| Life In 19x19 http://lifein19x19.com/ |
|
| Which way to select a winner, opinions? http://lifein19x19.com/viewtopic.php?f=45&t=11713 |
Page 1 of 1 |
| Author: | Matti [ Wed Apr 15, 2015 7:33 am ] |
| Post subject: | Which way to select a winner, opinions? |
We have a tournament of N rounds. The primary goal is to find the winner. The initial grouping and the pairng is done in a way that after N-1 rounds we have two players at top. We have either with full score (wins or McMahon points) and one with one point less (1;1) or we have no player with full score and two players with one point less (0;2). They may or may not have played against each other. The question is how to determine the winner. Below are some options, but you may suggest more. Case (1;1), no mutual game:
Play a final game with the two players and use tie breakers to decide the winner of the tournament. Case (1;1), mutual game already played:
Play a final game with the two players and declare the winner to the winner of the tournament. Play a game with the two players and use tie breakers to decide the winner of the tournament. Case (0;2), no mutual game:
Case (0;2), mutual game already played:
Declare the player having won the mutual game the winner and give both players a free round. Pair both players down and use tie breakers to decide the winner of the tournament. In all cases I prefer the first choice. What do you prefer? |
|
| Author: | HermanHiddema [ Wed Apr 15, 2015 7:45 am ] |
| Post subject: | Re: Which way to select a winner, opinions? |
Case (0;2): Always play a final game, winner wins the tournament Case (1;1): Always play a final game, winner wins the tournament, but give the player who is ahead black and give white no komi. |
|
| Author: | Kirby [ Wed Apr 15, 2015 9:25 am ] |
| Post subject: | Re: Which way to select a winner, opinions? |
Case (1;1), no mutual game: Play a final game with the two players and declare the winner as the winner of the tournament. Case (1;1), mutual game already played: Play a game with the two players and use tie breakers to decide the winner of the tournament. Case (0;2): Play a final game with the two players and declare the winner to the winner of the tournament. |
|
| Author: | Matti [ Wed Apr 15, 2015 11:26 pm ] |
| Post subject: | Re: Which way to select a winner, opinions? |
HermanHiddema wrote: Case (1;1): Always play a final game, winner wins the tournament, but give the player who is ahead black and give white no komi. An interesting idea. |
|
| Author: | HermanHiddema [ Thu Apr 16, 2015 12:01 am ] |
| Post subject: | Re: Which way to select a winner, opinions? |
@Matti: BTW, my opinion for (1;1) here is very specifically for those cases where there absolutely must be a single winner (i.e title game or large/important prize that cannot be shared), and where there is only time for one game. If there is time for more games, I would prefer for the (1;1) case that the players play a regular game, which will result in either (1;0) which has a winner or (0;2) and another game between them (effectively, this is like double KO) |
|
| Author: | Matti [ Thu Apr 16, 2015 4:26 am ] |
| Post subject: | Re: Which way to select a winner, opinions? |
About the (1;1) case. Your opinion is that whether the players have played before or not, makes no difference. You prefer them to play a game or two. Let's change the setup a bit. Players play N rounds (not N-1) before we get any of the choices listed in my first post. We have time to play one more game if needed. So the difference is that the lower players would be free during this game. Would you make the same choices in each case? |
|
| Author: | HermanHiddema [ Thu Apr 16, 2015 4:39 am ] |
| Post subject: | Re: Which way to select a winner, opinions? |
It depends. If a tournament is advertised as N rounds in advance, and there is a single leader after N rounds, then I would consider that player the winner. If two players are tied, then I prefer a play-off above other tie-breaking methods. (again, in the context of a single significant indivisible prize) If, on the other hand, the tournament is advertised as e.g. "double elimination Swiss" with "at least N rounds for all players", then I would hold these preferences (because neither player has been eliminated by two losses yet in either the (0;2) or the (1;1) case). |
|
| Author: | Matti [ Fri Apr 17, 2015 1:24 am ] |
| Post subject: | Re: Which way to select a winner, opinions? |
I am planning a title tournament. Double elimination Swiss would be ideal. If we can use 4 days for the tournament it is doable, but if we have only 3 days it is more difficult. Strict double elimination means that one cannot predict exactly the number of rounds needed. When we reach (1;1) players at top we might finish on the next round with (1;0) or we might reach (0;2) and need another game. We might by chance avoid (1;1), but we cannot count on that. There are different ways on handling the one round uncertainty of deciding the title:
2) Have an option to miss one game on the last round, if the title is already decided uniquely (only with number of wins or McMahon score). 3) Define that two finalists are selected with N-1 rounds, after they play a final to decide the title 4) combination of 2) and 3) It seems that other writers in this thread don't like choice 2) or did not understand it. If I had four days for the tournament, I would choose option 1). If I had three days, I would choose differently. |
|
| Author: | HermanHiddema [ Fri Apr 17, 2015 2:58 am ] |
| Post subject: | Re: Which way to select a winner, opinions? |
@Matti: If it is important to control the number of rounds, you can use McMahon instead of Swiss, and optionally byes, to exactly control the group sizes. More important than the number of days, perhaps, is the number of rounds. How many rounds can you hold? I am coincidentally currently writing a proposal myself for a double elimination McMahon type system with 8 rounds (4 days, two rounds per day), which works like this: The field is divided into groups of exactly 16 players (McMahon). Only the strongest two groups have a chance of winning the title (the second group effectively start with one loss for the purpose of the double elimination). With this system, it is entirely predictable that there will be one player on 4 points after 4 rounds and there will be 5 with 3 points (including those with 4 wins who started in the second group). Then if you pair those 6 players among themselves, you will get to either (1;2) or (0;4) after 5 rounds. The second case is easy, play two rounds KO. In the first case, pair the player with 5 points down aganst someone with 3, while those with 4 play each other. Then you will get either (1;1) or (0;2) and are at your situation. My proposal has room for two more rounds, and can therefore do proper double elimination. To guarantee 7 rounds, you could do the (1;1) case as a play-off without komi, as I suggested earlier. If you need fewer rounds, you can start with (8;16). Then after 3 rounds there will be (1;5) and you can proceed as above. If the number of players does not fill the groups, you can use byes to manage it. (e.g if there are only 28 players, so only 12 players in the second group, then after 2 rounds you'll be at (4;13) and I would give one of the 13 a bye to get to (2;9), then another bye to get to (1;5). |
|
| Author: | Matti [ Fri Apr 17, 2015 5:06 am ] |
| Post subject: | Re: Which way to select a winner, opinions? |
Your tournament proposal is almost the same to what I have in mind. 8 rounds in 4 days would be easy. If I get three days, I would try to squeeze 7 rounds in it, possibly 6 rounds and a separate final. Later I am going ask the relevant players which way they prefer: 4 days with longer thinking times or 3 days with shorter ones. |
|
| Author: | Matti [ Tue Apr 21, 2015 4:04 am ] |
| Post subject: | Re: Which way to select a winner, opinions? |
HermanHiddema wrote: If the number of players does not fill the groups, you can use byes to manage it. (e.g if there are only 28 players, so only 12 players in the second group, then after 2 rounds you'll be at (4;13) and I would give one of the 13 a bye to get to (2;9), then another bye to get to (1;5). One could pair a player down instead of giving him a bye. Bye may be needed only very late say with (0;3) if you want to avoid finding the winner prematurely by getting (0;1) instead of (0;2). |
|
| Page 1 of 1 | All times are UTC - 8 hours [ DST ] |
| Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group http://www.phpbb.com/ |
|