Page 5 of 6

Re: 17th Nongshim Cup

Posted: Tue Mar 01, 2016 1:29 pm
by jeromie
emeraldemon wrote:Here is goratings prediction for each player left in the table:

Ke Jie beats Lee Sedol : 64.271 %
Ke Jie beats Lian Xiao : 70.098 %
Ke Jie beats Iyama Yuta : 61.314 %
Ke Jie beats Murakawa Daisuke : 92.115 %
Lee Sedol beats Lian Xiao : 56.582 %
Lee Sedol beats Iyama Yuta : 46.838 %
Lee Sedol beats Murakawa Daisuke : 86.656 %
Lian Xiao beats Iyama Yuta : 40.337 %
Lian Xiao beats Murakawa Daisuke : 83.286 %
Iyama Yuta beats Murakawa Daisuke : 88.053 %

Now some smart person tell me the odds for each team to win based on these...
Based on these numbers, Korea has an 8.2% chance of winning, Japan has a 22.3% chance of winning, and China has a 69.5% chance of winning.

I'd show my work, but it's too messy. :-)

Re: 17th Nongshim Cup

Posted: Tue Mar 01, 2016 2:10 pm
by yoyoma
I thought I would write a quite python script to iterate over the combinations and calculate it... It seems like it should be simple but I gave up. :o If someone writes one post the code. :geek:

Re: 17th Nongshim Cup

Posted: Tue Mar 01, 2016 2:58 pm
by wineandgolover
Maybe the players can just roll Elo-weighted dice, and finish the whole series tonight without any messy byo yomi problems. ;-)

Re: 17th Nongshim Cup

Posted: Tue Mar 01, 2016 3:40 pm
by jeromie
yoyoma wrote:I thought I would write a quite python script to iterate over the combinations and calculate it... It seems like it should be simple but I gave up. :o If someone writes one post the code. :geek:
I can share my spreadsheet if you want. :-)
wineandgolover wrote:Maybe the players can just roll Elo-weighted dice, and finish the whole series tonight without any messy byo yomi problems. ;-)
Yeah, even though I calculated the odds I think that putting too much weight on such things is kind of silly. Each game still has to be played on the board. All the odds really do is confirm (and quantify) what we knew: China is still the favorite to win the tournament, but they aren't a lock.

Re: 17th Nongshim Cup

Posted: Wed Mar 02, 2016 2:23 am
by trout
11th game;
Lee Sedol defeated Murakawa Daisuke by resign.

Remaining players;
Korea:Lee Sedol, Choi Cheolhan, Min Sangyeon, Baek Chanhee, Park Junghwan
China:Ke Jie, Lian Xiao, Wu Guangya, Fan Yunruo, Gu Li
Japan:Iyama Yuuta, Murakawa Daisuke,Ida Atzshi, Ichiriki Ryo, Kono Rin
bold - Remaining players.



3rd round: 3/1/2016~3/5/2016

Re: 17th Nongshim Cup

Posted: Wed Mar 02, 2016 8:20 am
by hyperpape
jeromie wrote:Yeah, even though I calculated the odds I think that putting too much weight on such things is kind of silly. Each game still has to be played on the board. All the odds really do is confirm (and quantify) what we knew: China is still the favorite to win the tournament, but they aren't a lock.
Taking current ratings and a cut-off of 50 ELO points difference as the dividing line between a coin-flip and a surprise, we get the following:

Predicted: Ichiriki over Baek
Surprise: Ichiriki over Fan
Coin Flip: Ichiriki over Min
Predicted: Chen over Ichiriki
Predicted: Choi over Chen
Predicted: Choi over Ida
Coin Flip: Gu over Choi
Predicted: Gu over Kono
Surprise: Gu over Park
Surprise: Murakawa over Gu
Predicted: Lee over Murakawa

So we have 6 accurate predictions and 3 surprises, which I'm guessing is itself not surprising. Maybe I need to tighten up my criterion of "surprise": 50 ELO is really not that big.

What that underscores is that most of these predictions are pretty modest. Take the Chen over Ichiriki game: it's big enough that I called it predicted, but the prediction is just that Chen has something like a 63% chance of winning. That's good, but if it had gone the other way, we wouldn't be shocked.

Re: 17th Nongshim Cup

Posted: Wed Mar 02, 2016 8:25 am
by hyperpape
yoyoma wrote:I thought I would write a quite python script to iterate over the combinations and calculate it... It seems like it should be simple but I gave up. :o If someone writes one post the code. :geek:
I almost got it done yesterday evening in JS. I've been mulling over making a little app to simulate tournaments, and maybe this will be the kick to do it.

Wouldn't call it too simple: if you use straightforward recursion, there's a fair bit of book-keeping.

Re: 17th Nongshim Cup

Posted: Wed Mar 02, 2016 2:25 pm
by yoyoma
Got my python script working. :ugeek: I get similar numbers to jeromie, differences could be rounding errors, or because I looked at today's goratings and they probably changed some. Anyways from before Lee Sedolbeat Murakawa Daisuke:
J = 0.2273
C = 0.6946
K = 0.078

pastebin of the python: http://pastebin.com/MARi39r0 (Oops I messed up the paste, line 57 is repeated, delete one of them if you try this)

Odds before Murakawa Daisuke > Gu Li. Impact of Murakawa's win in parenthesis

Code: Select all

    before   after  difference
J = 0.1527  0.2273 (+0.0746)
C = 0.7885  0.6946 (-0.0939)
K = 0.0588  0.078  (+0.0192)
And finally after Lee Sedol > Murakawa Daisuke:
J = 0.2201
C = 0.6887
K = 0.0913

Re: 17th Nongshim Cup

Posted: Wed Mar 02, 2016 4:48 pm
by seigenblues
We'll be broadcasting Lian Xiao vs Lee Sedol tonight on the AGA Youtube & twitch! the usual...

Re: 17th Nongshim Cup

Posted: Wed Mar 02, 2016 9:13 pm
by Uberdude
seigenblues wrote:We'll be broadcasting Lian Xiao vs Lee Sedol tonight on the AGA Youtube & twitch! the usual...
Thanks for doing all these broadcasts. It's great that there is now too much quality English Go content for me to consume!

Re: 17th Nongshim Cup

Posted: Thu Mar 03, 2016 2:22 am
by trout
12th game;
Lee Sedol defeated Lian Xiao, by resign.

Players;
Korea:Lee Sedol, Choi Cheolhan, Min Sangyeon, Baek Chanhee, Park Junghwan
China:Ke Jie, Lian Xiao, Wu Guangya, Fan Yunruo, Gu Li
Japan:Iyama Yuuta, Murakawa Daisuke,Ida Atzshi, Ichiriki Ryo, Kono Rin
bold - Remaining players.



3rd round: 3/1/2016~3/5/2016

Re: 17th Nongshim Cup

Posted: Thu Mar 03, 2016 3:23 am
by PlaySlow
seigenblues wrote:We'll be broadcasting Lian Xiao vs Lee Sedol tonight on the AGA Youtube & twitch! the usual...
Thank you for your coverage! It really provides us great insight. However it would be very good if you can start coverage from the first move. You are skipping the opening phase and start commenting on variations starting from mid game generally. I hope you would cover fuseki too in the incoming games:)

Re: 17th Nongshim Cup

Posted: Thu Mar 03, 2016 5:15 am
by RedStick
@playslow. I haven't watched yesterday's game yet but I've seen most of their streams and they pretty much always get to the fuseki at some point although it can sometimes be like an hour in and usually when the players are doing a lot of thinking. A review that started from the first move would be very long and would drag a lot more. I like the approach of starting with a middle game position and then backing up to see how they got there.

Re: 17th Nongshim Cup

Posted: Thu Mar 03, 2016 9:33 am
by jeromie
So now we get Lee Sedol vs Iyama Yuta with the winner going on to face Ke Jie. What a great end to this tournament!

Re: 17th Nongshim Cup

Posted: Fri Mar 04, 2016 2:14 am
by trout
13th game;
Lee Sedol defeated Iyama Yuuta by resign.

Players;
Korea:Lee Sedol, Choi Cheolhan, Min Sangyeon, Baek Chanhee, Park Junghwan
China:Ke Jie, Lian Xiao, Wu Guangya, Fan Yunruo, Gu Li
Japan:Iyama Yuuta, Murakawa Daisuke,Ida Atzshi, Ichiriki Ryo, Kono Rin
bold - Remaining players.



3rd round: 3/1/2016~3/5/2016