|
On February 23 2013 08:32 monkybone wrote:Show nested quote +On February 23 2013 08:25 Vorenius wrote:On February 23 2013 08:11 monkybone wrote:On February 23 2013 08:03 Die4Ever wrote:On February 23 2013 07:50 monkybone wrote:On February 23 2013 07:46 BrokenMirage wrote: Isn't the assumption that blind picking on Liquibet would give you an average of 50% false, since for the GSL group picks, you have to pick the two players coming out of the group, giving you odds of 1 over (4 choose 2) or 1/6. The expectancy value of points you get when picking at random will still be 50% of the maximum amount. I'm not sure that's correct. Each group seems to be worth 1 point if you get it right. At random I think you have a 1/6th chance of getting the group right. So if you only bet on the GSL groups blind then you would be at about 16.6%? Well, that would be a very odd way of calculating the percentage of correct liquibets. My guess is that guessing two players correctly to advance contributes double that of guessing one player correctly. If you only get one player right, you didn't get it right, and you don't get any points. Oh, I didn't know that. And how many points do you get for getting it right, compared to one vs one matches? just 1 point, I think they should change it to 1 point per correct player instead of 1 point for the whole group, but w/e
|
On February 23 2013 08:32 monkybone wrote:Show nested quote +On February 23 2013 08:25 Vorenius wrote:On February 23 2013 08:11 monkybone wrote:On February 23 2013 08:03 Die4Ever wrote:On February 23 2013 07:50 monkybone wrote:On February 23 2013 07:46 BrokenMirage wrote: Isn't the assumption that blind picking on Liquibet would give you an average of 50% false, since for the GSL group picks, you have to pick the two players coming out of the group, giving you odds of 1 over (4 choose 2) or 1/6. The expectancy value of points you get when picking at random will still be 50% of the maximum amount. I'm not sure that's correct. Each group seems to be worth 1 point if you get it right. At random I think you have a 1/6th chance of getting the group right. So if you only bet on the GSL groups blind then you would be at about 16.6%? Well, that would be a very odd way of calculating the percentage of correct liquibets. My guess is that guessing two players correctly to advance contributes double that of guessing one player correctly. If you only get one player right, you didn't get it right, and you don't get any points. Oh, I didn't know that. And how many points do you get for getting it right, compared to one vs one matches? 1 point. When you look at the liquibets, the number on the left is how many points you get. So you are left with a lot bets that are very hard to predict since there are 6 outcomes, and then all the Code A matches between relatively unknown players that obviously are gonna be hard to predict if you don't know half the players.
I don't think SC2 is significantly more volatile than BW, and the champs' winrates seem to support that. In both games, dominating players rarely go much above 70% over longer periods of time.
|
also using peoples' liquibet scores is kinda iffy this season because of proleague
|
|
1 big thing you gotta factor in is that for SC2 the format is completely different from typical sports like the NBA/NHL/MLB.etc Most of the liquibets and predictions are centered around GSL, which is a league where each player has a lot of time to prepare for their opponent. This makes it harder to predict as one player may prepare more than the other, which could slant things in their favour. This is especially true if one of the players plays in a lot of foreign tournaments as well as they have less time to focus and study/prepare for their GSL opponent. These are all things none of us can really account for when we do predictions as none of us really know how much time they spend preparing for a specific opponent.
|
Canada10904 Posts
On February 23 2013 01:30 Evangelist wrote:
The only reason BW stabilised is because they stopped patching it. In the process they left half the units in the game completely useless. Starcraft 2 is constantly getting rebalanced and the metagame shifts ever so slightly every single time someone discovers a new build.
I don't know what game you play, but this is not remotely true. Even if one is being extremely generous by discounting wraiths and Dark Archons that do have pretty specific use.
|
On February 23 2013 08:11 monkybone wrote:Show nested quote +On February 23 2013 08:03 Die4Ever wrote:On February 23 2013 07:50 monkybone wrote:On February 23 2013 07:46 BrokenMirage wrote: Isn't the assumption that blind picking on Liquibet would give you an average of 50% false, since for the GSL group picks, you have to pick the two players coming out of the group, giving you odds of 1 over (4 choose 2) or 1/6. The expectancy value of points you get when picking at random will still be 50% of the maximum amount. I'm not sure that's correct. Each group seems to be worth 1 point if you get it right. At random I think you have a 1/6th chance of getting the group right. So if you only bet on the GSL groups blind then you would be at about 16.6%? Well, that would be a very odd way of calculating the percentage of correct liquibets. My guess is that guessing two players correctly to advance contributes double that of guessing one player correctly. It seems to me that you only need to correctly predict 1 player getting through to get the point. Guessing both players correctly still yields only 1 point.
So, in a GSL group you have 4 players, and 2 will advance. You select 2 players, and if at least one of your picks advances, you receive the point. I don't know the math, but looking at it intuitively that looks like a 50% chance of getting the point.
In previous seasons of liquibet the voting system for GSL groups was a little different, but this season they simplified it.
|
On February 23 2013 08:44 Falling wrote:Show nested quote +On February 23 2013 01:30 Evangelist wrote:
The only reason BW stabilised is because they stopped patching it. In the process they left half the units in the game completely useless. Starcraft 2 is constantly getting rebalanced and the metagame shifts ever so slightly every single time someone discovers a new build.
I don't know what game you play, but this is not remotely true. Even if one is being extremely generous by discounting wraiths and Dark Archons that do have pretty specific use. You mean scout? I thought wraiths were a staple.
On February 23 2013 08:46 Gatesleeper wrote:Show nested quote +On February 23 2013 08:11 monkybone wrote:On February 23 2013 08:03 Die4Ever wrote:On February 23 2013 07:50 monkybone wrote:On February 23 2013 07:46 BrokenMirage wrote: Isn't the assumption that blind picking on Liquibet would give you an average of 50% false, since for the GSL group picks, you have to pick the two players coming out of the group, giving you odds of 1 over (4 choose 2) or 1/6. The expectancy value of points you get when picking at random will still be 50% of the maximum amount. I'm not sure that's correct. Each group seems to be worth 1 point if you get it right. At random I think you have a 1/6th chance of getting the group right. So if you only bet on the GSL groups blind then you would be at about 16.6%? Well, that would be a very odd way of calculating the percentage of correct liquibets. My guess is that guessing two players correctly to advance contributes double that of guessing one player correctly. It seems to me that you only need to correctly predict 1 player getting through to get the point. Guessing both players correctly still yields only 1 point. So, in a GSL group you have 4 players, and 2 will advance. You select 2 players, and if at least one of your picks advances, you receive the point. I don't know the math, but looking at it intuitively that looks like a 50% chance of getting the point. In previous seasons of liquibet the voting system for GSL groups was a little different, but this season they simplified it. That can't be right, since a random bet would have a 5/6 chance of getting 1 point.
|
|
On February 23 2013 08:47 achan1058 wrote:Show nested quote +On February 23 2013 08:44 Falling wrote:On February 23 2013 01:30 Evangelist wrote:
The only reason BW stabilised is because they stopped patching it. In the process they left half the units in the game completely useless. Starcraft 2 is constantly getting rebalanced and the metagame shifts ever so slightly every single time someone discovers a new build.
I don't know what game you play, but this is not remotely true. Even if one is being extremely generous by discounting wraiths and Dark Archons that do have pretty specific use. You mean scout? I thought wraiths were a staple. Show nested quote +On February 23 2013 08:46 Gatesleeper wrote:On February 23 2013 08:11 monkybone wrote:On February 23 2013 08:03 Die4Ever wrote:On February 23 2013 07:50 monkybone wrote:On February 23 2013 07:46 BrokenMirage wrote: Isn't the assumption that blind picking on Liquibet would give you an average of 50% false, since for the GSL group picks, you have to pick the two players coming out of the group, giving you odds of 1 over (4 choose 2) or 1/6. The expectancy value of points you get when picking at random will still be 50% of the maximum amount. I'm not sure that's correct. Each group seems to be worth 1 point if you get it right. At random I think you have a 1/6th chance of getting the group right. So if you only bet on the GSL groups blind then you would be at about 16.6%? Well, that would be a very odd way of calculating the percentage of correct liquibets. My guess is that guessing two players correctly to advance contributes double that of guessing one player correctly. It seems to me that you only need to correctly predict 1 player getting through to get the point. Guessing both players correctly still yields only 1 point. So, in a GSL group you have 4 players, and 2 will advance. You select 2 players, and if at least one of your picks advances, you receive the point. I don't know the math, but looking at it intuitively that looks like a 50% chance of getting the point. In previous seasons of liquibet the voting system for GSL groups was a little different, but this season they simplified it. That can't be right, since a random bet would have a 5/6 chance of getting 1 point. Yup, you're right. After thinking about it for a minute, a random guess yields 5/6.
4 players, pick 2.
A B C D
you pick A B
possible outcomes:
A B A C A D B C B D C D
if you pick any two players, there is only 1/6 chance neither of those players will win.
But I am pretty sure that's how liquibet works. When I pick 2 players and 1 of them goes through, I receive 1 point. If both my picks are correct, I believe I still only received 1 point.
|
Canada10904 Posts
On February 23 2013 08:47 achan1058 wrote:Show nested quote +On February 23 2013 08:44 Falling wrote:On February 23 2013 01:30 Evangelist wrote:
The only reason BW stabilised is because they stopped patching it. In the process they left half the units in the game completely useless. Starcraft 2 is constantly getting rebalanced and the metagame shifts ever so slightly every single time someone discovers a new build.
I don't know what game you play, but this is not remotely true. Even if one is being extremely generous by discounting wraiths and Dark Archons that do have pretty specific use. You mean scout? I thought wraiths were a staple. Like I said, I'm trying to be really, really generous. The closest you can get is with zerg simply because of how few units they have to begin with so the percentages change easily. Protoss is over 90% in 1v1 anyways.
|
On February 22 2013 15:38 Gatesleeper wrote: I'm just gonna throw out an unpopular idea here that's been sitting at the back of my mind for a long time while watching Starcraft.
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?
In any game that requires this much concentration, psychological factors are also huge. Injuries, bad nights sleep, etc all affect it.
But there's a bigger gripe here: How do you know that a certain player is 'better'? Who decides who 'should win'?
I know Idra goes on about how lesser players can beat better players, but he literally thinks he's one of the top players who shouldn't ever be beaten. His standards for deciding skill are totally biased, are much of the fanbases.
That is the main problem: The only was to determine how good a player is, is to look at who he beats. Then, by comparing "Person A beat Person B who beat Person C", you end up with "Person A beats Person C". That's not how Starcraft works.
TSLCenter vs IM_Seed. OMG WHAT AN UPSET, LESSER PLAYER BEATS THE GUY WHO SHOULD HAVE WON THIS GAME BLOWS.
Who's to say Center isn't actually better?
Just because Seed has a championship, it actually doesnt matter. People focus too much on past history, when all that matters in Starcraft is that game, right then and there. Center went on to beat Seed again, showing it wasn't a fluke, and that he actually is the better players.
Skill level is always developing. So is meta-game. Stop setting ridiculous rigid standards for who is better. People lose because they aren't good enough. Not because of the game.
If you actually watch the games, you'll see that 9/10, the person who wins actually plays better than the loser. Amazing, right?! Nothing to do with game design.
Consider this: Artosis, perhaps the most knowledgable guy in all of Starcraft II, is constantly predicting certain players to do well and constantly getting it wrong. Hence, the "curse".
Lol.
Pretty sure he is nowhere near. Sure he knows a lot about the game itself, but I don't think he's in touch with the pro scene at all. Hence why Wolf and Khaldor can actually predict scores, because they talk to players and know who is up and coming through all the stuff they cast.
Artosis just turns up and tell you who he thinks is a sick nerd baller based off....god knows what.
Pretty sure he trashed Bogus in his first season, now he's all over his dick saying he's the best ever.
|
i think if it's too luck based we wouldn't have had 3 time and 4 time GSL champions, we'd have 1 or 2 at the absolute most, especially when won of those championships was won without dropping a game.
theres other factors at play here namely the lack of ways to outplay your opponent with better control and mechanics. Something HOTS is addressing in some ways with fast units, more reliability on detection ect ect which will give people with better control, multi-tasking and mechanics the edge over those that dont.
|
If player A is a complete beginner to games and player B is a 2 to 1 favourite against player A, then we can say that players A and B are on different "levels of skill". Then of course there's player C who is a 2 to 1 favourite against player B, player D who is a 2 to 1 favourite against player C, and so on until we hit the very top. How many "levels of skill" are there in Starcraft 2?
In chess there's about 20 levels of skill, whereas in a game like Go there's as many as 40. Does this mean that Go is a more skillful game? It's not necessarily so - in a given position finding the best move in chess is about as difficult as finding the best move in Go, and I don't think anyone could argue that reaching the top level is harder to achieve in Go than in chess. But the difference is that a single game of Go is much longer, and can last for well over a hundred moves, whereas a single game of chess lasts maybe 40 moves. In Go, a better player has many more opportunities to distinguish himself.
So the issue with chess here may be that just a single game, a bo1, is too small a unit to be measuring player's levels by, and this is much, much, more so in Starcraft 2. A single game isn't nearly enough for a player to distinguish himself. If all liquibets were best of 5s then you'd have a significantly better success rate than 53%.
|
The game is simply played to fast for players to be consistently better. Take just one example: marines in a mineral line. In SC2 drones pop like bubbles, while in SC they take a few hits before dying. Something like this changes a game too much and even the best and fastest players won't always be able to react fast enough to keep game changing things from happening.
|
Ugh...liquidbet, I knew it was a mistake not to bet on all the zergs making Ro4. As a result I only got 1/4 right on the quarterfinals. Dropped down to rank 40 something now
|
Because you are putting a volatile game ON TOP of volatile tournament formats.
In a game where one small micro mistake can cause you the game, or one cheese/all-in can win/lose you a game, a Bo3 format to see who advances is NOT good enough (and sometimes even Bo1's like up and downs).
The OP compares it to other sports; well look at NBA or NHL for example, the teams play a entire season, multiple games against every other team; and then during the play-off, they play a Bo9. Take Tennis for example, each player has to win at least 2-3 SETS to advance, and within those sets, they have to win 7 games, and within each games, they must score 4 points; and even then, the additional rule where you must win by at least 2 points for games/sets/match makes it even less volatile. Do upset games happen in these sports? Of course, but the fact the formats prevent the better team/player from advancing due to a simple upset makes the sport feels alot less volatile.
Starcraft would have similar result too if everyone were to play a year long league, and then play a Bo7 or Bo9 single elimination playoff. But the length of the game makes it impractical, and there is really no way around it from a tournament format standpoint. So the only thing really is to make the game itself less volatile
|
a drastically lesser player wouldn't be in the round of 8, so the entire underlying assumption is wrong to begin with.
Whenever artosis predicts X stomping Y just makes me laugh. Artosis knows absolutely zit about each pro's condition and their practicing regime, as those are house secrets. And when two players meet at ro8, or 4, they are already best of the best sc2 have to offer and it is actually not that likely for one side to completely stomp the other.
Meanwhile, if this game is anywhere close to volatile as people are suggesting it to be. Koreans would not be able to destroying the open bracket of every foreign tournament they enter and then proceed to destroy the rest of the tourneys. The sheer mass of foreigners they go through should get lucky and knock the koreans out, but they don't. And we wouldn't be seeing the same faces over and over in code S where there are hundreds in code A and B trying to take their spots
|
On February 23 2013 08:57 BigKahunaBurger wrote: That is the main problem: The only was to determine how good a player is, is to look at who he beats. Then, by comparing "Person A beat Person B who beat Person C", you end up with "Person A beats Person C". That's not how Starcraft works.
TSLCenter vs IM_Seed. OMG WHAT AN UPSET, LESSER PLAYER BEATS THE GUY WHO SHOULD HAVE WON THIS GAME BLOWS.
Who's to say Center isn't actually better?
Just because Seed has a championship, it actually doesnt matter. People focus too much on past history, when all that matters in Starcraft is that game, right then and there. Center went on to beat Seed again, showing it wasn't a fluke, and that he actually is the better players.
I feel like your argument is a bit contradictory, because according to you there is no way to determine who is a better player. You can't then use the argument of Center beating Seed to say that Center is a better player.
And winning a major tournament does matter - it clearly implies that the player is quite skilled, so people expect those people to play better than those who haven't shown good results in a long time.
You're talking as if history doesn't matter at all, that its impossible to judge skill except for in the moment. Do you expect pro-athletes like Djokovic to do well in grand slam tennis tournaments, or do you really have no expectations about how well the world number 1 will do against some random player ranked 220th in the world?
You can't just say "that's not how Starcraft works", but not back it up with any reasoning. Every sport has the psychological pressures you're talking about, and they all have their own revolutions as far as play styles and strategies are concerned. But we observe consistency in every other sport; its not a "ridiculous standard", its standard practice.
So clearly there is something different with the game. Maybe it has to do with the over reliance on the so-called "metagame" (aka statistically glorified guesswork), leading to flaws in a playstyle that can be exposed (metametagaming)? But then why would they do it so much if it was so flawed? Maybe its the best option?
Maybe there is too much of a luck element, in terms of how difficult it is to scout your opponent (i.e. Stephano losing to a dark templar rush that wasn't done since the beta, just because he missed scouting the dark shrine with his overlord). You say in 9/10 games a player just plays "better". But this is a vague and arbitrary appeal to your authority about what constitutes "better" play. How much do we *really* know about playing SC2 at the top level?
The point remains, why would a top player somehow not play as well as a less accomplished player, and why would it happen with such regularity?
I'm not sure how bad it really is anyway. MVP won several GSL titles, and so did Nestea. And they still do fairly well; so there is some constancy. But we certainly have seen the rise and fall of many, many progamers.
|
Koreans have been raping foreigners pretty consistently, and increasingly consistently as WoL matured.
Also half the time when good players are beaten by "inferior" players you can usually see exactly why in the games themselves, where the good player was either figured out or was just slumping/outplayed.
|
|
|
|