:On July 02 2013 02:57 BronzeKnee wrote:On July 02 2013 02:55 SlixSC wrote:On July 02 2013 02:52 BronzeKnee wrote:On July 02 2013 02:49 SlixSC wrote:On July 02 2013 02:43 BronzeKnee wrote:
Exactly as MasterOfPuppets said, you don't know what blackmail means.
This statement I made, is not blackmail:
More interesting is the fact that a guy named "ChaosTerran" came up with the statistics saying that Terran has a losing winrate in both mirrors... conflict of interests?
Yeah I'm sorry I misused the word blackmail. It's actually defamation, you are basically implying that the person has skewed the statistics out of personal interest. Before making such accusations though shouldn't you be presenting at least some evidence? And I won't reply to MasterOfPuppets anylonger, because his argument simply doesn't hold water. "They are at least master and thus relevant to pro level statistics". I'm sorry but I am high master and my level of play is absolutely not relevant to the pro level play, they are world's apart. "At least master" is such a terrible parameter to use to decide wether or not a game is relevant because pro level (even low GM) and masters are worlds apart. The question mark is there for a reason. If the digging that Plansix did is correct, then he indeed does have a Terran bias. That is incredibly important, because he "filtered" the games. He filtered out qualifiers for tournaments, but used the tournament statistics. That is really important. If Terran is truly OP, then they'll dominate the qualifiers and will let in a lot of "bad " Terran players. Then those bad Terrans will lose to good Protoss and Zerg, and there we have the reason why the statistics are the way they are. Actually, the parameters used were the same as in every other month. In the first two months they showed Terran as the most winning race, this month they didn't. SAME PARAMETERS 100%. But you missed my point. If Terran is truly OP, then they'll dominate the qualifiers and will let in a lot of "bad " Terran players. Then those bad Terrans will lose to good Protoss and Zerg, and there we have the reason why the statistics are the way they are.
That could completely explain the statistics. I'll go through and add in the qualifier data later. Time to go to work. Ok? But that's exactly what would happen every month, regardless of what the person compiling these statistics does? Not only these statistics, the same logic could be applied to aligulac's win rates too. So then I ask, why even bother in the first place? (and the effect could be even amplified in aligulac's statistics seeing as they include all kinds of low level matches and qualifiers?) Ah, we reached a consensus. Why should they bother? Both ChaosTerran and Aligulac have a filtering process that may actually render their data invalid. Let's go back briefly to GSL October 2011, Code S. This was the height of GomTvT, where the final four were all Terrans. But if we go back, we see that the winrate looks okay, and if we look at the games in Code S, we see that in the 19 matches between Terran and Zerg, Terran won 10 and Zerg won 9. Looks great right? But 20 of the players in Code S were Terran, while only 7 were Zerg. And those 7 Zergs were an elite group (for the time). Nestea, Leenock, Losira, DRG, July, Coca and Zenio (Zenio was knocked out 0-2 in the group stage too, winning nothing). So we have elite Zergs taking on lesser Terrans and winning, which keeps the win rate close. But when the elite Zergs hit the elite Terrans, they were swiftly knocked out. In fact, of the 5 who made the round of 16, only two advanced, and one was against a Protoss opponent. None of them made the round of 4. So where were all the lesser Zergs, who's elimination would show the imbalance in the winrate? They didn't even qualify! And that is why filtering qualifiers can lead to problems! So looking purely at winrates is silly to be honest, and this example shows how filtering can skew results. Which is why I suggested this:
What would really be a worthwhile exercise (and I'll probably contact the Aligulac guy about this later) would be to use the Aligulac rating of each player, and then compare winrates by race. Thus if a heavily favored Protoss player like Dear defeated QXC, it wouldn't necessarily show that Protoss was imbalanced. However if QXC (who is a big underdog) defeated Dear, it might indicate imbalance, especially if Terran underdogs were defeating Protoss favorites on a regular basis when they normally did not.
And TheBB delivered. http://aligulac.com/reports/And now we can see that when I selected Protoss I made a massive error. This bottom graph looks almost identical like top one... Also this not contain data from ladder so i does not affect you. Also you can change the race if you see your decision as "massive error"
Indeed.
|
On July 02 2013 14:24 BronzeKnee wrote:Show nested quote +On July 02 2013 14:15 Thrillz wrote:On July 02 2013 14:06 BronzeKnee wrote:On July 02 2013 13:58 Thrillz wrote:On July 02 2013 13:54 BronzeKnee wrote:On July 02 2013 13:51 Thrillz wrote:On July 02 2013 13:46 BronzeKnee wrote:On July 02 2013 13:40 Thrillz wrote:On July 02 2013 13:33 BronzeKnee wrote:On July 02 2013 02:59 SlixSC wrote:
[quote]
Ok? But that's exactly what would happen every month, regardless of what the person compiling these statistics does? Not only these statistics, the same logic could be applied to aligulac's win rates too. So then I ask, why even bother in the first place? (and the effect could be even amplified in aligulac's statistics seeing as they include all kinds of low level matches and qualifiers?) Ah, we reached a consensus. Why should they bother? Both ChaosTerran and Aligulac have a filtering process that may actually render their data invalid. Looking purely at winrates is silly to be honest. Which is why I suggested this:
What would really be a worthwhile exercise (and I'll probably contact the Aligulac guy about this later) would be to use the Aligulac rating of each player, and then compare winrates by race. Thus if a heavily favored Protoss player like Dear defeated QXC, it wouldn't necessarily show that Protoss was imbalanced. However if QXC (who is a big underdog) defeated Dear, it might indicate imbalance, especially if Terran underdogs were defeating Protoss favorites on a regular basis when they normally did not.
And TheBB delivered. http://aligulac.com/reports/And now we can see that when I selected Protoss I made a massive error. That doesn't say Protoss is horribly bad right now at all..........not to mention there are a few problems with that method, just like the previous ones. The method is far better than the old one, because it somewhat correct for player skill, while the old one does not. Perfect? No. But better. Anyway, I like your edit from Protoss isn't bad at all to Protoss isn't horribly bad right now. Protoss has been massively underperforming in SC2 compared to the other races, and right now high level Protoss players are losing more to lower level players from other races, than high level Terrans or Zergs do. But evidence show the contrary. Sorry but -11 says nothing to me, and aligulac has admitted that it isn't perfect, (see forgg nearly being #1, Soulkey not even top #10. Not jumping on the hysteria bandwagon, but nice try anyways. You literally just posted nothing. And I just said Aligulac wasn't perfect either. Why are you telling me it isn't? Then why are you drawing a conclusion when you only have a few points from one method on a hypothesis? I personally think it's pretty good right now. Pretty good relative to the past? Sure. Balanced... well you know the answer to that. -11 is the difference between how much worse Protoss players perform compared to Zergs. -48 is the difference between how much worse Protoss players perform compared to Terran. But -22 is how much Protoss players are underperforming. That's it? You can see my skepticism here, one method from an admittedly imperfect method. Even still those numbers don't conclude imba, not to mention I'd rather wait as they seem to be trending upwards. You have not met enough criteria to conclude imba, and thank god Blizzard isn't looking to make big chances, other than those to tweak gameplay. EDIT: Also just read : "However, as ratings catch up to the performances of the players, this chart will tend toward equilibrium, even if balance never changes." Reminds me of the arguments against global warming... "Well there is all this imperfect science that we can poke holes in, and were going to ignore what is actually happening, and conclude nothing is happening. Excellent work everyone!" I mean, there is just no evidence at all that Protoss hasn't performed as well as the other races... Anyways, what does the edit have to do with anything? The chart will tend toward equilibrium only if balance, player skill, maps and the metagame never change. In other words if everyone plays exactly how they are playing now, ratings will catch up, so it will no longer be an upset for Sjow to defeat Life, because Sjow will repeat that performance over and over, and he will climb, while Life will fall for the same reason. Then the underdog, who would skew the ratings, is now the favorite and the ratings will no longer be skewed. So the chart does tend toward equilibrium the longer an imbalance isn't fixed.
Nothing like global warming, it has mountains of support from multiple sources and data points, whereas you have a few, thus far one, and even that one doesn't say imba. Also why do I care about balance of the past? How does past balance affect current balance? How does WoL balance affect us now? The chart tends to equilibrium, regardless of balance, so how does that conclude it's imbalance, it merely states that protoss is underperforming with respect to their ratings. This isn't too surprising given the trend it's following in HotS. A race could very well be imbalance, but if they are not performing to their expect level, there will be a period of "underperformance" until going back to equilibrium. This isn't making sense to me: "So the chart does tend toward equilibrium the longer an imbalance isn't fixed" but you're using it to conclude imbalance in the first place aren't you?
|
Nevermind.
I enjoyed the discussion, but it is time for bed.
Goodnight all!
EDIT:
On July 02 2013 14:32 Thrillz wrote:
This isn't making sense to me: "So the chart does tend toward equilibrium the longer an imbalance isn't fixed" but you're using it to conclude imbalance in the first place aren't you?
When he wrote that I was worried it would be misconstrued, and it in fact has.
Maybe he will clarify it.
Re-read my post above regarding Sjow vs Life. Assume that Sjow won because of imbalance and that Life is a better player for the sake of this example. Now assume Sjow repeated his victory over and over. Eventually Sjow's rating would go up and while he wouldn't be the better player, his rating would show that he is the better player (because he won due to imbalance, so he isn't better). So if he won over and over, eventually it wouldn't be an upset because of how they are rated, despite Life being a better player.
Thus when it isn't an upset (when Sjow has won enough due to imbalance for his rating to be higher than Life's), it won't skew this chart, people will be performing as expected, and the chart will tend toward equilibrium if the imbalance isn't fixed and Sjow continues to win.
|
On July 02 2013 12:28 HuKPOWA wrote:Show nested quote +On July 02 2013 11:47 Lunareste wrote:On July 02 2013 11:41 HuKPOWA wrote:Problem with using this as a balance thing...is...protoss has highest win %...yet what has protoss won lately?  Dreamhack. 2nd at WCS Grand finals. 1ST PLACE at one event while others races have 2-3 like i said,,,
I keep track of the premier tournaments and I try to learn how far from perfection are we.
There were 10 premier HotS tournaments so far. Every race has a good representation in the ro8, P made it into ro8 26 times (80/3 = 26.3(3)) and that is pretty good.
P is missing 2 players in ro4 (expected: 13.3(3), actual: 11), 1 in the ro2 (expected: 6.6(6), actual: 5) and one winner (expected: 3.3(3), actual: 2).
That's it. If 4 games (2 in the ro8, 1 in the ro4 and 1 in the ro2) went in P favor it would be really great. Too bad HerO messed up at MLG Spring, he could have fixed it himself.
RO8 | RO4 | Ro2 | WIN
P T Z | P T Z | P T Z | P T Z
--------------------------------------------------
WCS S1 KR 2 2 4 | 1 1 2 | 0 1 1 | 0 0 1
IEM 7 WC 4 2 2 | 2 2 0 | 1 1 0 | 0 1 0
MLG winter 3 4 1 | 1 2 1 | 0 1 1 | 0 0 1
WCS EU s1 3 4 1 | 0 2 2 | 0 1 1 | 0 1 0
DH STKHML 2 1 5 | 2 0 2 | 1 0 1 | 0 0 1
WCS US S1 3 2 3 | 1 2 1 | 1 0 1 | 1 0 0
WCS S1 F 2 4 2 | 1 2 1 | 1 1 0 | 0 1 0
DHS 1 3 4 | 1 2 1 | 1 0 1 | 1 0 0
HSC 2 1 5 | 0 1 3 | 0 1 1 | 0 1 0
MLG Spring 4 2 2 | 2 1 1 | 0 1 1 | 0 1 0
--------------------------------------------------
total 26 25 29 |11 15 14 | 5 7 8 | 2 5 3
D -1 +2 -2 +1 -1 +1 -1 +1
|
On July 02 2013 14:35 BronzeKnee wrote:Nevermind. I enjoyed the discussion, but it is time for bed. Goodnight all! EDIT: Show nested quote +On July 02 2013 14:32 Thrillz wrote:
This isn't making sense to me: "So the chart does tend toward equilibrium the longer an imbalance isn't fixed" but you're using it to conclude imbalance in the first place aren't you? When he wrote that I was worried it would be misconstrued, and it in fact has. Maybe he will clarify it. Re-read my post above regarding Sjow vs Life. Assume that Sjow won because of imbalance and that Life is a better player for the sake of this example. Now assume Sjow repeated his victory over and over. Eventually Sjow's rating would go up and while he wouldn't be the better player, his rating would show that he is the better player (because he won due to imbalance, so he isn't better). So if he won over and over, eventually it wouldn't be an upset because of how they are rated, despite Life being a better player. Thus when it isn't an upset (when Sjow has won enough due to imbalance for his rating to be higher than Life's), it won't skew this chart, people will be performing as expected, and the chart will tend toward equilibrium if the imbalance isn't fixed and Sjow continues to win.
Exactly which means in your example the chart isn't a good indicator of balance at a given point, so why are you using the chart with -22 as "evidence" of imbalance when the chart is about performance. Those are separate entities. You are assuming imbalance in your example to cause changes in the ratings, but in your "evidence of toss up" you're using the chart to say there is a imbalance. Thus your evidence of the imbalance is underperformance?The statement:"However, as ratings catch up to the performances of the players, this chart will tend toward equilibrium, even if balance never changes exactly" states that the chart doesn't always prove balance or imbalance, so why are you trying to do that? Perfomance does not equal to balance, they are not one-to-one.You could be imba but under performing because you weren't as imba before, for a brief period of time, before stabilizing.
Protoss is UP, -22 rating
- No -22 is performance, not indicative of balance as they are not linear. As you already stated, you could be OP but in equilibrium. But you could also be UP but in equilibrium, or OP but underperforming, or UP but overperforming. Clearly you need a bigger picture of everything, and not just a few methods/data points.
Over time if imbalanced is not fixed it will like fine in performance chart
-Ok prove that there is an imbalance/toss UP that is causing this in the first place and that they remain that way even now, which you haven't done, as your evidence thus far IS the -22 underperformance in the chart and nothing else.
Even still a -22 underperformance from this month is not even that signifcant lol. You haven't been convincing and predictably Blizzard doesn't look to be doing anymore changes other than game play adjustment until there is mounting evidence such as Terran early WoL or Zerg late WoL. Sorry but they aren't crazy so deal with it.
|
On July 02 2013 13:55 FabledIntegral wrote:Show nested quote +On July 01 2013 20:05 Ghanburighan wrote:
As I posted in the DB thread, I'll repeat here:
A pretty low number of games. Around 300 per MU. I would have expected more fluctuation due to the numbers.
The Korean sub 100 MU's show's greater fluctuation. But still remains roughly balanced.
Nice to see all the MU's are nicely balanced, and that's roughly the feel you get from watching tournaments too. We've had a whole lot of P, T and Z win tournaments recently.
Thanks to ChaosTerran. Since when is 300 per MU a low number of games... or do you just mean relative to other months?
If you think back to TLPD winrate graphs, they generally had ~1000 international games, ~300 Korean games per MU. And everyone bemoaned that there aren't enough games in Korea to avoid fluctuation.
|
terran doing the worst in korea, wow..... and protoss seems to be on top
|
Protoss stomping down Terran in Korea! Nice
|
On July 01 2013 21:26 DarkLordOlli wrote:Show nested quote +On July 01 2013 21:24 MasterOfPuppets wrote:On July 01 2013 21:23 DarkLordOlli wrote:On July 01 2013 21:19 MasterOfPuppets wrote:On July 01 2013 21:02 DarkLordOlli wrote:
I find it weird that protoss is statistically doing so well without ever winning anything at the absolute highest level. Not counting WCS AM because HerO was just better than everybody else. Oh boy I lol'd so hard... Spoken like a true HerO fanboy... Top 8 were more or less in the same ballpark in terms of skill, with top 4 being fairly equal. Not the place to discuss this. Not that I'd even care about your opinion if we discussed it anywhere else. Then why should anyone care about your opinion that you shamelessly state as though it were an indisputable fact and then refuse to bring arguments in its favor?  I didn't say anyone had to care about it. I said I wasn't counting WCS AM because HerO was the best player there. Feel free to think differently. However the only one so far who took the time to be a jackass about my personal opinion was you. Good on you? Be proud. User was warned for this post
and in exactly way could you possibly disregard the results from hero to not be relevant to balance discussion if he was the best player there?
On July 01 2013 21:30 thezanursic wrote:Show nested quote +On July 01 2013 20:36 Tuczniak wrote:Wow, haha, that totally doesn't correspond from what I see in games  . I haven't thought zvp and pvt were so bad in korea, maybe those are just proleague maps. Zvt is strange, maybe terrans haven't accustomed to roachbane busts yet. They should be again above 55% fast though. 55% isn't a big imbalance.
thats right, its gigantic
On July 01 2013 23:03 GeNi wrote:
Stardust won with 2 base all ins pretty much the whole tournament. I can't count that. So pretty much HerO is the only toss that has won using legit mechanics and macro and whatever
balance includes anything and everything...
On July 02 2013 00:24 aldochillbro wrote:
I don't understand why anyone would include proleague in these statistics, it's just illogical. These stats either mean something or mean nothing. when you put a metagame and sniping based proleague in there it most definitely means nothing. Looking at the real statistics(posted in this thread), the game looks really balanced. it's weird to see protoss's actually keeping up
lol
|
And this doesn't even take into full effect the warp prism buff.
Typically in Korea there are less statistical variation in the matchups than in the rest of the world but this time around the gap is actually bigger.
|
So are people still crying about hellbats and how Terran is imba even after seeing this?
|
funny how everyone is complaining bout terran and yet they have the lowest winrate ....
|
How many people remember a time in War3 when undead in particular would race pick? Madfrog comes to mind. He severely disliked playing UvO so he would race switch to orc. The orc player, not wanting to play OvO would race pick to night elf. In turn, the undead player would go back to picking undead, prompting the orc player to go back to orc and thus starting the cycle of race pick all over again.
I'm already seeing some of it with KawaiiRice and Tilea. It seems TaJea is playing more protoss than terran these days. Yes I know a lot of pros off race anyway. But we are starting to see it in more than just ladder games.
|
|
|
|
|
|
EPT
