EPTBlizzard's SC2 race stats - Page 3
| Forum Index > SC2 General |
|
Pyrrhuloxia
United States6700 Posts
| ||
|
VonLego
United States519 Posts
On October 08 2010 08:14 Mutarisk wrote: 3% is a lot. Think about how many millions of games have been played. 3% doesn't become more just because it is extrapolated across millions of games. You're essentially saying "the small margin of error isn't small enough because the source is large as a whole," which makes very little sense. Now if you want to argue that 3% is just too large in the first place thats a different discussion. | ||
|
Assymptotic
United States552 Posts
| ||
|
Jerubaal
United States7684 Posts
On October 08 2010 08:36 Pyrrhuloxia wrote: Why is the thread about win percentages - is there a thread somewhere that I am not seeing that is talking about BARRACKS NOW REQUIRING A DEPOT and REAPER SPEED NOW REQUIRING A FACTORY? Those seem much more discussion-worthy. On the contrary, I find individual balance discussions pale in comparison to discussing the methods of balancing. In fact, I would almost call it a META balance discussion, but I have been sternly rebuked by others far my superior and told to keep my impertinent suggestions to myself. | ||
|
darmousseh
United States3437 Posts
"The system does account for this to an extent. After analyzing thousands of games and compiling trends based on player skill and race balance, our metrics will attempt to predict who the winner in a given match should be based on player skill and what it thinks is the favored race. These calculations will "correct" themselves over time as trends change. We do welcome feedback about the announced changes, but there are a considerable number of people outright disregarding the win/loss percentages we provided. Saying we conjured up these numbers to support our own ends, or that our numbers must be incredibly flawed, doesn't really add much to the discussion. The math was developed to disregard the way in which the matchmaking system attempts to keep players as close to a 50% win/loss ratio as possible. As stated in the blog, it's certainly not the only way we analyze balance, but it's very useful to us. I would really like to know their algorithm. It would deserve a placement in a mathematics thesis if this was possible. With the algorithm the way it's designed it is IMPOSSIBLE to determine a win/loss ratio without equal conditions or somewhat equal conditions. Predicting who the winner is is based on ELO (or glicko-2 or trueskill), determining balance does not derive from this at all. If only blizzard gave us access to a list of all games in an api then we could determine this ourselves. Note: Technically sc2 is a combination of 6 different games. pvp pvt pvz tvt tvz zvz so this makes it even harder. | ||
|
LOLtex
United States148 Posts
On October 08 2010 08:31 Waxangel wrote: Why is anyone talking about variance when blizzard has THE ENTIRE FUCKING SAMPLE OF EVERY GAME PLAYED? Because when you're calculating whether or not X race is going win against Y race, you're going to get a +/- variance in the model that can't be accounted for unless you literally include every possible variable in your model (ie. a meteor killing a player while he's playing the game). This is impossible, variance is present regardless of access to population data or not, because you're attempting to pigeonhole something that normally can't be explained via modeling into a nice and neat mathematical equation. You do the best you can, but it's not going to be 100% perfect. | ||
|
StimCraft
United States144 Posts
-Don't quote a top-level foreigner to improve your flawed logic, -pull a sample size of 10 (and one that isn't even close to real top 10), -use %'s with 10 users, -absolutes -cussing -flame incoherently...etc. -digress from what OP wants Quit being full of yourself and contribute to discussion or flame somewhere else. I doubt you actually feel a major imba problem to be frank. ================================== 3% is significant with the degree of information available. How the present the information is the problem. *adjusted for player skill is quite questionable, but probably means if standardized for ELO of ____ example: [Race] V [Race] of 8,000 users for ELO of [Range]. They could have said okay Bronze ELO is 0-1000 or something Silver 1000-1300 etc. and let's pull 60% percentile of each division +/- 50 ELO to show a point. One can do so many things to "prove" a point in stats. We all are just frustrated at their deceptive behavior. Being a Zerg, in ZvP I do feel I need to be a lot better to gain a small advantage. This is the only statistic that I feel is close to the current meta-game (the point of bringing up the stat). I still enjoy ZvP the most, though. I think the PvT is the biggest problem, though, it says 50/50. Hmm. | ||
|
smegged
Australia213 Posts
| ||
|
Ruthless
United States492 Posts
The skill that blizzard has hidden or not still is a function of the the race being played and the players true skill. Defining either of these two quantities or how they relate is too complicated to do and is also somewhat subjective. I just dont see how stats that assume that a players rating reflects their skill can possibly be used to determine balance of the game. Now If you could quantify those two traits then racial strength would depend on true skill and would be a complicated function that would incorporate how the balance between the races shift at as true skill varies and as experience with the race varies. | ||
|
KentHenry
United States260 Posts
Note: Technically sc2 is a combination of 6 different games. pvp pvt pvz tvt tvz zvz so this makes it even harder. Mirror match ups are not counted because they are balanced for either player. | ||
|
Mutarisk
United States153 Posts
On October 08 2010 08:39 VonLego wrote: 3% doesn't become more just because it is extrapolated across millions of games. You're essentially saying "the small margin of error isn't small enough because the source is large as a whole," which makes very little sense. Now if you want to argue that 3% is just too large in the first place thats a different discussion. You are correct. What I am arguing is that as the number of games increases, the number should become closer to zero difference which would represent "balance" (assuming they have an accurate measure of skill) For the number of games that have been played, I think 3% is rather large. | ||
|
Adeny
Norway1233 Posts
On October 08 2010 08:35 mutantmagnet wrote: The conceit of all of you who complain Blizzard can't make judgments without knowing how the stats are generated even though they are the ones who designed them are ridiculous. Get over it. Blizzard knows the numbers better than you do. Whether or not they are making a good decisions on those numbers is open to debate. Wait are you trying to disagree with me? Blizzard would prefer to make themselves look good. You're assuming that a multi-billion dollar company with hundreds of thousands of stockholders would not try to shift the statistics in a way that would make them look better. I'm well aware that Blizzard knows the numbers better than I do, THAT'S EXACTLY WHY THEY ARE USELESS. Only if they openly released the methods they used to arrive at those statistics would they be relevant in any way. - Never does it state if the entire pool of players are used, for all you know it's only EU, only US etc. - "Accouting for skill" could mean any number of things, but in essence it probably means "we tried every comparison and this is the one that makes us look best". I'm not trying to claim I know more about Blizzards numbers than they do, but blinding trusting their very own statistics is extremely naive, especially with the lack of information. | ||
|
refraxion
Canada88 Posts
On October 08 2010 07:47 KingAce wrote: Someone needs to go into Blizzard, and kick everyone on the balance team out. If these are the numbers they use to balance the game, it makes perfect sense why they keep nerfing protoss. I think TL needs to make it's own balance notes and force them down blizzards throat. What are they doing? If you think Protoss deserves a buff, I laugh. | ||
|
TaKemE
Denmark1045 Posts
Aint that how the system work? | ||
|
Adeny
Norway1233 Posts
On October 08 2010 08:48 StimCraft wrote: Adeny, don't flame. TL is not flame ground. -Don't quote a top-level foreigner to improve your flawed logic, -pull a sample size of 10 (and one that isn't even close to real top 10), -use %'s with 10 users, -absolutes -cussing -flame incoherently...etc. -digress from what OP wants Quit being full of yourself and contribute to discussion or flame somewhere else. I doubt you actually feel a major imba problem to be frank. WHAT?! Wait, are you sure you are reading my post? Cussing where? Flame incoherently, where? And digress? My post is very on point, let me break this down for you. The point I was trying to prove was that any statistic is complete garbage if the method used to generate said statistics is released in full. I'm using a sample size of 10 and %'s with 10 users to show that you can make factual statistical statements and make them look exactly the way you want them to. Real top 10 or not it was just an example, but now I understand how Darwin must have felt. You're blindly assuming Blizzard is 100% correct, I'm demanding evidence to support their claims. Again, I'm not "full of myself", I am rationally and logically explaining why the numbers actually mean nothing, so stop your higher-than-thou attitude, please. | ||
|
Happy Frog
Australia490 Posts
What we're also aware of is that, while the numbers don't necessarily support the need for zerg changes across all leagues, the feedback from the community as well as our own play experience tells us that improvements are necessary to make zerg matchups feel and play better. Regardless of how they compile the data for racial stats, they've shown the ability to ignore them in the same post they released them, evidenced by that quote and all 5 previewed changes being Zerg buffs. | ||
|
xephon
Canada38 Posts
On October 08 2010 07:29 Klive5ive wrote: Win % in Diamond (accounting for player skill) 49.6% win rate for Protoss when fighting Terran. 52.8% win rate for Protoss when fighting Zerg. 49.6% win rate for Terran when fighting Zerg. the number tells ZvT and PvT is fine than, why BZ want to buff zerg ?? very strange | ||
|
eloist
United States1017 Posts
On October 08 2010 07:41 nemahsys wrote: I want to know how they are "accounting for player skill" when coming up with these numbers. Not really sure how you go about doing that. They could determine the true relative ranking of players by looking at just the results from mirror matches. Not hard at all to do. | ||
|
universalwill
United States654 Posts
the only statistics that indicate the imbalance are the very top ranked terrans, because there are almost no zergs good enough to match them (had to say almost no because you all would go WHAT ABOUT FRUITDEALER). if they looked at the very top 50 players from each region, i think they'd get a better sense of balance than "oh well our matchmaking system that is designed to place players against players they will lose 50% of their games to is telling us that they are, in fact, losing 50% of their games. game balanced, guys!" | ||
|
tianGO
Argentina591 Posts
This could affect things like early marauder pressure in tvp just because they want to change the reaper timings. | ||
| ||
