EPTOn August 03 2009 01:30 tirentu wrote:
Yes.
Why?
JAEDONG.
Yes.
Why?
JAEDONG.
pretty much
Do you like the new Super-Ace PL Format? - Page 5
| Forum Index > Polls & Liquibet |
|
StorrZerg
United States13910 Posts
pretty much | ||
|
ThePhan2m
Norway2739 Posts
| ||
|
thopol
Japan4560 Posts
STX should NOT have lost in the first round. | ||
|
Doso
Germany769 Posts
| ||
|
7mk
Germany10156 Posts
It is bad for 'depth' since teams like Oz have two tries to force it to an ace match. But I voted yes because it's great for us viewers cause it adds so much excitement. | ||
|
FreeDoM[YA]
Canada855 Posts
+1. I've never liked that one game decides everything, it seems really... off to me. | ||
|
CharlieMurphy
United States22895 Posts
| ||
|
ShAsTa
Belgium2841 Posts
On August 04 2009 06:34 FreeDoM[YA] wrote: +1. I've never liked that one game decides everything, it seems really... off to me. | ||
|
ghostWriter
United States3302 Posts
| ||
|
StarBrift
Sweden1761 Posts
On August 03 2009 12:12 Gustav_Wind wrote: Hot Bid's reasoning is correct. I'll just make up some numbers for the sake of making the example easier. Let's say Oz's chance of winning a Bo7 vs CJ is 40%, because CJ is a deeper team, and that Jaedong's chances of winning an ace match is 80%. In a normal single Bo7 format, Oz's chance of winning is 40%. Simple. In this format, Oz's chance of winning is (chance of winning both sets) + (chance of winning 1 of the two sets)*(chance of winning super ace). That is, (.4*.4) + (.4*.6*2)*.8. This comes out to .544, or 54.4%. Extremely flawed equations. Your numbers are taken out of thin air. There is no way to define a number of success in exact percentages unless you take into account all the factors (like the stats of all players, their recent performances, their practise hours, their other commitments, map balance, specific matchup preparation and the list goes on and on and on). But if indeed you are making up your own numbers please have some that are likely and have some grounds to them. Oz has a 40% chance of winning overall? Where did you get that number? Off of the top of your head? And what makes you estimate Jaedongs likelyhood to win Ace matches int he proleague playoffs to 80%. You think Bisu/Fantasy or Effort only stood a 20% chance of beating him? I know you agree with HotBid but you can't prove his point by numbers. It's impossible. | ||
|
Aznleeman
United States208 Posts
| ||
Gnaix
United States438 Posts
| ||
|
Dagobert
Netherlands1858 Posts
1) Probability to win. There are countless ways to calculate this, and none of them refers to probability to win in a single case. 80% probability of winning doesn't mean he's got .8 of the game in the bag, it means he is expected to win (for instance) 80 out of 100 games under the given hypothetical circumstances as depicted by the data. 2) Equal footing. Players will never stand equal chances at winning because we do not have the measures necessary to check that. You could standardize all you want (same living quarters, same mouse, practice time, number of other games, etc.) and they would *still* not be on equal grounds. 3) Achievement. "But now they only have to win 2 non-ace games". Sure. If you think so lowly of one team's members that you consider 2 (3) wins worth little to nothing, why are you arguing at all? 4) The "Ace Player" argument: "Better Ace player wins it, even if bla wins more games." This argument doesn't hold, since not everyone gets to play. The Super Ace would be obsolete if there were 7 games played on a day. If they wanted a second day so bad, they could award points for wins, and only if these turned out 7 : 7 would there be a Super Ace Match necessary. However, since the series are stopped at 4:x, we never know how the others would have fared. Maybe Oz would've won the first day 6:1 and lost the second 3:4, they would've won 9:5. Simply 'having more points' doesn't mean much right now. "But a Bo7 always ends when one team has 4 wins." - "Yes, and the team that wins the super-ace wins, too. Rules can be a pain if everyone's gotta stick to them, right?" | ||
|
Kong John
Denmark1020 Posts
| ||
|
hyst.eric.al
United States2332 Posts
| ||
|
gumbum8
United States721 Posts
| ||
|
Gustav_Wind
United States646 Posts
On August 04 2009 07:51 Dagobert wrote: Gustav is mixing up things here. 1) Probability to win. There are countless ways to calculate this, and none of them refers to probability to win in a single case. 80% probability of winning doesn't mean he's got .8 of the game in the bag, it means he is expected to win (for instance) 80 out of 100 games under the given hypothetical circumstances as depicted by the data. I'm not mixing anything up. I understand that 80% chance to win a game means in the long run 80% of games will be won. This can still be applied to a single game. My calculations used exactly this fact. I don't see people making posts like this when the people who write the TL news come out with stats articles that give probabilities of players winning single games. On August 04 2009 07:00 StarBrift wrote: Extremely flawed equations. Your numbers are taken out of thin air. There is no way to define a number of success in exact percentages unless you take into account all the factors (like the stats of all players, their recent performances, their practise hours, their other commitments, map balance, specific matchup preparation and the list goes on and on and on). But if indeed you are making up your own numbers please have some that are likely and have some grounds to them. Oz has a 40% chance of winning overall? Where did you get that number? Off of the top of your head? And what makes you estimate Jaedongs likelyhood to win Ace matches int he proleague playoffs to 80%. You think Bisu/Fantasy or Effort only stood a 20% chance of beating him? I know you agree with HotBid but you can't prove his point by numbers. It's impossible. ...Nice job using the fact that I made up numbers to prove my argument wrong when the first thing I said was that I made up numbers as a demonstration, not a proof. Choose any numbers that are comparable and do the calculations. You will see that Oz's chances improve in the Super Ace format vs the Single Bo7 format. THE SPECIFIC NUMBERS DON'T MATTER. I only made up numbers because I didn't think you would actually read a real proof. Here it is: + Show Spoiler + Let p = the probability of a team winning vs the other team in a single Bo7 Let a = and let a be the probability of that team's Ace winning an ace match vs the other team's Ace. Since I am proving that the Super Ace format favors teams with dominant aces, we are going to assume that a > .5. The probability of that team winning in the Super Ace format is this: (chance of winning both sets) + (chance of winning 1 of the two sets)*(chance of winning super ace) using our notation, it's p^2 + 2p(1-p)a so let us subtract the probability of the single Bo7 win from this number. If the result is positive, then this will show that the Super Ace format gives them an increased chance of winning from the single Bo7. p^2 + 2p(1-p)a - p using algebra: p^2 + 2ap - 2ap^2 - p p^2 - 2ap^2 - p + 2ap p^2(1-2a) - p(1-2a) (1-2a)(p^2 - p) (1-2a)(p-1)p (2a-1)(1-p)p Final Result: (2a-1)(1-p)p we know p and 1-p have to be positive, since p is a probability and therefore is between 0 and 1. Since a is > .5, 2a-1 is also positive. Therefore, the net result is positive. The Super Ace format favors the teams with more dominant aces. The more dominant the ace is, the better it is for them. The rest of the team is still very relevant, but to say that the format doesn't favor Aces is false. | ||
|
Dagobert
Netherlands1858 Posts
Lol did you even write what I posted? I understand that 80% chance to win a game means in the long run 80% of games will be won. This can still be applied to a single game. My calculations used exactly this fact. I don't see people making posts like this when the people who write the TL news come out with stats articles that give probabilities of players winning single games. No, I didn't *write* what you posted. And obviously, you do not understand. Besides, I don't care who writes the stat articles. Again, you're confusing some things here. a) Probability - how often is the observation expected to be made given certain data? b) Predictions - which observation will be made? Now what's the difference between the two? By using a), you can summarize data ("if I were to repeatedly [and randomly] pick one game out of Bisu's matches against Zerg, I would probably end up with 67 of 100 games where he wins it"). That's neat, but doesn't help if you want to make a prediction ("Will Bisu win vs Jaedong?"). To make a prediction, you need to *test* the predictive formula you're using. Note: This is called a model, and it is not chosen on the grounds of how well it describes existing data (e.g. winning percentages) but on how well it predicts future outcomes. | ||
|
Gustav_Wind
United States646 Posts
On August 04 2009 09:17 Dagobert wrote: No, I didn't *write* what you posted. And obviously, you do not understand. Besides, I don't care who writes the stat articles. Again, you're confusing some things here. a) Probability - how often is the observation expected to be made given certain data? b) Predictions - which observation will be made? Now what's the difference between the two? By using a), you can summarize data. That's neat, but doesn't help if you want to make a prediction. To make a prediction, you need to *test* the predictive formula you're using. Note: This is called a model, and it is not chosen on the grounds of how well it describes existing data (e.g. winning percentages) but on how well it predicts future outcomes. I'm not creating a model or a prediction because I'm not trying to actually figure out winning percentages. I'm saying that no matter what those specific percentages are, the Super Ace format favors teams with dominant aces more than the standard single Bo7 format. Nothing needs to be tested because no predictions are being made. It's a proof. | ||
|
lueiGi2
Canada24 Posts
| ||
| ||
