|
On October 13 2014 23:48 antiRW wrote:Show nested quote +On October 13 2014 23:44 Thieving Magpie wrote:On October 13 2014 23:40 antiRW wrote:On October 13 2014 23:30 Thieving Magpie wrote:On October 13 2014 23:16 Big J wrote:On October 13 2014 22:41 Thieving Magpie wrote:On October 13 2014 22:29 HEADD wrote:
I see TVZ strongly unbalanced right now.With recent Thor buff and widow mine buff without reverting hellbats buff.
I think this will balance the game:
1-revert helbatt buff-This prevent zerg from having good early economy+forcing him to banelings
2-revert thor buff-this is just stupid buff.Complete anti muta buff.Not necessary at all and pretty much forcing zerg not use mutas at all.
3-widow mine slightly nerf-Not in same state 3 moths ago, but slightly nerf, because i seen so many times 1 widow mine kills 20banelings in 1 hit.This is just not right.1 little mistake by zerg and ist game ending.
I think 70-80% of current splash damage would be OK.
Terrans will have still decent widow mines, but not that good that they can wipe 20 banelings with 1 hit and zergs still need micro.
Then why does Zerg win more often than Terran does according to aligulac? Sorry, but most of the stats of aligulac are not interesting for the highest level of balance. Even if it is only 50% it is a big problem if we judge balance by this. Here are the stats of all the tournaments minus WCS America that took place in the last two days: Seacraft Weekly #24: TvZ 7–1 (87.50%) Notable TvZs: ??? WECG Qualifier Korea Playoffs: TvZ 3–2 (60.00%) Notable TvZs: Dark vs Ty Gfinity 1v1 Cup #31 TvZ 0–0 (0%) Notable TvZs: --- go4sc2 Cup Europe #413 TvZ 8–9 (47.06%) Notable TvZs: YoDa 1-2 TargA Dragon 0–2 Tefel PxL-Lan #41 TvZ 1–4 (20.00%) Notable TvZs: ??? --> 25/35 (71%) games are not even worth mentioning when discussing balance on a high level in that periode of time. It's complete bonkers to watch aligulac stats and then make so exact statements as "Zerg is winning more than Terran because 48%". Yes, in the aligulac coverage it is. If we don't arbitrarily cut by tournament participation and take a different "arbitrary" cut, namely WCS Premier Leagues, this is what we get: WCS AM: 27-15 (64%) WCS EU: 25 - 21 (54%) GSL Code S: 19-13 (59%) And I do not like the practice of "these numbers that don't agree with my conclusions are wrong" bible thumper mentality. You take all of the data, all of the different degrees of the data, and have each aspect of the data inform different parts of the conclusion. Individual player performance shows what is possible. Totality of player performances shows what is happening. You have to integrate what is possible with what is happening to be able to make conclusions about anything. ... which was exactly his point. Namely that statements such as "nothing is broken because aligulac's 48%" are not useful, because aligulac along is not reflective of the full picture. You don't choose one over the other. Aligulac shows that zerg, as a race, is doing well. And the individual stats shows that 5-6 zerg players are not doing as well as the overall stats. That does not mean that Zerg, as a race, is weak. Only if you think that the success of zerg as a race is measured by Aligulac. You can make a case for that, but one can just as well use other metrics: Overall or M/GM ladder stats, WCS tournaments only, GSL. Obviously we don't want to have that debate every single time and the honest approach would be to always contrast the different indicators.
I'm all in favour of getting more statistical tools in play to find out what's going on. But 17 sets of games is ridiculously few.
I'll show you why. (this is the double-digit-dance)
Let's assume that all those sets were Bo3 for the sake of argument. That means we had a maximum of 51 games (all sets ended 2-1).
Let's assume that T>Z, so the winrate is ~60% (30 games were won by T, 21 by Z). Now, let's say history was different by the smallest amount, one T who previously beat Z, lost to T. This is a change of a single game in a set as all were 2-1. Previously it was T 2, Z1 , now it's Z 2, T 1. So the numbers are 29, 22, which makes it 57%. That's a huge swing from a single game in a set.
Assuming that we have one game where a player clicked on an OC with his entire army, that one click swings the winrate by an averaged 3% That's insane. It shows nothing about balance that this person clicked on the OC.
In fact, it takes only 6 games to have a different outcome to go from 60% to full parity (actually, full parity is at 5.5 games, 6 is slightly Z favoured, 5 is slightly T favoured). And this is if we assume that all games were 2-1. If some of the sets were sweeps, the numbers take even larger swings per game.
And that's why we shouldn't be making conclusions based on small numbers of games.
|
On October 13 2014 23:25 Thieving Magpie wrote:Show nested quote +On October 13 2014 23:04 Meavis wrote:On October 13 2014 22:41 Thieving Magpie wrote:
Zerg win more often than Terran does according to aligulac? you realize how silly that sounds right? What's silly about it? Statistically Zerg is ahead right now by a few percentage points while some terrans are doing better than others. You can't really say that a match is imbalanced when you don't even have enough statistic to stake that claim. Especially when the cherry picked evidence contradicts the totality of the data set. You can say that top end Zerg players have been unable to adapt to top end Terran play, because the stats does show that. But saying that Zerg, as a race, is obviously weak due to a buff to a unit that wasn't OP the last time it had the stats it has right now is silly. Sometimes, it really is just the players and not the race. Heck, depending which 17 games of my ladder history you get I'm either really bad vs Zerg or really good vs Zerg. You have to include as many aspects and dimensions of the data to be able to make conclusions from them. Go back to what I said months ago when terrains complained that TvP was imba, and I pointed out that Maru was able to keep a good winrate vs Protoss suggesting that it is possible for anyone to do well vs Protoss, it's just that it's very hard. And I argued that you don't make balance patches just because things are hard. A patch is not a quality of life fixer.
I don't think your conclusion is... good? Maru is able to beat Protoss, therefore Protoss can be beaten. That's ok, in my opinion. But for ZvT it's the other way round: lower level Zerg manage to beat Terrans, but higher level Zerg don't. Basically this means the opposite of what you said when Maru won against Protoss: Once a certain level of skill is obtained, Zerg doesn't win against Terran anymore. Proof? Top level Zerg lose.
I'm not saying that, but that's basically the same conclusion you drew from the "Maru-Case"... just the other way round.
And yes, at the top level, Terran have quite a high win rate in general:
Terran:
INnoVation: 66.66%
Cure: 74%
Flash: 75%
Reality: 66.67%
Maru: 60%
(All the Terrans in Code S)
Zerg:
soO: 52.6%
DRG:45.45%
Solar: 62%
Soulkey: 54.55%
EffOrt: 72.73%
Life: 48%
TRUE: 40.9%
(All the Zerg in Code S Ro16 + Life because I like Life :-P )
No Code S Terran is below 60% in the matchup since the patch was released. Only 2 Zerg manage to be above 60%. One could of course include foreign Koreans... but that would most definitely not help the Zerg :-P
EffOrt and Solar... no idea where they got all those wins. In GSL Solar won 1 map against Terran, EffOrt not a single one.
|
The game should be balanced on the hightest tier possible, thats what i always say. Highter level of play is more relevant. This was discussed many times here.
Thats why tournament results is an important data. Aligulac includes a much wider range of players, not just the one on the top, this is clear.
If TvZ is imbalanced in the top and not imbalanced in lower levels... then i just call it imbalanced.
The only problem with the data shown for recent tournaments is small sample.
Increasing the sample to have numbers but no content analysis is not really helpful.
Also, before saying that banshee and tank buffs are irrelevant do check if terrans used them, because they did in some games. Directly or indirectly it affected the MU. Not that i think the problem is in those units.
|
On October 13 2014 23:50 Ghanburighan wrote:Show nested quote +On October 13 2014 23:41 Big J wrote:On October 13 2014 23:28 Ghanburighan wrote:
Let's do the your-numbers-are-in-double-digits dance. BigJ, you know very well why such a small number of games shows next to nothing. Until we see a proper number of games (especially in Korea - Flash v Snute in my mind is a bonkers disparity in skill), we just can't make such claims. But that's the point. You will NEVER get a huge number of games in a few weeks when you only consider the games we are actually interested in. And adding matches that we are not interested in is NOT a substitute for a huge number of highlevel games. It is a dilemma which we do not solve by closing our eyes to the flaws that aligulac has, just as it isn't solved by just looking at WCS winrates. Our best bet is to watch as many indicators as possible simultanously and in particular, also watching the games. There are easy solutions, a) we wait for more games, b) we try to explain why the numbers are as they are. We are not constrained to looking at a few weeks. We need to give players time to experiment (or in this case, for zergs to learn the new timings and to learn to micro against WM again). Once the dust settles, we'll know what's up. We have only recently seen the new corruptor-style come out, we'll see in some time where the game is going with it. Also, there is a story for why the 17 sets were T favoured. Some of the MUs were heavily T favoured (Heart v Hyun, Flash v Snute, etc), and in others major mistakes were made on the Z side which were not reciprocated on the T side. I'd like to add that Dark played a great game against TY today that he should have won, had he not misclicked 50 supply of banes on an OC that he didn't even kill. These sorts of unforced errors happen, and they show nothing about balance. They can skew balance, though, if you don't have enough numbers to make the claim. Regarding Aligulac, I'm not convinced that the numbers show nothing. For example, Tefel beating Yoda is a significant result in my mind as I consider Yoda miles ahead in terms of skill. But the real point is, if you want to make the claim that Aligulac stats do not support the claim that Z is weak to T, then you have to explain the mechanism by which Z gets better results in those tournaments. Aligulac takes all tournaments, so in a balanced MU, the winrate should be 50%.
I don't think a) is a good solution. Say you wait some time and the metagame changes. Of what value are the older stats then? Like, say XvY is 55-45. 3months later it is played differently. The stats are 55-45 again. How to judge? And how do you judge "how far" the metagame developed?
The "micro against WM" I don't even want to comment. It's so condescending... As if every Zerg bacome stupid over night and now months after the widow mine rebuff is still too stupid...
Yup, Heart vs Hyun is heavily favored. In favor of the former GSL finalist, WCS AM Champion, long time aligulac leader, WCS points second place, Blizzcon participant Hyun. But yeah, Heart has also been to a GSL Code S once...
The winrate will never be exactly 50-50. There is quite some variance in winning. There are metagame developments. There are differences in which tournaments are being played and who participates. In the very, very longrun we could maybe judge the balance with statistical criteria like that when we can form a year long trend with a very stale metagame. But tiny +/- a few percent movements can mean lots of things.
|
If the only solution is to balance the metagame, then this game will never be balanced.
How about blizzard balances the game instead of the metagame.
|
On October 14 2014 00:14 Big J wrote:Show nested quote +On October 13 2014 23:50 Ghanburighan wrote:On October 13 2014 23:41 Big J wrote:On October 13 2014 23:28 Ghanburighan wrote:
Let's do the your-numbers-are-in-double-digits dance. BigJ, you know very well why such a small number of games shows next to nothing. Until we see a proper number of games (especially in Korea - Flash v Snute in my mind is a bonkers disparity in skill), we just can't make such claims. But that's the point. You will NEVER get a huge number of games in a few weeks when you only consider the games we are actually interested in. And adding matches that we are not interested in is NOT a substitute for a huge number of highlevel games. It is a dilemma which we do not solve by closing our eyes to the flaws that aligulac has, just as it isn't solved by just looking at WCS winrates. Our best bet is to watch as many indicators as possible simultanously and in particular, also watching the games. There are easy solutions, a) we wait for more games, b) we try to explain why the numbers are as they are. We are not constrained to looking at a few weeks. We need to give players time to experiment (or in this case, for zergs to learn the new timings and to learn to micro against WM again). Once the dust settles, we'll know what's up. We have only recently seen the new corruptor-style come out, we'll see in some time where the game is going with it. Also, there is a story for why the 17 sets were T favoured. Some of the MUs were heavily T favoured (Heart v Hyun, Flash v Snute, etc), and in others major mistakes were made on the Z side which were not reciprocated on the T side. I'd like to add that Dark played a great game against TY today that he should have won, had he not misclicked 50 supply of banes on an OC that he didn't even kill. These sorts of unforced errors happen, and they show nothing about balance. They can skew balance, though, if you don't have enough numbers to make the claim. Regarding Aligulac, I'm not convinced that the numbers show nothing. For example, Tefel beating Yoda is a significant result in my mind as I consider Yoda miles ahead in terms of skill. But the real point is, if you want to make the claim that Aligulac stats do not support the claim that Z is weak to T, then you have to explain the mechanism by which Z gets better results in those tournaments. Aligulac takes all tournaments, so in a balanced MU, the winrate should be 50%. I don't think a) is a good solution. Say you wait some time and the metagame changes. Of what value are the older stats then? Like, say XvY is 55-45. 3months later it is played differently. The stats are 55-45 again. How to judge? And how do you judge "how far" the metagame developed?
This is simple. You take it period by period and estimate trends based on them. You can then see when changes occured. Which trends were sustainable (and thus significant) and which winrate fluctuations are merely down to variance.
The "micro against WM" I don't even want to comment. It's so condescending... As if every Zerg bacome stupid over night and now months after the widow mine rebuff is still too stupid...
Knowing is only half the battle. Micro requires practice, and practice takes time. Even before the WM nerf, few zergs had gotten truly good at WM micro. I still see great players like Dark mess up their WM micro and lose tons of units carelessly because they didn't presplit. That's going to improve still because, lord knows, there's room for it.
Yup, Heart vs Hyun is heavily favored. In favor of the former GSL finalist, WCS AM Champion, long time aligulac leader, WCS points second place, Blizzcon participant Hyun. But yeah, Heart has also been to a GSL Code S once...
A glib comment receives a glib response: then Mvp is clearly favoured over Hyun.
The winrate will never be exactly 50-50. There is quite some variance in winning. There are metagame developments. There are differences in which tournaments are being played and who participates. In the very, very longrun we could maybe judge the balance with statistical criteria like that when we can form a year long trend with a very stale metagame. But tiny +/- a few percent movements can mean lots of things.
I concur, but currently we don't know what the winrate is. The best we have is an aligulac list trend, and for the last months, the trend has been from T > Z to Z < T. Every single list, the post-balance terran advantage has eroded away, and now it's going to swing to a slight Z advantage. That's significant. It shows that a) zergs are adapting to the changes, b) terrans are repopulating tournaments, c) the meta is shifting. We learn from those numbers.
|
On October 14 2014 00:30 Foxxan wrote:
If the only solution is to balance the metagame, then this game will never be balanced.
How about blizzard balances the game instead of the metagame.
Lack of computers that are capable of calculating the true balance all possibilities included.
|
Its not rocket science, the meta is changing. We see corruptors, swarmhosts, lotsa cheese, early speed, tons of stuff that makes no sense, and more roach hydra.
That also indicates that zerg is lost and not relying in what we would call standard play. Wait and see if the new adapted standard play (whatever its going to be) is enough to even things up. If not, then its a strong indication of imbalance (aka biomech in TvZ and SCV pulls in TvP some months ago, pre patches).
Im under the assumption that if there is a viable playstyle pros will figure out when under pressure to do so.
Roach Hydra is reliable for now, but if this becomes standard terran can just not prepare for mutalingbane, and shut it down hard.
edit: lets stop taking individual results into account please? you can analyse the games, but just saying" X is better, so he should win" won't help at all O_O
|
On October 14 2014 00:14 Big J wrote:
Yup, Heart vs Hyun is heavily favored. In favor of the former GSL finalist, WCS AM Champion, long time aligulac leader, WCS points second place, Blizzcon participant Hyun. But yeah, Heart has also been to a GSL Code S once...
I think this argument lacks an understanding of how SC2 works. Regardless of past results, the meta keeps wheeling about. If you don't keep up with it, you'll most likely be beaten by someone who did keep up. Heart was in Korea training at the top level, while not going to tournaments, he was keeping up with the meta. Hyun was travelling and depending on his previously accumulated skill in order to perform, which works against some players, but not all. When Heart played this weekend, he played exceptionally, and in the circumstances, that's what matters. Never once did I think that it was due to balance. He showed gameplay which was much more impressive than even the top terrans, recently.
The worst thing that haunts this discussion is to use ALL match outcomes as evidence for whatever opinion you have. Even though there is a marked difference in skill between the two players. soO showed at Dreamhack that balance was not even a factor when there was a discrepancy between player skills. So why should we use many of these Terran wins as arguments for balance being a factor. Was Yoda-Zest a display of Protoss Imba? No, because we acknowledge the inherent difference in skill between players. We should be consistent and give many of the recent Terran wins the same kind of consideration when it comes to judging them.
One thing I will concede, is that Terran's seem to have an easier time incorporating changes into their core gameplay, which explains how a lot of Terran's perform so well while travelling, Taeja and Polt comes to mind.
|
On October 14 2014 00:33 Ghanburighan wrote:Show nested quote +On October 14 2014 00:14 Big J wrote:On October 13 2014 23:50 Ghanburighan wrote:On October 13 2014 23:41 Big J wrote:On October 13 2014 23:28 Ghanburighan wrote:
Let's do the your-numbers-are-in-double-digits dance. BigJ, you know very well why such a small number of games shows next to nothing. Until we see a proper number of games (especially in Korea - Flash v Snute in my mind is a bonkers disparity in skill), we just can't make such claims. But that's the point. You will NEVER get a huge number of games in a few weeks when you only consider the games we are actually interested in. And adding matches that we are not interested in is NOT a substitute for a huge number of highlevel games. It is a dilemma which we do not solve by closing our eyes to the flaws that aligulac has, just as it isn't solved by just looking at WCS winrates. Our best bet is to watch as many indicators as possible simultanously and in particular, also watching the games. There are easy solutions, a) we wait for more games, b) we try to explain why the numbers are as they are. We are not constrained to looking at a few weeks. We need to give players time to experiment (or in this case, for zergs to learn the new timings and to learn to micro against WM again). Once the dust settles, we'll know what's up. We have only recently seen the new corruptor-style come out, we'll see in some time where the game is going with it. Also, there is a story for why the 17 sets were T favoured. Some of the MUs were heavily T favoured (Heart v Hyun, Flash v Snute, etc), and in others major mistakes were made on the Z side which were not reciprocated on the T side. I'd like to add that Dark played a great game against TY today that he should have won, had he not misclicked 50 supply of banes on an OC that he didn't even kill. These sorts of unforced errors happen, and they show nothing about balance. They can skew balance, though, if you don't have enough numbers to make the claim. Regarding Aligulac, I'm not convinced that the numbers show nothing. For example, Tefel beating Yoda is a significant result in my mind as I consider Yoda miles ahead in terms of skill. But the real point is, if you want to make the claim that Aligulac stats do not support the claim that Z is weak to T, then you have to explain the mechanism by which Z gets better results in those tournaments. Aligulac takes all tournaments, so in a balanced MU, the winrate should be 50%. I don't think a) is a good solution. Say you wait some time and the metagame changes. Of what value are the older stats then? Like, say XvY is 55-45. 3months later it is played differently. The stats are 55-45 again. How to judge? And how do you judge "how far" the metagame developed? This is simple. You take it period by period and estimate trends based on them. You can then see when changes occured. Which trends were sustainable (and thus significant) and which winrate fluctuations are merely down to variance. Show nested quote +
The "micro against WM" I don't even want to comment. It's so condescending... As if every Zerg bacome stupid over night and now months after the widow mine rebuff is still too stupid...
Knowing is only half the battle. Micro requires practice, and practice takes time. Even before the WM nerf, few zergs had gotten truly good at WM micro. I still see great players like Dark mess up their WM micro and lose tons of units carelessly because they didn't presplit. That's going to improve still because, lord knows, there's room for it. Show nested quote +
Yup, Heart vs Hyun is heavily favored. In favor of the former GSL finalist, WCS AM Champion, long time aligulac leader, WCS points second place, Blizzcon participant Hyun. But yeah, Heart has also been to a GSL Code S once...
A glib comment receives a glib response: then Mvp is clearly favoured over Hyun. Show nested quote +
The winrate will never be exactly 50-50. There is quite some variance in winning. There are metagame developments. There are differences in which tournaments are being played and who participates. In the very, very longrun we could maybe judge the balance with statistical criteria like that when we can form a year long trend with a very stale metagame. But tiny +/- a few percent movements can mean lots of things.
I concur, but currently we don't know what the winrate is. The best we have is an aligulac list trend, and for the last months, the trend has been from T > Z to Z < T. Every single list, the post-balance terran advantage has eroded away, and now it's going to swing to a slight Z advantage. That's significant. It shows that a) zergs are adapting to the changes, b) terrans are repopulating tournaments, c) the meta is shifting. We learn from those numbers.
I think the most important part just is that we keep the greater picture in mind. One of the main indicators for Terran being weak wasn't really winrates in the first half of the year, but Code S participation. I think there really can be many factors pointing towards an "OK balance", but when 1-2 are significantly off we shouldn't be ignoring them in favor of others.
On October 14 2014 00:45 Superbanana wrote:
Its not rocket science, the meta is changing. We see corruptors, swarmhosts, lotsa cheese, early speed, tons of stuff that makes no sense, and more roach hydra.
That also indicates that zerg is lost and not relying in what we would call standard play. Wait and see if the new adapted standard play (whatever its going to be) is enough to even things up. If not, then its a strong indication of imbalance (aka biomech in TvZ and SCV pulls in TvP some months ago, pre patches).
Im under the assumption that if there is a viable playstyle pros will figure out when under pressure to do so.
Roach Hydra is reliable for now, but if this becomes standard terran can just not prepare for mutalingbane, and shut it down hard.
edit: lets stop taking individual results into account please? you can analyse the games, but just saying" X is better, so he should win" won't help at all O_O
Well, if you think Zerg is currently lost then don't you agree what should be done is:
step 1: wait until Zerg has settled
step 2: see wether what Zergs settle at is actually good enough (by observing for some time)
Personally I don't believe Zergs are lost at all. They make some mistakes. But it's not like you can perfectionize anti-WM micro, the way people pretend to (could I please for once see an example for that? Where are those perfect players from pre-nerf? And how frequently did they achieve that? And how do metagame changes - e.g. Thor usage which mainly started after the patch - into that?). Just like with splitting marines vs banelings, there is no real perfection to be reached. The game is fast, you need to multitask a lot and can't always babysit each and every step of every unit you have.
Muta/ling/bling - possibly into some Ultralisks, though I believe this is partly a question of personal taste - is going to be the main style. Same as with Terrans, there are going to be some weirdos that try to incorporate something else. But that's no different from a Terran player incorporating the one or other siege tank, going a little Thorheavier (e.g. Ty today) or still for hellbats instead of mines or together with them.
And as always, since the game has never really solved its rushproblems there is always going to be the easy solution out that lets you default to a "good enough for me personally" winrate with cheap options, in the case of ZvT roach and baneling busts.
On October 14 2014 00:46 TokO wrote:Show nested quote +On October 14 2014 00:14 Big J wrote:
Yup, Heart vs Hyun is heavily favored. In favor of the former GSL finalist, WCS AM Champion, long time aligulac leader, WCS points second place, Blizzcon participant Hyun. But yeah, Heart has also been to a GSL Code S once... I think this argument lacks an understanding of how SC2 works. Regardless of past results, the meta keeps wheeling about. If you don't keep up with it, you'll most likely be beaten by someone who did keep up. Heart was in Korea training at the top level, while not going to tournaments, he was keeping up with the meta. Hyun was travelling and depending on his previously accumulated skill in order to perform, which works against some players, but not all. When Heart played this weekend, he played exceptionally, and in the circumstances, that's what matters. Never once did I think that it was due to balance. He showed gameplay which was much more impressive than even the top terrans, recently. The worst thing that haunts this discussion is to use ALL match outcomes as evidence for whatever opinion you have. Even though there is a marked difference in skill between the two players. soO showed at Dreamhack that balance was not even a factor when there was a discrepancy between player skills. So why should we use many of these Terran wins as arguments for balance being a factor. Was Yoda-Zest a display of Protoss Imba? No, because we acknowledge the inherent difference in skill between players. We should be consistent and give many of the recent Terran wins the same kind of consideration when it comes to judging them. One thing I will concede, is that Terran's seem to have an easier time incorporating changes into their core gameplay, which explains how a lot of Terran's perform so well while travelling, Taeja and Polt comes to mind.
Sorry, Innovation is not going to get beaten by Heart, regardless of how much he travels. Hyun isn't Innovation, but really, Heart is a complete noname in term of results lately (the ones he had are WCS AM, and HyuN was equal or better in those) and favoring him over the Zerg equivalent to Polt sounds quite like a stretch to me. I actually don't even think that HyuN is such a great player these days, but come on, Heart is not known for being a tournament favorite.
But I didn't really aim towards saying HyuN should have won the game. It's rather that I find it ridiculous that when two players reach a very late round of a tournament and thereby prove to be strong, that you'd just go ahead and in hindsight dismiss results. "because it was obvious". That's just stupid.
|
I'm not saying explicitly that it was obvious that Heart would win. I'm saying I'm not surprised given the circumstances. And I don't think tournament results is always a good indicator of skill. It doesn't mean that it is worthless, but skill can fluctuate without tournament participation as well, which would mean it wouldn't translate to tournament results. A lot of talking about 'favourites' is often just banter, and I think casters tend to overrate past results. Well, it's their job to hype stuff.
Anyway, just because tournament results are not there, doesn't mean that a player has to stagnate in skill. I'm just saying that it makes sense that a player training in Korea will perform better than a player that didn't have the same amount of training/quality. It's like when the Kespa Protosses started to come over and dominated. They too didn't have huge tournament results, but does it mean that they were unfavoured and a proof of imbalance? Was Flash's IEM Toronto win a result of imbalance? Because Flash was never a tournament favourite?
|
On October 14 2014 01:37 Big J wrote:Show nested quote +On October 14 2014 00:33 Ghanburighan wrote:On October 14 2014 00:14 Big J wrote:On October 13 2014 23:50 Ghanburighan wrote:On October 13 2014 23:41 Big J wrote:On October 13 2014 23:28 Ghanburighan wrote:
Let's do the your-numbers-are-in-double-digits dance. BigJ, you know very well why such a small number of games shows next to nothing. Until we see a proper number of games (especially in Korea - Flash v Snute in my mind is a bonkers disparity in skill), we just can't make such claims. But that's the point. You will NEVER get a huge number of games in a few weeks when you only consider the games we are actually interested in. And adding matches that we are not interested in is NOT a substitute for a huge number of highlevel games. It is a dilemma which we do not solve by closing our eyes to the flaws that aligulac has, just as it isn't solved by just looking at WCS winrates. Our best bet is to watch as many indicators as possible simultanously and in particular, also watching the games. There are easy solutions, a) we wait for more games, b) we try to explain why the numbers are as they are. We are not constrained to looking at a few weeks. We need to give players time to experiment (or in this case, for zergs to learn the new timings and to learn to micro against WM again). Once the dust settles, we'll know what's up. We have only recently seen the new corruptor-style come out, we'll see in some time where the game is going with it. Also, there is a story for why the 17 sets were T favoured. Some of the MUs were heavily T favoured (Heart v Hyun, Flash v Snute, etc), and in others major mistakes were made on the Z side which were not reciprocated on the T side. I'd like to add that Dark played a great game against TY today that he should have won, had he not misclicked 50 supply of banes on an OC that he didn't even kill. These sorts of unforced errors happen, and they show nothing about balance. They can skew balance, though, if you don't have enough numbers to make the claim. Regarding Aligulac, I'm not convinced that the numbers show nothing. For example, Tefel beating Yoda is a significant result in my mind as I consider Yoda miles ahead in terms of skill. But the real point is, if you want to make the claim that Aligulac stats do not support the claim that Z is weak to T, then you have to explain the mechanism by which Z gets better results in those tournaments. Aligulac takes all tournaments, so in a balanced MU, the winrate should be 50%. I don't think a) is a good solution. Say you wait some time and the metagame changes. Of what value are the older stats then? Like, say XvY is 55-45. 3months later it is played differently. The stats are 55-45 again. How to judge? And how do you judge "how far" the metagame developed? This is simple. You take it period by period and estimate trends based on them. You can then see when changes occured. Which trends were sustainable (and thus significant) and which winrate fluctuations are merely down to variance.
The "micro against WM" I don't even want to comment. It's so condescending... As if every Zerg bacome stupid over night and now months after the widow mine rebuff is still too stupid...
Knowing is only half the battle. Micro requires practice, and practice takes time. Even before the WM nerf, few zergs had gotten truly good at WM micro. I still see great players like Dark mess up their WM micro and lose tons of units carelessly because they didn't presplit. That's going to improve still because, lord knows, there's room for it.
Yup, Heart vs Hyun is heavily favored. In favor of the former GSL finalist, WCS AM Champion, long time aligulac leader, WCS points second place, Blizzcon participant Hyun. But yeah, Heart has also been to a GSL Code S once...
A glib comment receives a glib response: then Mvp is clearly favoured over Hyun.
The winrate will never be exactly 50-50. There is quite some variance in winning. There are metagame developments. There are differences in which tournaments are being played and who participates. In the very, very longrun we could maybe judge the balance with statistical criteria like that when we can form a year long trend with a very stale metagame. But tiny +/- a few percent movements can mean lots of things.
I concur, but currently we don't know what the winrate is. The best we have is an aligulac list trend, and for the last months, the trend has been from T > Z to Z < T. Every single list, the post-balance terran advantage has eroded away, and now it's going to swing to a slight Z advantage. That's significant. It shows that a) zergs are adapting to the changes, b) terrans are repopulating tournaments, c) the meta is shifting. We learn from those numbers. I think the most important part just is that we keep the greater picture in mind. One of the main indicators for Terran being weak wasn't really winrates in the first half of the year, but Code S participation. I think there really can be many factors pointing towards an "OK balance", but when 1-2 are significantly off we shouldn't be ignoring them in favor of others.
I'll sign this dotted line.
|
On October 14 2014 01:58 Ghanburighan wrote:Show nested quote +On October 14 2014 01:37 Big J wrote:On October 14 2014 00:33 Ghanburighan wrote:On October 14 2014 00:14 Big J wrote:On October 13 2014 23:50 Ghanburighan wrote:On October 13 2014 23:41 Big J wrote:On October 13 2014 23:28 Ghanburighan wrote:
Let's do the your-numbers-are-in-double-digits dance. BigJ, you know very well why such a small number of games shows next to nothing. Until we see a proper number of games (especially in Korea - Flash v Snute in my mind is a bonkers disparity in skill), we just can't make such claims. But that's the point. You will NEVER get a huge number of games in a few weeks when you only consider the games we are actually interested in. And adding matches that we are not interested in is NOT a substitute for a huge number of highlevel games. It is a dilemma which we do not solve by closing our eyes to the flaws that aligulac has, just as it isn't solved by just looking at WCS winrates. Our best bet is to watch as many indicators as possible simultanously and in particular, also watching the games. There are easy solutions, a) we wait for more games, b) we try to explain why the numbers are as they are. We are not constrained to looking at a few weeks. We need to give players time to experiment (or in this case, for zergs to learn the new timings and to learn to micro against WM again). Once the dust settles, we'll know what's up. We have only recently seen the new corruptor-style come out, we'll see in some time where the game is going with it. Also, there is a story for why the 17 sets were T favoured. Some of the MUs were heavily T favoured (Heart v Hyun, Flash v Snute, etc), and in others major mistakes were made on the Z side which were not reciprocated on the T side. I'd like to add that Dark played a great game against TY today that he should have won, had he not misclicked 50 supply of banes on an OC that he didn't even kill. These sorts of unforced errors happen, and they show nothing about balance. They can skew balance, though, if you don't have enough numbers to make the claim. Regarding Aligulac, I'm not convinced that the numbers show nothing. For example, Tefel beating Yoda is a significant result in my mind as I consider Yoda miles ahead in terms of skill. But the real point is, if you want to make the claim that Aligulac stats do not support the claim that Z is weak to T, then you have to explain the mechanism by which Z gets better results in those tournaments. Aligulac takes all tournaments, so in a balanced MU, the winrate should be 50%. I don't think a) is a good solution. Say you wait some time and the metagame changes. Of what value are the older stats then? Like, say XvY is 55-45. 3months later it is played differently. The stats are 55-45 again. How to judge? And how do you judge "how far" the metagame developed? This is simple. You take it period by period and estimate trends based on them. You can then see when changes occured. Which trends were sustainable (and thus significant) and which winrate fluctuations are merely down to variance.
The "micro against WM" I don't even want to comment. It's so condescending... As if every Zerg bacome stupid over night and now months after the widow mine rebuff is still too stupid...
Knowing is only half the battle. Micro requires practice, and practice takes time. Even before the WM nerf, few zergs had gotten truly good at WM micro. I still see great players like Dark mess up their WM micro and lose tons of units carelessly because they didn't presplit. That's going to improve still because, lord knows, there's room for it.
Yup, Heart vs Hyun is heavily favored. In favor of the former GSL finalist, WCS AM Champion, long time aligulac leader, WCS points second place, Blizzcon participant Hyun. But yeah, Heart has also been to a GSL Code S once...
A glib comment receives a glib response: then Mvp is clearly favoured over Hyun.
The winrate will never be exactly 50-50. There is quite some variance in winning. There are metagame developments. There are differences in which tournaments are being played and who participates. In the very, very longrun we could maybe judge the balance with statistical criteria like that when we can form a year long trend with a very stale metagame. But tiny +/- a few percent movements can mean lots of things.
I concur, but currently we don't know what the winrate is. The best we have is an aligulac list trend, and for the last months, the trend has been from T > Z to Z < T. Every single list, the post-balance terran advantage has eroded away, and now it's going to swing to a slight Z advantage. That's significant. It shows that a) zergs are adapting to the changes, b) terrans are repopulating tournaments, c) the meta is shifting. We learn from those numbers. I think the most important part just is that we keep the greater picture in mind. One of the main indicators for Terran being weak wasn't really winrates in the first half of the year, but Code S participation. I think there really can be many factors pointing towards an "OK balance", but when 1-2 are significantly off we shouldn't be ignoring them in favor of others. I'll sign this dotted line.
Probably should have started with that instead of nitpicking why which stats may be flawed. 
|
Czech Republic12128 Posts
Just to note one more important thing - we have a stupid map pool ahead. The map pool, which does not respect HotS rules and its metagame. OK, some of you think - cool, old maps, blah blah blah, nostalgic story. Also think about this - HOW is Blizzard supposed to balance any problems(and I think that the recent Terran success is a problem in the same way I thought the P success is a problem(though instead of balance whine I whined about map pool)), when the map pool will move those problems away for a season until the new map pool is revealed? Statistically you can, as Blizzard, take numbers of the pro games from ladder, because you can select the correct barcodes... but with THIS map pool? I am worried nothing will change until next RO32 or maybe even RO16 which will lead to the same problem - over buffing some race because they will hurry...
|
@BigJ: I do think a lot of zerg players are lost (maybe the majority in the top). Yes, i do agree with steps 1 and 2, thats one of the things i said in my post.
I don't think the muta ling bling that we know is going to be standard, if its the same composition the builds are going to change, but its just a guess after all.
Yes, i do agree that "cheap" play can even winrates, i think its already happening.
|
We need to wait until more terrans figure out their new timings perfectly and especially the new combinations of bio + x, with x=mine, thor, hellbat, tank (the 4 times in a row buffed terran units) or all of them together, following each other and whatever else.
By then we gonna see some tournament wins of zergs such as at dreamhack stockholm where basically every single top 10 zerg in the world participated and almost none of the top 10 terrans, but hardly anywhere else.
After ~2 months from now the state of balance should be clearly figured out as terran favoured probably in all matchups but at least in TvZ as more of lower than very top tier terrans will have figured stuff out and learned how to benefit.
p.s: in about 2 month blizzard is gonna announce an infestor buff cause it will benefit all 3 zerg matchups.
|
Another interesting statistic would be the win rates of the top X aligulac players just against each other. This would mean we are only looking at the highest level of play, yet we still get a good number of data points (since we are not only looking at the handful of WCS or GSL games).
Looking at games played between the best players only is in line with balancing the game at the highest possible level.
I wrote a quick python script to do this via aligulacs API.
For the period from Sep-01 until today:
Top 50 active players: ['Solar', 'herO', 'YoDa', 'Sacsri', 'INnoVation', 'TaeJa', 'Zest', 'jjakji', 'Rain', 'Life', 'Flash', 'MMA', 'HyuN', 'Polt', 'Bomber', 'soO', 'San', 'sOs', 'Cure', 'Scarlett', 'First', 'Bunny', 'ForGG', 'Dear', 'Maru', 'KingKong', 'Soulkey', 'TY', 'Leenock', 'PartinG', 'Stats', 'Bbyong', 'Classic', 'StarDust', 'Heart', 'Jaedong', 'Patience', 'VortiX', 'Sora', 'Snute', 'viOLet', 'Hurricane', 'TargA', 'GuMiho', 'Happy', 'DongRaeGu', 'Ourk', 'Super', 'Pet', 'Mvp']
171 meetings between these players
Total games: 485
PvT - 47.90% - 119
PvZ - 50.94% - 106
TvZ - 61.96% - 92
35 PvP, 75 TvT, 57 ZvZ
=> PvT and PvZ look balanced within the expected variations, TvZ looks quite a bit off.
If we include the top 100:
Top 100 active players: ['Solar', 'herO', 'YoDa', 'Sacsri', 'INnoVation', 'TaeJa', 'Zest', 'jjakji', 'Rain', 'Life', 'Flash', 'MMA', 'HyuN', 'Polt', 'Bomber', 'soO', 'San', 'sOs', 'Cure', 'Scarlett', 'First', 'Bunny', 'ForGG', 'Dear', 'Maru', 'KingKong', 'Soulkey', 'TY', 'Leenock', 'PartinG', 'Stats', 'Bbyong', 'Classic', 'StarDust', 'Heart', 'Jaedong', 'Patience', 'VortiX', 'Sora', 'Snute', 'viOLet', 'Hurricane', 'TargA', 'GuMiho', 'Happy', 'DongRaeGu', 'Ourk', 'Super', 'Pet', 'Mvp', 'ByuL', 'MajOr', 'Trap', 'Jim', 'Rogue', 'Lilbow', 'Dark', 'Hydra', 'MC', 'RagnaroK', 'TRUE', 'Impact', 'Center', 'Nerchio', 'Reality', 'KangHo', 'Golden', 'Petraeus', 'ShoWTimE', 'HeRoMaRinE', 'Pigbaby', 'Journey', 'Sen', 'EffOrt', 'MaNa', 'SuperNova', 'MaSa', 'Symbol', 'Avenge', 'Dayshi', 'Kas', 'Trust', 'Ryung', 'Hack', 'Check', 'HerO', 'EnDerr', 'Curious', 'Stork', 'FanTaSy', 'Starbuck', 'Thermal', 'Revival', 'Squirtle', 'ParalyzE', 'TLO', 'Spear', 'MorroW', 'HuK', 'Sorry']
466 meetings between these players
Total games: 1279
PvT - 47.37% - 247
PvZ - 52.49% - 261
TvZ - 54.44% - 349
85 PvP, 150 TvT, 186 ZvZ
=> Z looks weak, T looks strong. TvZ is not as off as before.
The additional data supports the idea that TvZ has been T-favored for the last weeks, in particular at the very highest levels of play.
(All data via aligulac's API.)
|
This is a nice new data point. Although it's clear why the top 50 gives a weird result when you compare it to top100. It has people like Targa, Jaedong, Leenock, etc over people like Hydra, Dark, Curious.
I do dislike that both lists bring in Mvp, Kas and some guy called EnDerr. Perhaps it's possible to refine the search a bit more.
|
On October 14 2014 02:49 Ghanburighan wrote:
This is a nice new data point. Although it's clear why the top 50 gives a weird result when you compare it to top100. It has people like Targa, Jaedong, Leenock, etc over people like Hydra, Dark, Curious.
I do dislike that both lists bring in Mvp, Kas and some guy called EnDerr. Perhaps it's possible to refine the search a bit more.
Easy to refine in principle. For starters I just took the aligulac top X. Could be something like current GSL players or some mixed group instead.
Agreed, players like Kas and EnDerr are why I personally prefer the top 50 over the top 100. It cannot be reduced much below 50 if we want to have a decent number of games (and thus valid statistics).
|
On October 14 2014 03:11 antiRW wrote:Show nested quote +On October 14 2014 02:49 Ghanburighan wrote:
This is a nice new data point. Although it's clear why the top 50 gives a weird result when you compare it to top100. It has people like Targa, Jaedong, Leenock, etc over people like Hydra, Dark, Curious.
I do dislike that both lists bring in Mvp, Kas and some guy called EnDerr. Perhaps it's possible to refine the search a bit more. Easy to refine in principle. For starters I just took the aligulac top X. Could be something like current GSL players or some mixed group instead. Agreed, players like Kas and EnDerr are why I personally prefer the top 50 over the top 100. It cannot be reduced much below 50 if we want to have a decent number of games (and thus valid statistics).
Only GSL players sounds quite viable on paper. Does that yield anything different from Top50 or GSL games only?
|
|
|
|
|
|
EPT
