|
On February 22 2013 15:56 Popkiller wrote:Show nested quote +On February 22 2013 15:53 MCXD wrote: But if the same player won every single competition and had a 90% winrate, wouldn't everyone complain about the fact the outcome of every tournament is boring and obvious? I'm pretty sure they would. if it was just one good player who won everything and everyone else sucked, it'd be boring. But if there were 3-4 players (or even two) who were at the top of the game and constantly battling for trophies, everyone would love it, like what we've had in tennis over the past years.
I agree with this, but he never said everyone other than the 1 dominating sucks.
It is more fun when there's 2-4 guys who usually beat everyone else (let's say Mvp, Nestea, MC) (Federer, Nadal, Djokovic), but play amazing games vs each other.
|
It's simply that the game is to volatile. There is no player like "Flash" in BW because the game doesn't allow it (yet at least, this could change)
|
Woa never checked the Liquibet rankings ^^ I'm 33rd with a percentage of 58.7% :o Cool post OP
|
Ugh, I know this is frowned upon but I havent read the entire thread, just the first page of replies.
Nevertheless for anyone still struggling with this, I have a one word answer that should set your minds at ease.
Golf.
Just think of starcraft in terms of golf and you will have no trouble accepting a 60% winrate and a 60% prediction rate.
|
Pro mahjong is more volatile than SC2, I can tell you that at least.
|
When you talk about a format like the GSL too it isn't as cut and dry as it seems. Factors like available replays of your opponent, one player doing a style you didn't prepare for, and what practice partners you have toss things up in the air. For example, Curious beating Parting 3-0 was so huge because StarTale doesn't have any Protoss players left at Parting's level for Curious to practice against, and it didn't involve Curious hard-countering the wonwonwon in any games. On the other hand, there are tons of Parting replays available, and Curious' style was "standard," so his early aggression and timing attacks probably threw Parting off. Another example; look at the success of Innovation and Mvp. Their performances outside the GSL lately have been lackluster at best, but Mvp's ability to prepare and throw any of his opponents knowledge of him off by tossing out all ins got him another gold pin. Likewise, Innovation went up against Stephano twice, and Stephano, who in a spontaneous BoX could compete with anyone in the world, used the same build he does in pretty much every game. Stephano won a few games when Innovation let him get to the late game, but otherwise Innovation just took advantage of how greedy Stephano opens and killed him with a timing attack. So you can't just rely on pure skill in game for a GSL win, knowledge and preparation beforehand matter just as much
|
On February 22 2013 15:48 Kaitokid wrote: Artosis is not the most knowledgable guy in all of starcraft lol. the best players are.
the main reason why predictions are often wrong is the people who predict stuff dunno what the hell they are talking about because they know the players based on a couple games they have seen of them. even if they watched hundreds of games of a certain player it's still just like 0.5% of all the games the players have played.
the best people to predict stuff would be other good players who know both 2 players very well and played them both a shit ton of times.
i completely agree with this post !
i just noticed that i very rarely write something myself, i tend to just read the comments till i find one that says what i want to say :D thanks Kaitokid
|
United Kingdom35820 Posts
On February 22 2013 19:37 Grumbels wrote: Tennis is random: 1. top players take points off each other all the time, because you can take risks. 2. some points are a lot more important than others. You can win games without winning the majority of the points.
Reading through this thread, came across this. It's an extremely good point. In tennis matches between closely matched players, total points scored are often extremely, extremely close, like 51-49%, 52-48%.
Even in a straightforward 6-4, 6-3 affair (say), the final points score will be around 55-45%. Even in the most outrageously one-sided match, it's extremely rare for the points difference to be greater than a 2:1 ratio.
Also, even in a very one-sided tennis match, it's likely to take 1h15 or 1h30, something like that. Say a SC2 player beat someone 5-1, with games of 10-15 real-time minutes each (say 15-20m average SC2 time). That's a best of 9. If every match was BO9, BO11, BO13... things would be different. If tennis matches were best of 1 set, which is the functional equivalent of a BO3 or even BO5 in SC2, then tennis would look remarkably different, too.
|
I don't feel like results are that random, however it does changes a lot. StarCraft is a game with a lot of variables (race, map, composition, openers, players, state of mind ...) and as expected it is very hard to be prepared for all of these aspects, especially when knowing that if one of these aspects crumbles, the rest usually follow. I think every player has had that feeling in a game where you play vs a build that you don't like and immediately tilt a bit. So it's a bit obvious and even expected that the results shift wildly. Even in BW, predicting the winner wasn't as easy. Yes you had the very good players that have always been very consistent but the other match were still there, even if you see way more upsets in SC2. I feel like in SC2 we haven't seen a champion that was ... hum flawless maybe ? And by flawless I don't mean their play but factors exterior to the game. Like every dominant player has had some sort of flaw drawing back their play. Nestea was older than the rest of the competition, MKP has always been too emotional, goes on tilt too easily. Life still had to focus on school, MVP was very good but he was injured, same with Taeja. Only exception would be MC. Maybe we've just been unlucky.
Now for the betting part, is it random or not. It's probably more random that we seem to think. As spectators we are exterior to the game and have limited information prior to match. We don't know what player are practicing (even IF they are practicing), we don't know their state of mind and so on. We lack a lot of information that would help us make a "good" guess. The only thing we can rely on is the previous winner interview (which is PR, so not that reliable but it can still help, if a player publicly admits he's not confident for his next match maybe he's not a safe bet) and previous results. We only see the past.
Now onto my opinions of liquibet : Liquibet in themselves are volatile. There is no benefit to not vote on match. If I have no idea, I should go at random because even if I don't vote it doesn't matter. The way they distribute points is also another factor. Like for example, the quarterfinals were worth 3 points each. I correctly got both Curious vs Parting and Symbol vs Bogus, putting me 1st by a "large" margin even tho it was very close for the past two months. We always were at like one or two points of each other max. These two match were east for me while MC vs Roro and Taeja vs Soulkey was a way bigger toss up. We don't really know what Taeja's medical condition is, so he's taken pleasure at fucking up my bets all season long so he's a wildcard here. Then you have MC vs Roro, MC just popped out of nowhere going all like "sup im back" and you don't really know for how long he's gonna keep it up. Same with Roro, he's good and somehow he's in the quarterfinals. I'm not saying they are not deserving, just that I wouldn't have thought they would make it this far looking a Ro32 groups. So this match is another wildcard. If we look at yesterday Curious vs Parting was a close one (despite the result) and Bogus vs Symbol was very clear in my mind.
So yeah, some times you have easy match and you have to get them right to stay in the race, then you have the ones that can go both ways and you have to "get lucky" to stay first. I got lucky yesterday, and got put in a comfortable lead as a result. I got unlucky today and will end up second if anyone that is currently second gets both bet correctly ( so I'm probably second as soon as the results are updated, kinda sad, I held this position for a while, I like the smiley ).
Betting is a lot like gambling. You have good gamblers, but you kinda have to be lucky to become one.
|
I don't really get it. Have you looked at MVP's liquipedia page? Do you realize that 12 of the 17 GSLs have had either MVP, MC, or Nestea in the code S finals? Did you look at any of their liquipedia pages? Did you watch Hyun win 14 consecutive best of 9s in IPL Fight Club? Did you see Parting and Life in the past 6 months win multiple major tournaments each? Do you not recognize every name in code S and almost every name in code A?
The GSL, which you are referencing, as with many other SCII tournaments, has you play 2-3 best of 3s for your tournament life. Some favorites getting knocked out early should not be surprising. That does not mean the player is on the decline, and it does not mean that the game has too low of a skill ceiling. It is not the MLB, NBA, NHL ect.. There is no regular season with 80ish (MLB 160ish) games. Remember when the Red Sox didn't win a world series for like 100 years, despite being an excellent team and bigtime championship contenders for many of them (even in the finals for a few)? Did "The Curse of the Billy Goat" with the Cubs mean baseball was broken? Tennis has been mentioned in this thread; a professional mens tennis match is a best of 5 in sets, each of which is a best of 13 games. Do you think that playing at most 50+ games per match as opposed to at most 5 will reduce the variance?
Do you know what sport has variance? How about the extremely popular college football, where like 10 team all finish the season with 1 loss and they literally have a computer just pick the teams. Apparently that doesn't stop college football from being awesome, so I don't know why Parting losing a best of 5 vs Curious (another consistently top 32 in the world player in case anyone forgot) would be so much worse.
Its amazing to me that there isn't way MORE variance. Anyway, I would wait on the whole 'SCII has too much variance" thing until people who haven't proven themselves to be top-caliber even qualify for the GSL (foreign code A qualification: probably 0 for 50ish), let alone consistently beat any of the pros.
|
On February 22 2013 16:43 SuperYo1000 wrote: sc2 has gotten significantly more stale in the last year. Lets not ruin it and have almost predetermined tourny outcomes as our goal as a community
That's your opinion. Except for the Las Vegas portion of the GSL, I think the games have gotten better and better.
|
My Personal Opinion is that SC2 isn't nearly as luck based as seems. Maybe it's a matter of perspective. Say, when you take people with a noticeable difference of skill level and pit them against each other, (such as a korean vs non-korean, GM league vs master league, Master vs Diamond, Diamond vs plat. and so on) then They'd have a much closer to 100% win rate.
Because Win rates go up so drastically when there is a difference of skill level, then perhaps luck isn't as big a factor after all. After all, It isn't that rare that best of 7's end up in a 4 - 0. Which should be unlikely that win rates were always at most 60%. There's people that have one the GSL more than twice, Which would also be very unlikely if win rates were capped low.
I believe the reason predictions are so hard, is that Liquibet based on its function pits people against people with similar skill level. That's just how the GSL system works, There's so many levels of the tournament that it seems like Skill level wise, the players tend to be where they belong.
So No, There isn't too much "luck" in sc2, The people at the top level in korea are just very similar in skill. Which isn't surprising, with how much time they spend on practicing, and the fact they all have access to a plethora of information involving build orders ect.
Edit: I'd simply love, however as far as matches go. That when possible tournaments should have a format, in which it is necessary for players to be ahead by at least 2 games to win a best-of. So when a Best of 3 was 1-1, it would turn into a BO 5, and if a BO 5 was 2-2, it would turn into a BO 7, and so on. Scores like 3-2, and 4-3, are so inconclusive as to which is the better player.
|
If the same 2 or 3 people won all the time people would complain about that too. MVP has been consistently dominant enough despite having no wrists for me to not worry about this too much.
|
Wasn't Parting totally outplayed by Curious yesterday? It looked so to me at least. I think it's more that players are not consitent than that the game is chancy. I didn't feel Curious won on a coin-flip. Also, MVP having won so many GSL championships surely tells a tale of that skill actually pays off. If it were a 50/50 chance for a player to win each game, what are the chances that MVP wins that many titles you think?
|
I'm scared to reply to this post because I've been banned before for posting anything approaching the negativity spectrum..
This whole post is bogus. It seems to me that you're looking for events that confirm what you already believe, that the game is too luck based, or the skill ceiling is low, or whatever.
You list Artosis as evidence of this. You choose to believe he must know who should or should not win a RoX game in a very highly competitive Code S, and when he's wrong, it must be due to some game shortcoming. Betraying your own argument though, as you mentioned, he is wrong most of the time. This would really suggest to me that Artosis is just bad at predictions in general, and never would I assume anything about Starcraft's gameplay, or even begin to even think about a "fix" to this imaginary problem.
With respect to Liquibet, your probability assumptions (and hence the whole argument) are complete bogus to it's not even worth discussing.
You go on to mention win rates of Mvp and Life in Starcraft as they compare to Federer in tennis. A literally 5 second Google search on Federer's lifetime win rate returns 69%, so I literally have no idea what you're on about here.
Is there something wrong with the game? If so, is there anything that can be done to fix it?
On February 22 2013 15:38 Gatesleeper wrote: Is there something wrong with the game? If so, is there anything that can be done to fix it?
No, stop thinking up imaginary problems and imaginary solutions to these problems.
To me the real gross assumption here is that you can even properly determine who the better on any 1 on 1 match is. Starcraft is inherently a vastly more complex game than say football, tennis, swimming, sprinting, whatever. Through statistics, we can formulate predictions on who is maybe the best player overall over some period of time, but I really don't think we can accurately predict any given pairing with very strong confidence.
On February 22 2013 15:49 Kennigit wrote:Show nested quote + Is there something inherently wrong with a game where the supposed best players so often lose to players who are widely considered weaker than them? Like, is the skill ceiling not high enough? Is it too luck based? Why don't the best players beat the slightly less good players more often?
This has been suggested by pros, writers, community figures since 2008 when we started playing alpha builds. It has been consistently complained about (especially by idra) for years now.
So the question's been asked before, okay.
|
suppose the best players have a 70% winrate. there is still a 30% chance to lose, even if it is a 0-3 loss
|
Artosis admitted at today's GSL Round of 8 day 2 broadcast that he's cursed after Soulkey, the player he wagered on, lost the fifth game to Taeja.
|
fer what you are saying is pretty dumb and ignorant to be honest. Referring to the 69% of Federer shows that you have no idea about the Tennissport. So please stop taking some no-sense stats from google. Federer isn't nearly in form anymore and Djokjovic is the current top player. Top 4 players in the world ranking list also in the top4 of the australien open shows how great of a sport it is.
Ofc it's true that you can't compare SC2 to Tennis because of the differences in each game. But at the same time it's just a bad specatator sport when you as a fan never know who is going to win in the most prestigious tournament (the GSL). I'm pretty sure it will continue to be like it and maybe get even worse when more players catch up to the level where it seems like it comes down more to luck then to skill.
I wasn't interested in BW so much but I wonder how much did the automining affected the skill ceiling? For me it seems like an impossible task to keep up through out the whole game especially when getting herrassed at the same time. It's such a little thing which lowers the skill cap immensely. Same thing goes for the smaller control groups. These are the things sepereating great players from good players. Right now every one in Master league can perform a certain build order without any problem. As there is nothing hard in remembering the supply counts you are supposed to build something. I'm not saying they can perform it at the same level as pros do but when you see some random GM player beating pro players like Stephano in ZvZ i really feel like there's a huge problem with the skill ceiling. Pro players with much more apm then these random GM players should have no problem defeating these guys.
|
I think the comparisons to Brood War are slightly unfair. You could say that Flash only had 70% winrate or whatever, but that's including proleague, where people specifically snipe him in a Bo1. Furthermore, he only plays vs the strongest players over and over. If you look at the current proleague there is nobody that gets to 70%. There is sOS, who has 68%, but the next best players are Hydra, Zero, Rain and RoRo at 63%. To take a random example, if you look at the 2010-2011 proleague, then over the entire season there are five players that have at least 65% or higher winrates, Flash, Bisu and Jaedong being far over 70%.
I did a recent count for this and the number of different (i.e. non-repeating) players each Ro8 in Starcraft 2 compared to Brood War is like 30% higher, so you can see that in Brood War there is more consistency and there are bigger stars.
I think if SC2 had balance issues and bad match-ups fixed that we would be in a much better position to judge whether it's really too volatile or not. I think the recent zerg balance issues have made it difficult to really say anything meaningful about the recent results.
|
I dont get what you are complaining about. Its not like they are loosing to some random guys( lets say XlorD,Bratok,Titan,Idra,qxc,masa .. any mid tier progamers) - they wouldnt even make it there. in GSL. They are playing vs the very very very best. As all, in this thread, mentioned there are so many factors - and so easily distrubable ones that its very hard to have yours consitency at level to not loose at such high level for extended periods of time.
|
|
|
|