|
Perhaps someone can explain the correlation between a balanced match-up and a 50% win rate to me, because I do not think aiming for a 50/50 win rate is how to make a game balanced. A perfectly balanced game to me means the best player should win every time with small random variables being the smallest factor in victory or defeat.
I will try to make an analogy that might apply to this case that I hope makes sense. I hope this isn't racist..
Lets say African Americans = Protoss and Asians = Zerg and the game of choice is basketball. After a week of games played between Africans and Asians we find Africans win 60% of the games and Asians win 40% of the games. Is Basketball a balanced game? By Blizzards definition the answer is no; therefore we need to:
a) change the basketball court (map) to make it more favored towards asians
b) nerf Africans (take away their ability to jump)
c) buff Asians (give them bouncy shoes)
Lets say A, B, and C are all applied to the game of basketball and now after a week of games between black people and Asians, we find the games are 50/50. So is basketball balanced now that we've taken away the natural skillsets of Africans and improved Asians artificially? By Blizzards thought process, yes; even though the Africans are having to fight against an unfair advantage.
This is what's going on exactly in my opinion whenever they balance match-ups and it doesn't make sense to me. They see a 50/50 and think, "This is a balanced game now, on to the next project."
On a side note; imagine PvZ comes to a point where Protoss is scouted by Zerg overlords to check for a certain 2 base timing. That 2 base timing looks exactly like a macro build also. Zerg can prepare for a 2 base timing and hold and win, or guess wrong and get outmacroed. Zerg can also macro up hoping its a macro build and stay even, or macro up and die to a 2 base all-in. Its a 50/50 chance in this scenario and therefore leads towards a 50/50 win rate for both sides. How is this balanced? Are we just admitting that is what Starcraft 2 has become?
|
God damn it, ffs the blizzard staff even goes to the effort to read and post in this thread and you post shit like that.
Fucking make me rage.
(Page break might remove the context, was in response to my previous post on the last page)
|
On January 10 2014 13:12 Gofarman wrote:Show nested quote +On January 10 2014 12:52 T.O.P. wrote:On January 10 2014 12:18 itsjustatank wrote:Please keep in mind these are not straight-up win percentages. They’re win percentages with player skill factored out. When we grab win/loss data for balance purposes, we categorize each game with 2 different variables per side: one being player skill and other being race strength. So by factoring the player skill out, we are able to more accurately check how each race is doing at each skill level. Without an explanation as to what this actually means, statistically speaking, this sounds like like linguistic flourish justifying cherrypicking one's data to one's advantage, a component of lying with statistics. Elucidation would be greatly appreciated, because I have to assume this isn't the case. Perhaps because good players play more games, if you just present the data unaltered good players would make influence the winrates more than their proportion in the population.
But knowing this is a post by the sc2 team, I wouldn't be surprised if they did some bullshit. What do you mean by that? the SC2 team is full of shit and looking to mislead? That little star doesn't give you a pass to be rude when you feel like it, if anything that shit shouldn't be tolerated from someone who is a 'quality poster'... Ingrate.
Chill out you self-righteous dick. Way to blow a little comment way out of proportion lol
|
Hong Kong9164 Posts
On January 10 2014 12:52 T.O.P. wrote:Show nested quote +On January 10 2014 12:18 itsjustatank wrote:Please keep in mind these are not straight-up win percentages. They’re win percentages with player skill factored out. When we grab win/loss data for balance purposes, we categorize each game with 2 different variables per side: one being player skill and other being race strength. So by factoring the player skill out, we are able to more accurately check how each race is doing at each skill level. Without an explanation as to what this actually means, statistically speaking, this sounds like like linguistic flourish justifying cherrypicking one's data to one's advantage, a component of lying with statistics. Elucidation would be greatly appreciated, because I have to assume this isn't the case. Perhaps because good players play more games, if you just present the data unaltered good players would make influence the winrates more than their proportion in the population.
Here's the thing though: all they do is present win-rates, and they already break down the win-rates by tier. Ignoring or weighting data based on their stated filters in this context is only done assuming you want to shape the data to your advantage.
|
On January 10 2014 13:16 Survivor61316 wrote:Show nested quote +On January 10 2014 13:12 Gofarman wrote:On January 10 2014 12:52 T.O.P. wrote:On January 10 2014 12:18 itsjustatank wrote:Please keep in mind these are not straight-up win percentages. They’re win percentages with player skill factored out. When we grab win/loss data for balance purposes, we categorize each game with 2 different variables per side: one being player skill and other being race strength. So by factoring the player skill out, we are able to more accurately check how each race is doing at each skill level. Without an explanation as to what this actually means, statistically speaking, this sounds like like linguistic flourish justifying cherrypicking one's data to one's advantage, a component of lying with statistics. Elucidation would be greatly appreciated, because I have to assume this isn't the case. Perhaps because good players play more games, if you just present the data unaltered good players would make influence the winrates more than their proportion in the population.
But knowing this is a post by the sc2 team, I wouldn't be surprised if they did some bullshit. What do you mean by that? the SC2 team is full of shit and looking to mislead? That little star doesn't give you a pass to be rude when you feel like it, if anything that shit shouldn't be tolerated from someone who is a 'quality poster'... Ingrate. Chill out you self-righteous dick. Way to blow a little comment way out of proportion lol
I'm self righteous eh?
lol
A little respect is well deserved.
|
Well its positive to note that he has thoughts right?
Please Activisionlizzards, finish the game so that I have it and don't have to care anymore when it is done. Its the only game that I am interested and not because I play it a lot or watch it a lot. I just dislike to have an uncompleted game on my PC.
And ofc its easier to balance a finished product. It has to suck to balance around and you know there is the last DLC coming that will screw it all again.
|
On January 10 2014 13:17 Gofarman wrote:Show nested quote +On January 10 2014 13:16 Survivor61316 wrote:On January 10 2014 13:12 Gofarman wrote:On January 10 2014 12:52 T.O.P. wrote:On January 10 2014 12:18 itsjustatank wrote:Please keep in mind these are not straight-up win percentages. They’re win percentages with player skill factored out. When we grab win/loss data for balance purposes, we categorize each game with 2 different variables per side: one being player skill and other being race strength. So by factoring the player skill out, we are able to more accurately check how each race is doing at each skill level. Without an explanation as to what this actually means, statistically speaking, this sounds like like linguistic flourish justifying cherrypicking one's data to one's advantage, a component of lying with statistics. Elucidation would be greatly appreciated, because I have to assume this isn't the case. Perhaps because good players play more games, if you just present the data unaltered good players would make influence the winrates more than their proportion in the population.
But knowing this is a post by the sc2 team, I wouldn't be surprised if they did some bullshit. What do you mean by that? the SC2 team is full of shit and looking to mislead? That little star doesn't give you a pass to be rude when you feel like it, if anything that shit shouldn't be tolerated from someone who is a 'quality poster'... Ingrate. Chill out you self-righteous dick. Way to blow a little comment way out of proportion lol I'm self righteous eh? lol A little respect is well deserved.
Sorry I forgot you were the respect police...and that you didn't have a sense of humor because he was clearly being facetious.
|
On January 10 2014 13:16 itsjustatank wrote:Show nested quote +On January 10 2014 12:52 T.O.P. wrote:On January 10 2014 12:18 itsjustatank wrote:Please keep in mind these are not straight-up win percentages. They’re win percentages with player skill factored out. When we grab win/loss data for balance purposes, we categorize each game with 2 different variables per side: one being player skill and other being race strength. So by factoring the player skill out, we are able to more accurately check how each race is doing at each skill level. Without an explanation as to what this actually means, statistically speaking, this sounds like like linguistic flourish justifying cherrypicking one's data to one's advantage, a component of lying with statistics. Elucidation would be greatly appreciated, because I have to assume this isn't the case. Perhaps because good players play more games, if you just present the data unaltered good players would make influence the winrates more than their proportion in the population. Here's the thing though: all they do is present win-rates, and they already break down the win-rates by tier. Ignoring or weighting data based on their stated filters in this context is only done assuming you want to shape the data to your advantage.
That assumes that your goal is to mislead the people you are presenting the data to, which David and Blizzard have almost no reason to do. In fact, they have numerous reasons to provide us with the most accurate information possible.
Unless you believe that DKim is trying to mislead us and sell us on the idea that the game is balance when he knows that one race is over powers.
|
So many bans in this thread...
1 week of statistics is hardly worthless. But at the same time, they're not really useful either I think. That's because thousands of ladder games are played each day, so they have at least a thousand matches worth of data for each day. But factoring out player skill... What the hell does that mean?
|
Hong Kong9164 Posts
On January 10 2014 13:20 Plansix wrote:Show nested quote +On January 10 2014 13:16 itsjustatank wrote:On January 10 2014 12:52 T.O.P. wrote:On January 10 2014 12:18 itsjustatank wrote:Please keep in mind these are not straight-up win percentages. They’re win percentages with player skill factored out. When we grab win/loss data for balance purposes, we categorize each game with 2 different variables per side: one being player skill and other being race strength. So by factoring the player skill out, we are able to more accurately check how each race is doing at each skill level. Without an explanation as to what this actually means, statistically speaking, this sounds like like linguistic flourish justifying cherrypicking one's data to one's advantage, a component of lying with statistics. Elucidation would be greatly appreciated, because I have to assume this isn't the case. Perhaps because good players play more games, if you just present the data unaltered good players would make influence the winrates more than their proportion in the population. Here's the thing though: all they do is present win-rates, and they already break down the win-rates by tier. Ignoring or weighting data based on their stated filters in this context is only done assuming you want to shape the data to your advantage. That assumes that your goal is to mislead the people you are presenting the data to, which David and Blizzard have almost no reason to do. In fact, they have numerous reasons to provide us with the most accurate information possible. Unless you believe that DKim is trying to mislead us and sell us on the idea that the game is balance when he knows that one race is over powers.
I question all use of statistics, and it leads me to question people's motives when I sense bad methodology. I don't make the assumption that all data and interpretations of data presented to me are kosher, precisely because I know how easy it is to shape any data set to say whatever it is I want it to say and phrase it in such a way that few people will question it.
|
Fiddler's Green42661 Posts
I thought we went over this with WoL. Just because PvZ was around a 50% winrate in a matchup, that didn't mean something wasn't wrong with the matchup.
|
On January 10 2014 13:20 Plansix wrote:Show nested quote +On January 10 2014 13:16 itsjustatank wrote:On January 10 2014 12:52 T.O.P. wrote:On January 10 2014 12:18 itsjustatank wrote:Please keep in mind these are not straight-up win percentages. They’re win percentages with player skill factored out. When we grab win/loss data for balance purposes, we categorize each game with 2 different variables per side: one being player skill and other being race strength. So by factoring the player skill out, we are able to more accurately check how each race is doing at each skill level. Without an explanation as to what this actually means, statistically speaking, this sounds like like linguistic flourish justifying cherrypicking one's data to one's advantage, a component of lying with statistics. Elucidation would be greatly appreciated, because I have to assume this isn't the case. Perhaps because good players play more games, if you just present the data unaltered good players would make influence the winrates more than their proportion in the population. Here's the thing though: all they do is present win-rates, and they already break down the win-rates by tier. Ignoring or weighting data based on their stated filters in this context is only done assuming you want to shape the data to your advantage. That assumes that your goal is to mislead the people you are presenting the data to, which David and Blizzard have almost no reason to do. In fact, they have numerous reasons to provide us with the most accurate information possible. Unless you believe that DKim is trying to mislead us and sell us on the idea that the game is balance when he knows that one race is over powers.
He used the same line of reasoning to state why BL/Infestor was balanced.
|
On January 10 2014 13:12 ZenithM wrote:Show nested quote +On January 10 2014 13:03 Brian333 wrote:On January 10 2014 12:54 ZenithM wrote:On January 10 2014 12:51 Brian333 wrote:On January 10 2014 12:48 ZenithM wrote:
I don't get how they would ever statistically know if something like "Protoss being 2 times stronger than both the other races" happens.
I mean, you just have no way of knowing, right? Except from watching how the actual game plays out :D You'd know because at a pro-level, Protoss would have like a 99% win rate vT and vZ. Hmm, no? In a bubble and in a scene built from a blank slate, you would be right. However, if suddenly Protoss was twice as strong as it is now, what would the win-rates of current Protoss suddenly be? They wouldn't lose. I guess the win-rate would still be around 50% given that eventually there would be a lot of bad protoss "pros" that emerge and drag the win-rate down. What I meant is that if Protoss is statistically significantly stronger than the other races (so, not a ridiculous number like 2 times as I've said in my first post :D), you probably couldn't tell from the pro-level, because there are just too few of those players. So whereas a 55% winrate on the whole ladder is concerning (like, significantly concerning), 55% tournament wins is completely fine. Obviously if suddenly you double the zealot's life and damage, things are not going to go well at all at high levels, but I was talking more of a slight edge that, while small, may be still there ;D I don't know how they detect that edge. They seem to pretend like they can ("race strength factored out" and all that shit), but so far we don't know how they do it. Edit: And btw, when DK pulls out the "skill adjusted winrates", they're never worse than 45-55%. Which is kinda weird because it would mean that Starcraft 2 was always almost balanced. We all know that isn't right :D
The way I see it, on a pro level, small differences in race strength will manifest itself in win-rates because the relevant pros all practice the same amount and with similar support. As long as there is no reason why one race would have significantly more talented individuals than another, racial strength is the reason for higher and lower win-rates.
On the level of ladder, the pure volume of games and the size of the player base balance things out to a point where racial strength is evident. Given the amount of players, you can assume that the players from the different races all have an equal amount of hard-working players that really practice and strive to improve, casuals that just play for fun, and everything in between. Unless there is a reason the more talented / skilled individuals are all playing a certain race, racial strength is the reason for higher and lower win-rates.
|
On January 10 2014 13:23 stuchiu wrote:
I thought we went over this with WoL. Just because PvZ was around a 50% winrate in a matchup, that didn't mean something wasn't wrong with the matchup.
It's almost like the balance team has learned nothing... Huh.
|
On January 10 2014 13:22 itsjustatank wrote:Show nested quote +On January 10 2014 13:20 Plansix wrote:On January 10 2014 13:16 itsjustatank wrote:On January 10 2014 12:52 T.O.P. wrote:On January 10 2014 12:18 itsjustatank wrote:Please keep in mind these are not straight-up win percentages. They’re win percentages with player skill factored out. When we grab win/loss data for balance purposes, we categorize each game with 2 different variables per side: one being player skill and other being race strength. So by factoring the player skill out, we are able to more accurately check how each race is doing at each skill level. Without an explanation as to what this actually means, statistically speaking, this sounds like like linguistic flourish justifying cherrypicking one's data to one's advantage, a component of lying with statistics. Elucidation would be greatly appreciated, because I have to assume this isn't the case. Perhaps because good players play more games, if you just present the data unaltered good players would make influence the winrates more than their proportion in the population. Here's the thing though: all they do is present win-rates, and they already break down the win-rates by tier. Ignoring or weighting data based on their stated filters in this context is only done assuming you want to shape the data to your advantage. That assumes that your goal is to mislead the people you are presenting the data to, which David and Blizzard have almost no reason to do. In fact, they have numerous reasons to provide us with the most accurate information possible. Unless you believe that DKim is trying to mislead us and sell us on the idea that the game is balance when he knows that one race is over powers. I question all use of statistics, and it leads me to question people's motives when I sense bad methodology. I don't make the assumption that all data and interpretations of data presented to me are kosher, precisely because I know how easy it is to shape any data set to say whatever it is I want it to say and phrase it in such a way that few people will question it.
Ok, but I am going to ask you the same question, why would they do that? You don't just imply that their data is incorrectly gathered, but they do so with purpose with the intent to mislead. Do you really think they want an imbalanced game? I don't really see any good reason for them to "cook the numbers" to make the game appear balanced.
|
Hong Kong9164 Posts
A reminder too, these are only descriptive statistics. They in no way make valid any inferences as to statistical significance of any hypothesis.
|
Hong Kong9164 Posts
On January 10 2014 13:25 Plansix wrote:Show nested quote +On January 10 2014 13:22 itsjustatank wrote:On January 10 2014 13:20 Plansix wrote:On January 10 2014 13:16 itsjustatank wrote:On January 10 2014 12:52 T.O.P. wrote:On January 10 2014 12:18 itsjustatank wrote:Please keep in mind these are not straight-up win percentages. They’re win percentages with player skill factored out. When we grab win/loss data for balance purposes, we categorize each game with 2 different variables per side: one being player skill and other being race strength. So by factoring the player skill out, we are able to more accurately check how each race is doing at each skill level. Without an explanation as to what this actually means, statistically speaking, this sounds like like linguistic flourish justifying cherrypicking one's data to one's advantage, a component of lying with statistics. Elucidation would be greatly appreciated, because I have to assume this isn't the case. Perhaps because good players play more games, if you just present the data unaltered good players would make influence the winrates more than their proportion in the population. Here's the thing though: all they do is present win-rates, and they already break down the win-rates by tier. Ignoring or weighting data based on their stated filters in this context is only done assuming you want to shape the data to your advantage. That assumes that your goal is to mislead the people you are presenting the data to, which David and Blizzard have almost no reason to do. In fact, they have numerous reasons to provide us with the most accurate information possible. Unless you believe that DKim is trying to mislead us and sell us on the idea that the game is balance when he knows that one race is over powers. I question all use of statistics, and it leads me to question people's motives when I sense bad methodology. I don't make the assumption that all data and interpretations of data presented to me are kosher, precisely because I know how easy it is to shape any data set to say whatever it is I want it to say and phrase it in such a way that few people will question it. Ok, but I am going to ask you the same question, why would they do that? You don't just imply that their data is incorrectly gathered, but they do so with purpose with the intent to mislead. Do you really think they want an imbalanced game? I don't really see any good reason for them to "cook the numbers" to make the game appear balanced.
Having the public on your side, especially in this community, when you are trying to change things in the game is politically useful. You make assumptions about what is rational and not rational that come off as being pretty naive.
|
On January 10 2014 13:25 Brian333 wrote:Show nested quote +On January 10 2014 13:12 ZenithM wrote:On January 10 2014 13:03 Brian333 wrote:On January 10 2014 12:54 ZenithM wrote:On January 10 2014 12:51 Brian333 wrote:On January 10 2014 12:48 ZenithM wrote:
I don't get how they would ever statistically know if something like "Protoss being 2 times stronger than both the other races" happens.
I mean, you just have no way of knowing, right? Except from watching how the actual game plays out :D You'd know because at a pro-level, Protoss would have like a 99% win rate vT and vZ. Hmm, no? In a bubble and in a scene built from a blank slate, you would be right. However, if suddenly Protoss was twice as strong as it is now, what would the win-rates of current Protoss suddenly be? They wouldn't lose. I guess the win-rate would still be around 50% given that eventually there would be a lot of bad protoss "pros" that emerge and drag the win-rate down. What I meant is that if Protoss is statistically significantly stronger than the other races (so, not a ridiculous number like 2 times as I've said in my first post :D), you probably couldn't tell from the pro-level, because there are just too few of those players. So whereas a 55% winrate on the whole ladder is concerning (like, significantly concerning), 55% tournament wins is completely fine. Obviously if suddenly you double the zealot's life and damage, things are not going to go well at all at high levels, but I was talking more of a slight edge that, while small, may be still there ;D I don't know how they detect that edge. They seem to pretend like they can ("race strength factored out" and all that shit), but so far we don't know how they do it. Edit: And btw, when DK pulls out the "skill adjusted winrates", they're never worse than 45-55%. Which is kinda weird because it would mean that Starcraft 2 was always almost balanced. We all know that isn't right :D The way I see it, on a pro level, small differences in race strength will manifest itself in win-rates because the relevant pros all practice the same amount and with similar support. As long as there is no reason why one race would have significantly more talented individuals than another, racial strength is the reason for higher and lower win-rates. On the level of ladder, the pure volume of games and the size of the player base balance things out to a point where racial strength is evident. Given the amount of players, you can assume that the players from the different races all have an equal amount of hard-working players that really practice and strive to improve, casuals that just play for fun, and everything in between. Unless there is a reason the more talented / skilled individuals are all playing a certain race, racial strength is the reason for higher and lower win-rates.
Yeah but on the ladder, it's all about MMR-based match-making. I want to know how they get their skill adjused winrates, because otherwise, the winrates are obviously all at 50% except on extreme leveled leagues.
On January 10 2014 13:25 ssxsilver wrote:Show nested quote +On January 10 2014 13:20 Plansix wrote:On January 10 2014 13:16 itsjustatank wrote:On January 10 2014 12:52 T.O.P. wrote:On January 10 2014 12:18 itsjustatank wrote:Please keep in mind these are not straight-up win percentages. They’re win percentages with player skill factored out. When we grab win/loss data for balance purposes, we categorize each game with 2 different variables per side: one being player skill and other being race strength. So by factoring the player skill out, we are able to more accurately check how each race is doing at each skill level. Without an explanation as to what this actually means, statistically speaking, this sounds like like linguistic flourish justifying cherrypicking one's data to one's advantage, a component of lying with statistics. Elucidation would be greatly appreciated, because I have to assume this isn't the case. Perhaps because good players play more games, if you just present the data unaltered good players would make influence the winrates more than their proportion in the population. Here's the thing though: all they do is present win-rates, and they already break down the win-rates by tier. Ignoring or weighting data based on their stated filters in this context is only done assuming you want to shape the data to your advantage. That assumes that your goal is to mislead the people you are presenting the data to, which David and Blizzard have almost no reason to do. In fact, they have numerous reasons to provide us with the most accurate information possible. Unless you believe that DKim is trying to mislead us and sell us on the idea that the game is balance when he knows that one race is over powers. He used the same line of reasoning to state why BL/Infestor was balanced.
Exactly. If I recall correctly the winrates he pulled out at that time were about of the same order as the current winrates.
|
On January 10 2014 13:09 Brian333 wrote:My point is your comment about unfavorable conditions and unequal skill is not true. Prime might be an exception due to how small their roster is but you are acting like these coaches are sending out Protoss with sub-par win-rates in map-specific PvT for shits and giggles. Who knows how good some of their PvT win-rates are on those specific maps in internal practice? The coach, not you.
I said all but 1 Terran being fielded has been Code A or better at one point or better and that 37% of the losses come from Prime (who everyone picked to finish dead last b/c of many reasons). Nowhere did I mention any of what you suggest I said.
|
On January 10 2014 11:55 CutTheEnemy wrote:
On EU, 24% of masters players are terran now compared with 35% and 38% for zerg and protoss. How can he say its balanced considering this? His appeal to win percentages within leagues is highly misleading.
He's also been speaking for years as though he's ignorant of our main complaint- its isn't balance per se, its how hard and stressful it is to play terran and win. We know terrans can win once they go pro, but most of us aren't capable of sustaining the serious damage to our personal relationships, grades, hands and paychecks it takes in order to play the race competitively.
^ I think this is the point
It's reflected everywhere that while there may be balance among the skilled players, people hate playing terran because it's too frustrating and hard. Although this sounds like QQ, it's shown by how many Terrans are still competing.
Some examples:
- Terran being under-represented in GM for several seasons in a row now
- Terran being under-represented in Proleague (source: http://wiki.teamliquid.net/starcraft2/2014_Proleague/Statistics)
8/41 players that played in Proleague so far are Terran. Once again, the same counter-argument is there, "it's a small sample size." Well, it's going to continue being a small sample size.
While I understand this small sample size argument... the truth is the entire SC2 pro-scene is a small sample size of the select few hundred players, and Terran is definitely under-represented in the pro scene. Whether win rates back it up or not, not many like playing Terran at the pro-level.
Like I said, it's fine if you're extremely good at the game (shown by the god-mode skilled terrans like Maru, Flash, TY all doing well).
The reason why we're seeing a lot of 50% in general is because you are matched with people your MMR. If one race struggles, the only impact this makes is every player of that race has lower MMR. It has nothing to do with "who is the better player" - in the game's perspective, the better player is the one with higher MMR, but we all know that's not true. The MMR system automatically adjusts your opponent so that you have a roughly ~50% win ratio.
The numbers are not cooked, just Ladder win ratios are a very misleading indicator of balance.
|
|
|
|
|
|
EPT
