EPTI call bullshit on those winrates though, I would like to know how David Kim computes these.
Feb 10: Balance Test Map Coming - Page 41
| Forum Index > SC2 General |
|
ZenithM
France15952 Posts
I call bullshit on those winrates though, I would like to know how David Kim computes these. | ||
|
Squat
Sweden7978 Posts
On February 12 2014 01:36 RampancyTW wrote: ...Yes, but because it's built to have balanced winrates, any racial imbalances will be clearly displayed because one race being overpowered or underpowered with respect to (an)other race(s) will be reflected in the matchup winrates. If the matchup-wide winrates are all close to 50% (which they are), then the game is pretty damn balanced. No, that is incorrect. In case of racial imbalances, the system will adjust for imbalances by matching the overpowered race with a increasingly stronger opponents until 50/50 parity is achieved. | ||
|
Plansix
United States60190 Posts
| ||
|
Ctone23
United States1839 Posts
On February 12 2014 00:34 Pirfiktshon wrote: You can't Balance a Game off of Bottom Tier brackets.... as many times as ppl say that it makes me cringe and puts a pit in my stomach thinking is the general populace this dull? You can't balance a game off of someone with 20 apm and just clicks on stuff and watches fights LOL Trust me I used to be one of those in BW when i was like 9 LOL You can only balance a game off of those that can use units to their maximum efficiency but the way Blizzard made SC2 is that alot of times you can't as a human perform what blizzard has set out for maximum efficiency for units.... IE the Automation 2000 marines vs Banes not a single marine dies and kills like 40 banes LOL or where a relatively small group of lings massacres like 40 tanks... as a human you can only do so much so fast even if you are korean..... So balancing really should only come into play when a certain strategy is used with complete ease and the counterpart that is reacting is facing an EXTREME unsurmountable task that can only be accomplished if the otherside makes a mistake.... LOL which TvP feels like that ALOT especially with blink stalker.... I don't know if you watch many T players stream but most of the time when terran defend blink all in at high GM level its because Protoss MADE HUGE HUGE HUGE mistakes.... May I suggest some paragraph breaks and fewer LOL statements... | ||
|
TheDwf
France19747 Posts
On February 12 2014 01:36 RampancyTW wrote: ...Yes, but because it's built to have balanced winrates, any racial imbalances will be clearly displayed because one race being overpowered or underpowered with respect to (an)other race(s) will be reflected in the matchup winrates. If the matchup-wide winrates are all close to 50% (which they are), then the game is pretty damn balanced. Nope, because players of the overpowered race are carried beyond their true skill level, and balance winrates by losing to superior players they shouldn't be matched against on an even basis. | ||
|
Rollora
2450 Posts
On February 12 2014 00:41 zelderan wrote: Balanced on what? Or just to answer your argument:The game is balanced. There are minor things here and there but overall Blizzard has the balance thing figured out. Most of these changes are in response to the Korean meta (msc+blink allins, wm not being strong enough, zerg AA bad etc.) *scratching head* no it is not. Simply said: no. No it isnt. And I can tell you why: it CANT be 100% balanced. The 3 races are too different from each other. The warpgate-mechanic, the larva mechanic, the mules. There are so many things that are TOTALLY different, so you can never 100% balance it. Just like, there will never be a 100% bugfree game while the code is several 100k lines big. It is just not possible. And moreover: Blizzard balances on winrates (wich are quite off atm, like 24% vor T in TvP when we look at Code A/Code S -> you wouldn't say it is just a minor balance thing). So they say, if a race has 50% winrate, it is balanced. Well that is ONE WAY you can say a game is balanced. But lets look how these 50% are reached. When a race or the current meta for a race, is inferior, pro-players work around that. Find another strat lets say. Maybe cheese, maybe something more complicated. And if a counter is more complicated or takes more effort than the initial tactics by an opponent, but overall it leads to 50% winrate, is it balanced. Is it balanced if countering a strat always takes more effort? How do you measure this "effort". APM? Thinking involved (how do you measure that)? For instance: Protoss have their warpin mechanics, so they can "beam in" Tier 1 armies wherever they want, whereas T has to move across the whole map. This is why so many P players can (successfully) play cheeses and allins and always could. So especially when it is about movement and macro, in early stages P could be called an easier race. But maybe because of the thinking process "which one is the right mix/composition of units" involved, and the perfect timing you need to hit, protoss is maybe balanced - even if it will never display in a chart (like APM). Similar things also happen in TvZ or Mirrormatches btw (where in ladder, I always curse my opponent for going for the eays tactics :D ) Current Problem in PvT is, that Protoss is very save in early game, and Protoss lategame (with colossi, MSC, Storms etc) is very very hard to break/takes more effort for the T it feels. Only in Midgame it can be easily broken (thats why we see so many SCV Pulls in midgame). And if you play cautious as T, you may be behind very fast - cause Toss can play greedy and still is in no real danger (Overcharge etc). But you cannot play greedy as T as well, cause Toss has many ways of punishing that (Oracle = almost always several kills if controlled, Blink-Allin etc). This is not intented to be a whine btw, this just points out why Pro Teams in Proleague don't nominate many Terrans (besides the 3-4 we see (Maru,TY, Flash..) who are the only ones able to deal with P. And there is also a reason, why there are so many eliminated T in Code A -> S. I do NOT say this game is inbalanced, it is Blizzards call to make and that of the Pro-Gamers. And maybe it is just the latter ones, we need to learn from. Hope they can find ways to deal with this, non-pro's can do as well, or ladder will get more and more frustrating (getting killed by a player with a third of APM (yeah, i know, APM doesnt say anything, but to me it shows of the effort (at least partially) put into the game) because of one better engagement or whatever. Game should get more like BW again: getting a lead through many small advantages in play and finish opponent off, not losing all the lead through a single stupid fight (that is over far too fast often). Or these current 45 min games... oh my... On February 12 2014 01:54 TheDwf wrote: Nope, because players of the overpowered race are carried beyond their true skill level, and balance winrates by losing to superior players they shouldn't be matched against on an even basis. That is basically what the problem with winrates is: they just show a part of the equation. But what, if it takes a better player for one race to reach that 50% compared to the other race. That is the problem we are seeing more and more. One player outplaying the other, but still losing in the end. Or generally the feeling, that one player plays it superiour, does more for the game... and still it is closer in the end than it "should"(personal feelings, attention) be. On February 12 2014 01:53 Plansix wrote: It doesn't. Cause the MMR is based on the assumption that the races are balanced on that level as well - which they aren't.At some point in every ladder session, you do play people close to your skill level. The MMR system seeks to find you harder and harder opponents until you lose, but isn't going to match a masters player against a silver leaguer just to make that happen. The system has to have some idea of how skilled players are or it wouldn't work at all. So it can always happen, that a one trick pony player, who has no clue about the game and only has a third of your apm or whatever, kills you with his allin/cheese which he learned to hide well (and you are no pro, so you cannot scout everything (btw even pro's are unaware of many cheeses)). So when I lose, and do some replay analyzis, just like day9 tought me  and many other players and casters and whatever I've learned over the past years, I see lots of players who do just the one thing right and me doing lots of things better, faster, etc but in the end I lose of maybe one misclick (which they count on, and which can happen, cause I'm rusty, cause frustration led to me not playing a lot anymore). So the MMR system does help, but it isn't the real thing either | ||
|
Squat
Sweden7978 Posts
On February 12 2014 01:53 Plansix wrote: At some point in every ladder session, you do play people close to your skill level. The MMR system seeks to find you harder and harder opponents until you lose, but isn't going to match a masters player against a silver leaguer just to make that happen. The system has to have some idea of how skilled players are or it wouldn't work at all. Yes it will, if you keep winning the system will promote you until you end up in a league with players that you can be expected to beat roughly 50% of the time. It is true that you may initially play against players either below or above your level, but over time it will even out. The only place on ladder where any observations about balance can be made is the absolute top of GM. | ||
|
Pirfiktshon
United States1072 Posts
Well, Kim the D claims that he can compute winrates "regardless of skill and race strength amirite olol". That might be the balanced winrates zelderan is referring to. I call bullshit on those winrates though, I would like to know how David Kim computes these. I was going to write this in my last post but I figured I wouldn't hahahaha Honestly there is no TRUE way to do that.... you could say well any match that is outside of a given league... so Silver vs Silver..... or possibly in the same percentage of a league like top 10% of Masters vs top 10% of masters..... Which doesn't work anyway because of what DWF pointed out ... if you are playing the race that puts you on a pedastool than it will raise you up past what your true skill level is.... which OUR definition of skill level is measured in Multitasking Micro macro decision making and game sense.... Edit: OH and Scouting ability.... which plays into EVERYTHING..... | ||
|
RampancyTW
United States577 Posts
On February 12 2014 01:54 TheDwf wrote: ...But this isn't the case, because your artificially inflated skill level will put you up against better players of BOTH other races, thus leading to a situation where winrates within matchups will be skewed to bring a racial population to 50%. The only way this isn't the case is if a race is completely equally overpowered in both matchups.Nope, because players of the overpowered race are carried beyond their true skill level, and balance winrates by losing to superior players they shouldn't be matched against on an even basis. | ||
|
Rollora
2450 Posts
On February 12 2014 02:03 RampancyTW wrote: ...But this isn't the case, because your artificially inflated skill level will put you up against better players of BOTH other races, thus leading to a situation where winrates within matchups will be skewed to bring a racial population to 50%. The only way this isn't the case is if a race is completely equally overpowered in both matchups. or underpowered. mind you | ||
|
RampancyTW
United States577 Posts
| ||
|
Squat
Sweden7978 Posts
On February 12 2014 02:03 RampancyTW wrote: ...But this isn't the case, because your artificially inflated skill level will put you up against better players of BOTH other races, thus leading to a situation where winrates within matchups will be skewed to bring a racial population to 50%. The only way this isn't the case is if a race is completely equally overpowered in both matchups. If a race is OP and the other two are not, it will likely be OP against both of them. Even if you lose more vs one of the other two races, overall increased win rates will put you into brackets with players you would not otherwise be playing against. Please stop saying the ladder having even winrates is indicative of good balance, it's nonsense. | ||
|
ZenithM
France15952 Posts
On February 12 2014 02:03 RampancyTW wrote: ...But this isn't the case, because your artificially inflated skill level will put you up against better players of BOTH other races, thus leading to a situation where winrates within matchups will be skewed to bring a racial population to 50%. The only way this isn't the case is if a race is completely equally overpowered in both matchups. Yes, that's what we're saying here. If a race was so OP that it completely crushes the other 2 equally easily, there would be no way of knowing from the general winrates. One way to go is to look at GM race representation, I think. | ||
|
Pirfiktshon
United States1072 Posts
If a race is OP and the other two are not, it will likely be OP against both of them. Even if you lose more vs one of the other two races, overall increased win rates will put into brackets with players you would not otherwise be playing against. Please stop saying the ladder having even winrates is indicative of good balance, it's nonsense. ^ This +1 | ||
|
WombaT
Northern Ireland26510 Posts
Not to say disregard statistics entirely, but sometimes just watch the games and some issues are patently obvious from even relatively small sample sizes. | ||
|
MstrJinbo
United States1251 Posts
On February 12 2014 02:23 Wombat_NI wrote: I know it's not really the flavour of the month, but with some things you have to go with your gut rather than use statistics which are proving increasingly difficult to use. For example, the weeklies have a lot of Korean big hitters stomping mediocre foreign players in the early rounds, which can disguise say, an issue with TvP winrates by creating a bunch of TvP victories when in reality it's that the Korean Terran is just far better than their P opponent. Not to say disregard statistics entirely, but sometimes just watch the games and some issues are patently obvious from even relatively small sample sizes. Quibbling over which meaningless statistic is better than another is like chocolate. It will always be a flavor around here. | ||
|
Rollora
2450 Posts
On February 12 2014 02:10 RampancyTW wrote: you are 100% right. I just wanted to say, winrates alone would not uncover if a race was generally OP or UP from the very beginning. There have to other stats to prove, that one race cannot be played with less or more effort to win 50%, -> which leads us to the discussion again, that it was always possible that one race was OP overall and other races' players had to play harder to reach that 50% (or the other way round with UP)True. The main point is though that a race has to have equal power relations with both of the other races for imbalances to not be reflected in the winrates. On February 12 2014 02:15 ZenithM wrote: Yes, that's what we're saying here. If a race was so OP that it completely crushes the other 2 equally easily, there would be no way of knowing from the general winrates. One way to go is to look at GM race representation, I think. True (and it is good that way). And that is why Blizzard fixes always one MU at a time. | ||
|
Plansix
United States60190 Posts
| ||
|
Rollora
2450 Posts
On February 12 2014 02:30 Plansix wrote: It's almost like Blizzard needs to rely on both what is going on in the professional scene and watch the ladder for any potential trends. No one source of information is going to give them the complete picture , but they can't just ignore any source if information either. yeah. It is also that EU GM games look a lot different than KR GM Games | ||
|
Plansix
United States60190 Posts
On February 12 2014 02:33 Rollora wrote: yeah. It is also that EU GM games look a lot different than KR GM Games GM has always been a questionable way to view balance because it's such a small sample set and players own multiple accounts. Watch region is going to have its own reasons for being for having the players it has in it. And those reasons may not reflect on balance at all. | ||
| ||
