EPTAlso, there's this thing on the ladder called MMR. Because it will have you play people on your MMR, you can be a lot better than someone and play Protoss, but beat them because you're just a lot better than them. How the ladder works is that it is supposed to keep things 50%. The ladder is by no means a demonstration of balance.
Blizzard Blog: Balance Snapshot - Page 7
| Forum Index > SC2 General |
|
Deleted User 183001
2939 Posts
Also, there's this thing on the ladder called MMR. Because it will have you play people on your MMR, you can be a lot better than someone and play Protoss, but beat them because you're just a lot better than them. How the ladder works is that it is supposed to keep things 50%. The ladder is by no means a demonstration of balance. | ||
|
maddogawl
United States63 Posts
On September 23 2011 02:42 headbus wrote: 7 Zergs, 5 Protoss in GSL code S. That should be enough to convince people that terran is stronger. Its too small of a sample size to make any accusations about balance. Out of 32 players all of them could be a single race although unlikely and still the game could be balanced. What matters more is the top players in a much larger quantity ( 100, 1000, etc ) even 100 is too small of a sample set, so looking at all players in Code S, Code A, and Code B with w/l ratios would yield better information because in theory all these players should be relatively close in skill. | ||
|
DNA61289
United States665 Posts
On September 23 2011 01:55 Brotocol wrote: Well said. I also want to point out that matchmaking makes all winrates tend toward 50%. * The only case where this does not apply is tournaments, where matchmaking is not done algorithmically. * The definition of balance means that, for 2 players of the same skill level, the game should be perfectly even. How do we isolate the skill level variable? We can't. But the closest we can do is make some assumptions: (1) There is less variability (lower std. deviation) in the highest tournaments and (2) The more up to date a given scene's metagame is, the more viable it is. Thus, we really should be looking at GSL for the closest picture of balance that we can get, as it's the closest situation in terms of reflecting equal skill levels (tapering off on the high end, and thus not as highly variable as other tournaments/scenes), and metagame advancement. PS. A common misconception of "balance for the highest levels" is that imbalance doesn't affect lower leagues. It affects any game with equal skill level players, where one has to work harder than the other to achieve the same result. The entire "balance only affects the highest leagues" is wrong. It affects high leagues consistently, but can also happen, albeit inconsistently, at lower leagues. Say, for instance, take 2 silver league players who are playing at skill levels 100 and 95, respectively (hypothetical values of course). Meaning that one is outplaying the other. And yet, the latter opponent wins. That is imbalance. Real life example: 1-1-1 happening at lower leagues. That is imbalanced if an inferior player beats a superior one. Verdict: This looks like damage control, as it misrepresents these stats as "raw stats," whereas in reality, matchmaking has influenced them heavily. This is really well thought out response, I can't help but come to the same conclusion. | ||
|
nemo14
United States425 Posts
On September 23 2011 02:24 Brotocol wrote: Misconception. What it really means to "balance for the highest level of play" is that the strongest play must be balanced, but this should automatically scale down to all levels because skill levels are relative. If it's balanced for players that have skill levels of 150, then it should automatically ensure balance for skill level 5. You're wrong. A diamond terran who can't remember to split his bio will lose to infestors constantly. A gold zerg who can't remember to scout properly will lose to a 6rax every time he sees it. A platinum protoss who can't remember to watch the minimap will lose to a ling/nydus all-in basically every time he encounters it. Not every strategy that silver-leaguers can pull off can be defended with silver-league map awareness, micro, etc. | ||
|
jinorazi
Korea (South)4948 Posts
| ||
|
Zaphid
Czech Republic1860 Posts
On September 23 2011 02:48 nemo14 wrote: You're wrong. A diamond terran who can't remember to split his bio will lose to infestors constantly. A gold zerg who can't remember to scout properly will lose to a 6rax every time he sees it. A platinum protoss who can't remember to watch the minimap will lose to a ling/nydus all-in basically every time he encounters it. Not every strategy that silver-leaguers can pull off can be defended with silver-league map awareness, micro, etc. That's exactly what makes you improve, people's skill isn't static | ||
|
Kazeyonoma
United States2912 Posts
blizz doesn't give info to help quell the masses, masses cry foul and claim blizzard doesn't know what it's doing. You can't go by just pure GM rankings. that's only 200 people per region, statistically it is too low to be able to balance upon, you have to include masters because at least in korea, lots of master level players are borderline pro or GM level but the 200 max system prevents them from being moved up until someone in GM falls out. Take what you want away from this blog post, but to flame blizzard for even giving us any information is horseshit. You can be skeptical, or you can try to interpret it into other subsets of data, but you have to believe that blizzard has ALL of the numbers. Our stats, even from the best of our community, is based off of partial data mining utilizing tools like sc2ranks (love you!) to try to collect the best data we can, but blizzard literally has it ALL. You have no clue why the neural parasite change happened. Maybe in their data, they can do a search where it says, in PvZ when games get past X time, and Infestors are out. Games with Neural parasite are won 90% of the time by zerg, games without are closer to 50%. See, a statistic like that, something that only blizzard can easily filter, would indicate there is some sort of disparity in PvZ when Neural is used. That's not to say they immediately made the change like the blog post said. It's possible this initial inference lead them to then further investigate the "possible problem" and look at it closer, looking further at variances of data, and upon their research and despite shifting meta game changes, the PvZ + Neural statistics have remained outside of the 55/45 win rate range they like to see. So they then hit the balance team to figure out a suitable fix, apply to PTR and test how that alters their data. It's easy to look at JUST these overall win rates and see PvZ isn't lop sided why nerf NP blizz? and QQ heavily, but we don't have the data, and we will never have all of the data. They have spreadsheets that are probably so sophisticated we couldn't dream of imitating them. As a former theorycrafter for WoW we used to try to create simulation spreadsheets to mimic that of blizzard, and several times we came close, but even blizzard employees would post that while impressed upon our reverse engineering of data to create our own spreadsheets, ours were fundamentally flawed for an overall viewpoint, but useful for our specific needs. SC2 has probably more possible iterations and changes due to unit compositions and macro/micro than in WoW, so I can only imagine their findings to be even harder to balance and their spreadsheets even more sophisticated. So please, while it's easy to try to judge blizzard upon this release and their recent patch changes, the blog itself says these numbers ARE NOT THE ONLY REASONING FOR CHANGES. /rant off | ||
|
Grummler
Germany743 Posts
On September 23 2011 02:44 JudicatorHammurabi wrote: Also, there's this thing on the ladder called MMR. Because it will have you play people on your MMR, you can be a lot better than someone and play Protoss, but beat them because you're just a lot better than them. How the ladder works is that it is supposed to keep things 50%. The ladder is by no means a demonstration of balance. Try to read what is written above these numbers. Are you seriously thinking that Blizzard forgot about their own matchmaking system? That would make you look very stupid. Actually i think this might be true, so i am going to quote the important part for you, in case you have trouble finding it: What's an adjusted win percentage? While the math behind calculating an adjusted win percentage is extremely complex, an adjusted win percentage can be summed up as the 'true' win percentage of a given race, produced by removing the skewing effects of the matchmaker and factoring in player skill. Source: See OP. | ||
|
nemo14
United States425 Posts
On September 23 2011 02:51 Zaphid wrote: That's exactly what makes you improve, people's skill isn't static Exactly. Game balance does not scale evenly down the skill curve, but when you can overcome that obstacle by just getting better then it isn't an issue. | ||
|
Serashin
235 Posts
On September 23 2011 02:38 maddogawl wrote: You do realize Blizzard does realize this. Theres a video that they did at Blizzcon where they discuss how they analyze balance, and they say themselves that they can't look at straight w/l ratios. They have behind the scenes skill values for each player that they analyze as well. Check out the video if you can. Blizzard designers are a lot smarter than people on the forums give them credit for. i figured how roughly their "skill values" work and they are nowhere near smart picked. Just because they dont wanna tell the comunity how exactly its measured ( what is of course wrong measurement) doesnt make it smart from them. I give you a comparision there are tests for IQ and stuff but hell they are wrong measurement aswell. There are roughly okish measurement but great differences are never point out able. Also how they can have a "oh so smart measurement" if their basic game maths and ladder system is wrong to show balance of the game. Also what Tyler said its nothing visible that forces them to show their statics. And i have it roughly figured after playing 8 k ladder games sir. All what a healthly mind learns after 8 k ladder games or their meassurement of 10 k games to be capable of knowing every little bit of the game is ... Congratulations you found out that this game is nowhere near balanced. Im honestly wanting that achievment and a Trollface as Avatar for it. | ||
|
tztztz
Germany314 Posts
there you have it. i don't get the all the blizzard hate. what do you want them to do? do you really believe you know more than blizzard about their matchmaking systen and about balancing the game? | ||
|
BronzeKnee
United States5219 Posts
On September 23 2011 02:39 BronzeKnee wrote: This is a great post. Obviously if Blizzard was balancing the game based in part on these stats, they would not have balanced it all in the way that they did. | ||
|
Sabu113
United States11078 Posts
Does explain the perceptions of zerg and Terrans on the forums. Wierd. I can think of a few stories that might explain it but they're all a bit too hand wavey. | ||
|
crms
United States11933 Posts
On September 23 2011 02:39 Rob28 wrote: Yeah, in the bronze league maybe. look at master/GM stats, they are better indicators since top level guys don't lose as much due to micro/macro or just inexperience... Read.. the.. tables.. The numbers are much better for Z in lower leagues. When you look at Master/GM as you suggested.. P was favored over Z 57% in NA and EU. It was 'balanced' *cough* in KR at 49%. | ||
|
Asshat
593 Posts
On September 23 2011 02:19 Zombie_Velociraptor wrote: And then how many of these 12/12 are advancing to Ro16, compared? l0l Actually most of the zerg players have been advancing so far. There will be 3 terrans at best in the ro8 and there could be as many as 6 zergs. 2 guaranteed for each race. | ||
|
EnderSword
Canada669 Posts
On September 23 2011 02:24 Brotocol wrote: Misconception. What it really means to "balance for the highest level of play" is that the strongest play must be balanced, but this should automatically scale down to all levels because skill levels are relative. If it's balanced for players that have skill levels of 150, then it should automatically ensure balance for skill level 5. That's non-sensical. Certain elements of the game require more and less skill, and there may be imbalance at different levels of play. Take a few examples... The MarineKing: A Terran and Zerg are fighting, and 10 banelings begin rolling towards 20 marines on creep. At Bronze level, 20 marines are going to die At Plat-Dia type level, some marines are likely going to die at MKP level, The 20 marines will be split into 20 groups and the banelings will do nothing. The HongUn: A Zerg and Protoss are fighting, 40 Roaches attack 30 Blink Stalkers At Bronze level, both armies basically get destroyed, the fight is simply determined by the concave. At Plat-Dia level, The Stalkers will edge it out, taking a lot of damage, blinking away some but losing many. At HongUn level all 30 stalkers will be blinked away perfectly as their shields run out, while HongUn reads a book about cashews. The DRG 6 Colossus and Gateway units march out as a deathball At Bronze level, the Protoss rolls over a horribly positioned zerg player, despite trapping his own zealots behind his Stalkers. 6 Infestors are killed with full energy because he A-moved them with the army. At Plat-Dia level, Several Colossus are NPed and FG coats the army...Zerg likely loses, but send Protoss packing back home to rebuild to fight again. At DRG level, you don't have a Deathball, you died at the 9-10 minute mark to a ling/roach faceroll supported by Infestors and ITs There's units and tactics the scale well, and those that do not. That's why it's so difficult to actually balance for everyone. | ||
|
ThaZenith
Canada3116 Posts
You can see those large differences in the Mast+GM category, but they don't mean anything because Master's can't actually play the game that well, and GM can be iffy. (on the NA/Europe servers more than Korean) | ||
|
Ownos
United States2147 Posts
On September 23 2011 02:22 Liquid`Tyler wrote: I'm kind of interested why Blizzard wants us to know this. As far as I know, there's no pressure on them to release these numbers. No one is waiting on them. And no one in the community is going to use them for anything. Blizzard simply wants us to know them... why? Because protoss balance in the GSL/Korea is a hot topic right now? I'm not really confident in these numbers. If I switched to terran and started doing really well, the system will just think I improved a lot since skill is measured only by win rates. But this is all guessing at a system we don't fully have the details of. Another example, if I didn't switch and due to a horrible build that terrans are using I start losing. But I eventually even out vs bad terrans who aren't aware of the build or can't execute well or just plain bad. Then my win rates will even out again. This might explain: "It’s fairly common, for example, for a new strategy or build order to skew the numbers in favor of a particular race for a brief period, until the metagame catches up and the counter strategies spread through the community. " Would it really just be a metagame shift or me facing worse opponents. These stats don't tell me that, hence: Still, while they may be interesting, it’s important to emphasize that these numbers aren't the last word in our balance analysis. People are reading too much into these stats and too eager to hop onto the very popular Blizzard hate train. But it is fairly obvious damage control as even TL is allowing threads and articles to cite balance issues. | ||
|
Welmu
Finland3295 Posts
On September 23 2011 01:51 R3N wrote: PvZ 57% on both NA and EU. I thought 'toss is UP? There are tons and tons of lower level Zerg players in EU ladder at least. I somehow manage to play almost every day against Z player who I havent seen before, but almost never against unknown T or P | ||
|
nemo14
United States425 Posts
On September 23 2011 02:59 Sabu113 wrote: So they aren't going to fix toss soon. Interesting. Does explain the perceptions of zerg and Terrans on the forums. Wierd. I can think of a few stories that might explain it but they're all a bit too hand wavey. When all anyone knows how to do is either cheese or slowly macro up to 200 and then a-move while frantically watching the battle, protoss is pretty good. | ||
| ||
