Basically because of the consecutive coin flips scenario. The chances of you flipping heads 10 times in a row are really really small, but that doesn't mean that after the 9th flip, the chances of the coin ending up heads are any smaller. It's still 50-50.

So if Idra has ~75% chance of beating someone, he will have that chance in every match (not to mention he only has to win 2 out of 3 or 3 out of 5, which gives a good room for minor errors). That intial chance that will always be high, and the room for error should turn up with a higher percentage than 43.

And even so, if Idra has 43% chance of winning, it means that others in the top 16 have ~5% to ~20% tops, which would STILL mean that Idra is twice as likely to win this than anyone else, were you to calculate everything in percentages that is.

So yeah, you can't really do that in a tournament because there are many factors to take into account, and even if you do, you won't have complete results unless you do it for other players too (or at least favorites and such).

Kyo Yuy

United States1286 Posts

December 28 2009 14:12 GMT

#70

It makes me happy that G5 is rooting for KawaiiRice in group H.

He's one of my favorite players. Up and coming, very young, and he likes Jpop which is an auto win in my book <3.

KAWAIIRICE FIGHTING!!!!!

FragKrag

United States11566 Posts

December 28 2009 17:36 GMT

#71

oh man White-Ra is a fighter ;D

I really hope Cloud gets a chance to advance from his group. <3 Cloud

Pablols

Chile519 Posts

December 28 2009 18:47 GMT

#72

killer will win the tsl 98% certain

Jaksiel

United States4130 Posts

December 28 2009 20:55 GMT

#73

That math doesn't work because the matches aren't Bo1s. (Also, you only need 4 wins from Ro16 on, not 5.)

White-Ra really is everywhere if there are lots of little White-Ras running around the Ukraine.

Cambium

United States16368 Posts

December 28 2009 23:59 GMT

#74

On December 29 2009 03:47 Pablols wrote:
killer will win the tsl 98% certain

94.324% certain you are wrong =\

GL G5, Cloud and White-ra!

Cambium

United States16368 Posts

December 29 2009 00:25 GMT

#75

On December 28 2009 02:29 oberon wrote:
"Either way I think IdrA will win this tournament with 90% certainty."

Argh. Stats-man ANGRY!

I think people greatly underestimate the role of chance in tournaments like this (same deal as the OSL/MSL/whatever). If someone is really good at a specific matchup in Korea, they tend to have a 70%+ win-rate. Let's give Idra an 85% win-rate in all three matchups, and a 95% chance at making the RO16. To win the tournament, he'll need to qualify for the RO16, then win 5 consecutive matches: 0.95 * (0.85^5) = 0.422, or less than 43%.

Obviously, this isn't a slight on Idra -- the point is that even 16 players is a lot, and all of the people who make the RO16 are going to be good. To have a 90% chance of winning, even just starting from the RO16, a player would need over a 97% winrate against other RO16 players.

Having now wandered wildly off-topic -- these interviews are always awesome. Are we going to all the groups before group play starts?

Except the games are not BO1...
+ Show Spoiler [for people who are bored (UPDATED)] +

Let's say idra has 85% chance to win in all three matchups against any opponent, and 95% chance of advancing to round of 16.

Assuming the tournament is BO5 all the way to the finals (I am ignorant of the format, i'm just doing this to prove a point), you need to win 3 in every BO5

You can win in either 3 games, 4 games or 5 games

Case 1, three games: x ^ 3 = A, P(case 1) = A
Case 2, four games: 4C3 * x ^ 3 * (1 - x) - x ^ 3 * (1 - x) = B
Case 3, five games: 5C3 * x ^ 3 * (1 - x) ^ 2 - 4C3 x ^ 3 * (1 - x) ^ 2 = C

You add all of those together, you get:
P(case 1) = 0.614125
P(case 2) = 0.276
P(case 3) = 0.083

Total = .97

So Idra has 97% chance of winning any BO5

There are 4 BO series (16 - 8 - 4 - 2 -> win), assuming 95% chance of getting into round of 16, the total chance of winning is

.95 * .973 ^ 4 = 85%

Jaksiel

United States4130 Posts

December 29 2009 00:59 GMT

#76

Cambium, your math is wrong as well. Just intuitively, it should be clear if Idra is 85% to win any one game he will be more than 85% to win a Bo5. The full math:

Idra wins in 3: .85^3 = .614 * 1 scenario (WWW) = .614
Idra wins in 4: .85^3*.15 = .092 * 3 scenarios (LWWW, WLWW, WWLW) = .276
Idra wins in 5: .85^3*.15^2 = .013 * 6 scenarios (LLWWW, LWLWW, LWWLW, WLLWW, WLWLW, WWLLW) = .083

So Idra's probability of winning a Bo5 would be .614 + .276 +.083 = .973, a massive 97%. Thus, his probability of winning the tournament would be .95 * .973^4, which comes out to 85%. Still not quite 90%, but very, very high.

Cambium

United States16368 Posts

December 29 2009 01:24 GMT

#77

On December 29 2009 09:59 Jaksiel wrote:
Idra wins in 3: .85^3 = .614 * 1 scenario (WWW) = .614
Idra wins in 4: .85^3*.15 = .092 * 3 scenarios (LWWW, WLWW, WWLW) = .276
Idra wins in 5: .85^3*.15^2 = .013 * 6 scenarios (LLWWW, LWLWW, LWWLW, WLLWW, WLWLW, WWLLW) = .083

So Idra's probability of winning a Bo5 would be .614 + .276 +.083 = .973, a massive 97%. Thus, his probability of winning the tournament would be .95 * .973^4, which comes out to 85%. Still not quite 90%, but very, very high.

edit: nvm he's right. Overlooked two things.

huameng

United States1133 Posts

December 29 2009 01:50 GMT

#78

The simplest way to do the math I think: find the probability that idra wins at least 3 games out of 5. Don't stop after idra has won 3 times. Obviously they would stop playing after idra wins 3, but this won't overcount anything since all permutations of form WWWxx will be counted. Nothing will be missed since idra can't both win 3 games and lose 3 games out of 5. The probability is what Jaksiel calculated, about 97 and 1/3rd %.

Also of course if he wins 1 game 85% of the time, he wins a BO5 more than 85% of the time. A 15% likely occurance has to happen 3 times out of 5, how could that happen 1-.75 = 1/4th of the time?