Author |
Topic: 2008 World Championship Format (Read 7336 times) |
|
omar
Forum Guru
Arimaa player #2
Gender: 
Posts: 1003
|
 |
Re: 2008 World Championship Format
« Reply #30 on: Oct 5th, 2007, 5:36am » |
 Quote  Modify
|
Another thing to think about is how to rank the players in the swiss tournament. Do we want to use any kind of ratings for this or not. One way to not use any ratings would be to just rank the players randomly in each round. Byes could also be given randomly from the set of players who have received the fewest byes so far.
There are also other methods of ranking that don't rely on initial ratings:
http://en.wikipedia.org/wiki/Buchholz_chess_rating
Should we look into using one of these?
Also exactly how many rounds swiss rounds should we have. Is 5 the right number if we know we want to select the 8 best at the end; or should it be something else? Of course it also depends on how many initial players there are. If there are ties in selecting the 8 best how do we break them?
Looks like I already have this swiss pairing program from before. Should we just go with this one?
http://arimaa.com/arimaa/tourn/compare/sim/formats/swissTypical
|
| « Last Edit: Oct 5th, 2007, 5:57am by omar » |
 IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: 2008 World Championship Format
« Reply #31 on: Oct 5th, 2007, 8:42am » |
Quote Modify
|
Sounds like a plan. If we don't want the qualifier to interfere with Christmas and New Year's, we should open registration and start publicizing soon, perhaps even before the format is finalized. I assume there will be a holiday break between tournaments, with the finals starting in January.
The "typical swiss" you linked to looks fine, and has the virtue of already being coded, which is important under scheduling pressure. That page recommends lg(N)+2 rounds, i.e. 6 rounds for 16 players and 7 rounds for 32 players. However, that formula probably arises from wanting a clear single champion, and is accounting for the high probability of draws in chess.
I like five rounds for the qualifier. There will always be a problem with ties for the final places, but the more rounds there are, the better the tiebreakers work. In a 24-player four-round swiss, you might have a situation where folks with 4-0 and 3-1 records are in, and exactly one of the nine 2-2 players enters on tiebreaks, with a fair chance that the top two 2-2 players had the same strength of schedule. With 24 players and five rounds instead, the 5-0 and 4-1 players will get in, plus half of the seven or eight 3-2 players. The typical tiebreak is sum of wins of opponents, and by round five it is clearer who got to 3-2 by a rough road and who backed in.
|
| « Last Edit: Oct 5th, 2007, 8:42am by Fritzlein » |
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: 2008 World Championship Format
« Reply #32 on: Oct 5th, 2007, 9:30am » |
Quote Modify
|
By the way, I know that some people don't want pre-tournament ratings to have any effect whatsoever on the process, but I think that would be an over-correction to the problems we had with the #1 seed in the past two World Championship (me) getting too many breaks. The "typical swiss" format uses the pre-tournament rating, but doesn't favor the top seed(s) nearly as much.
For example, the swiss pairing tries to be sliding, not folding. So instead of 1 vs. 16; 2 vs. 15; 3 vs. 14; etc. it pairs 1 vs. 9; 2 vs. 10; 3 vs. 11; etc. This not only means the games have a more equal amount of mismatch, it also make the post-tournament tiebreaks more informative.
Also, the swiss pairing gives any bye to the lowest-ranked player, not the highest.
Also, as the tournament progresses, the swiss pairing uses the pre-tournament ratings less and less, because it not only relies on in-tournament records first, but because it relies on strength of opponents second, and only uses pre-tournament rating third to distinguish rank.
Finally, if some player must "play down" against a player of a worse record, the swiss pairing code tries to have it be the lowest player in the higher group who plays down, not the highest player in the higher group.
In combination these factors mean that, while pre-tournament ratings are still influential in pairing early rounds, they by no means give an easy path to the top seed. And of course, for the finals the ratings will be completely forgotten, with seeding only on the basis of order of finish in the qualifier, so the influence of pre-tournament ratings will have almost completely attenuated by the time the champion is crowned.
Also I don't think the pre-tournament p8H ratings will be that bad. (By the way, to answer an earlier question, p8 ratings do have a time attenuation factor that gradually draws ratings of inactive players back towards 1500.) Maybe Omar can publish a list so we can see if they look reasonable.
[EDIT]
I should also reiterate the argument for using pre-tournament ratings in the qualification tournament: it makes the results more accurate. Random seeding could lead to silly situations such as the true #4 player facing the #1 player in the first round, losing, and then in the second round having to face the #3 player, who lost to the #2 player in the first round.
With two losses, the #4 player would then get paired against easier opponents, but potentially too easy. After beating players ranked #24, #21, and #17 in the final three rounds, the #4 player has a 3-2 record, but too few tiebreak points to finish in the top eight, and therefore doesn't make it into the final.
Contrast this to using ratings for the pairings, so that the #4 player is more likely to face tougher average opposition, say beating #16, beating #10, losing to #1, beating #8 and losing to #3. Although he had the same two losses and finished with the same 3-2 record, he had a chance to prove something against a tougher schedule, and will therefore have enough tiebreak points to qualify. (Also note that with swiss pairing seeded by ratings, it is reasonably likely that the #4 player will face only one higher-ranked player in five rounds, and therefore have a better shot to finish with a 4-1 record and a decent seed for the final. In the given example his final-round opponent will probably be #6 rather than #3.)
It's not that random seeding of the qualifier makes strange twists of fate inevitable, it just makes them more likely. Naturally random seeding make the selection of the final eight a little bit more random, and a little bit less related to skill.
|
| « Last Edit: Oct 5th, 2007, 7:42pm by Fritzlein » |
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: 2008 World Championship Format
« Reply #33 on: Oct 5th, 2007, 11:04am » |
Quote Modify
|
Floating double elimination would be well-behaved for the finals with the field guaranteed to be eight players, a power of two.
In the first round, there would be folding pairing with #1 vs #8, #2 vs #7, etc. This seems like a reasonable advantage to give to the top seeds, since they earned their seeds in the qualifying tournament, but can't carry their records forward. (Actually the folding pairing is much less of an advantage than carrying forward extra lives would be.)
In the second round, the four winners from round one would play each other, while the four losers play each other, again with folding pairing in each group. Two people will be eliminated in this round.
In the third round, the two undefeated players would necessarily play each other, while the four one-loss players will be paired in a way dictated by avoiding repeat matchups. There will always be a pairing without repeat matchups. Two more players will be eliminated.
By the fourth round, there will be one undefeated player and three one-loss players. It is possible but unlikely that the undefeated player has played each of the one-loss players in the three previous rounds. In that case there would have to be a repeat matchup. Most likely, though, the undefeated player will play against the one-loss player he hasn't beaten yet, while the other two one-loss players play each other.
If undefeated player wins in the fourth round, then it is simple: he plays the other remaining player for the championship, needing to win only once while the one-loss player needs to win twice. On the other hand, if the undefeated player loses in the fourth round, then we have three one-loss players left, and one bye must be assigned in the fifth round. This is the one possible ungraceful moment.
If any of the three possible matchups hasn't occurred yet, then it must occur in round five while the other player gets a bye. More likely, however, all the matchups have already occurred, so there must be a repeat pairing. In that case the highest seed gets the bye. This is one point where I would like to change the way floating double elimination works: instead of giving the bye to the highest seed, it should be given to whichever of the three faced the toughest opposition in the first four rounds, as measured by the total wins of their four opponents. Using the seeds should be a last resort when there is no other way to determine who deserves the bye.
|
|
IP Logged |
|
|
|
mistre
Forum Guru
Gender:
Posts: 553
|
|
Re: 2008 World Championship Format
« Reply #34 on: Oct 5th, 2007, 12:21pm » |
Quote Modify
|
Count me in for a 5-round swiss qualifying tournament as long as the time control is at 60 sec/move or less.
I don't play live much because of time constraints, but a short tournament format sounds like fun.
On another note, I am also interested in participating in the next Postal Tournament. I looked at the results from last year and I was disappointed that there is no ranking or winner. Is there a way to make the Postal Tournament more like the World Championship?
Maybe if there was a time control of 1d/1d/100/2d/0 - A fast postal if you will, then more of a tournament structure could be set up.
|
| « Last Edit: Oct 5th, 2007, 12:23pm by mistre » |
IP Logged |
|
|
|
chessandgo
Forum Guru
Arimaa player #1889
Gender:
Posts: 1244
|
|
Re: 2008 World Championship Format
« Reply #35 on: Oct 5th, 2007, 1:40pm » |
Quote Modify
|
on Oct 4th, 2007, 10:02pm, 99of9 wrote:| The first 5 rounds would be at a short time control so don't require as major time commitments. |
|
It seems to me that the hardest part, when scheduling a game, is to find a common spot, not really the potential length of the game. It's much more difficult to play two 45 sec/move games than one 90sec/move game, for two common slots with two opponents must be found. So I'd say the actual length of one game is not the problem, when it comes to making an arimma tourney fit in one's timetable ; it's rather the total number of games to be played.
So I don't think playing the premilinary round at a short time control would do much good.
Mistre : I'm not sure I understand your statement. Do you say you'd play in a 5-rounds swiss tourney but not in a 2-lives floating elimination tourney, as was used last year ? It seems to me that, even if the swiss tourney is played at a realatively short time control, it would require at least as much involvement as a 2-FTE tournament.
|
|
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: 2008 World Championship Format
« Reply #36 on: Oct 5th, 2007, 2:17pm » |
Quote Modify
|
on Oct 5th, 2007, 12:21pm, mistre wrote:| Count me in for a 5-round swiss qualifying tournament as long as the time control is at 60 sec/move or less. |
|
Awesome! And would you also play for the World Championship at 90 seconds per move if you qualified in the top eight? If not, the ninth-place qualifier might be happy to take your place, so there is no problem. I'm just curious if there are a lot of folks who would want to play in a swiss for fun, but wouldn't want a seat in the final even if they earned one.
|
|
IP Logged |
|
|
|
RonWeasley
Forum Guru
Harry's friend (Arimaa player #441)
Gender:
Posts: 882
|
|
Re: 2008 World Championship Format
« Reply #37 on: Oct 5th, 2007, 2:41pm » |
Quote Modify
|
Just so everyone knows, I have a relative who will be spending much of December in St. Mungo's Hospital. This will leave no time for me to play the WC or even be the TD. I hope to watch some of the games from the Gryffindor section of the stadium.
|
|
IP Logged |
|
|
|
mistre
Forum Guru
Gender:
Posts: 553
|
|
Re: 2008 World Championship Format
« Reply #38 on: Oct 5th, 2007, 2:56pm » |
Quote Modify
|
on Oct 5th, 2007, 1:40pm, chessandgo wrote:
Mistre : I'm not sure I understand your statement. Do you say you'd play in a 5-rounds swiss tourney but not in a 2-lives floating elimination tourney, as was used last year ? It seems to me that, even if the swiss tourney is played at a realatively short time control, it would require at least as much involvement as a 2-FTE tournament. |
|
Actually, it IS the length of 1 game that poses a problem for me and not multiple shorter length games. I do have sufficient time here or there to play Arimaa, just not in long chunks. Which is why I mostly play Postal vs. humans.
So, I am probably not the best candidate for a tournament, but I am interested - so I would try to make it work if I can.
|
| « Last Edit: Oct 5th, 2007, 2:56pm by mistre » |
IP Logged |
|
|
|
jdb
Forum Guru
Arimaa player #214
Gender:
Posts: 682
|
|
Re: 2008 World Championship Format
« Reply #39 on: Oct 5th, 2007, 3:29pm » |
Quote Modify
|
on Oct 5th, 2007, 2:56pm, mistre wrote:
Actually, it IS the length of 1 game that poses a problem for me and not multiple shorter length games. I do have sufficient time here or there to play Arimaa, just not in long chunks. Which is why I mostly play Postal vs. humans.
|
|
This is my situation also.
|
|
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: 2008 World Championship Format
« Reply #40 on: Oct 5th, 2007, 3:43pm » |
Quote Modify
|
on Oct 5th, 2007, 2:41pm, RonWeasley wrote:| Just so everyone knows, I have a relative who will be spending much of December in St. Mungo's Hospital. This will leave no time for me to play the WC or even be the TD. |
|
What if the qualifier is in January/February and the final is in February/March? This morning Omar was letting on that trying to squeeze in the qualifying tournament in November/December sounds a little rushed.
|
|
IP Logged |
|
|
|
chessandgo
Forum Guru
Arimaa player #1889
Gender:
Posts: 1244
|
|
Re: 2008 World Championship Format
« Reply #41 on: Oct 5th, 2007, 4:08pm » |
Quote Modify
|
on Oct 5th, 2007, 1:40pm, chessandgo wrote:
It seems to me that the hardest part, when scheduling a game, is to find a common spot, not really the potential length of the game. It's much more difficult to play two 45 sec/move games than one 90sec/move game, for two common slots with two opponents must be found. So I'd say the actual length of one game is not the problem, when it comes to making an arimma tourney fit in one's timetable ; it's rather the total number of games to be played.
So I don't think playing the premilinary round at a short time control would do much good.
Mistre : I'm not sure I understand your statement. Do you say you'd play in a 5-rounds swiss tourney but not in a 2-lives floating elimination tourney, as was used last year ? It seems to me that, even if the swiss tourney is played at a realatively short time control, it would require at least as much involvement as a 2-FTE tournament. |
|
Well, apparently everyone disagrees with my statement, so I withdraw this. Mistre, I'm glad that you will play (apparently the new formula will be prefered to the old one ?), same goes for jdb.
The preliminary starting only on january would be fine with me as well as in november ; if it allows a looser timetable, all the better.
|
|
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: 2008 World Championship Format
« Reply #42 on: Oct 5th, 2007, 5:16pm » |
Quote Modify
|
Actually, I agree with you Jean: I prefer a 90-second time control, and the difficulty is for me is not finding a long enough block of time but rather a mutually agreeable block of time. I could resign myself to a 60-second time control for the qualifier if the finals were still 90 seconds per move, but even at that level most games will be decided by tactical blunders, so I would rather it not slip further, in order that the games be of a higher quality.
Random thought: [moved to new thread]
|
| « Last Edit: Oct 5th, 2007, 7:45pm by Fritzlein » |
IP Logged |
|
|
|
Janzert
Forum Guru
Arimaa player #247
Gender:
Posts: 1016
|
|
Re: 2008 World Championship Format
« Reply #43 on: Oct 7th, 2007, 1:53pm » |
Quote Modify
|
I wish the last games of the WC were still played at 2 minutes per move as they used to be. But it's probably too many games to make the whole final tournament 2 minutes and I don't see where after that it would be appropriate to bump up the time.
Janzert
|
|
IP Logged |
|
|
|
chessandgo
Forum Guru
Arimaa player #1889
Gender:
Posts: 1244
|
|
Re: 2008 World Championship Format
« Reply #44 on: Oct 8th, 2007, 7:35am » |
Quote Modify
|
I feel that the games should be played with time control as long as possible for practical purpose too, to make game quality as high as possible, especially for the last games.
In my exeprience from last year, 90 sec is quite a good compromise, with games ending within a reasonnable of time (around two hours iirc), and yielding higher game quality than 45 sec a move, so I'd be for the last stage games to be played at at least 90 sec per move. 2 min seems quite good too, or maybe just for the last games (like when only 2 people remain, or 3, or 4 ?) ?
How about the preliminary stage ? I saw 45 sec and 1 mn mentionned, what is going to be chosen ? I hope we won't have a too short time control ; 45 sec is the very standard time control for training games H vs H throughout the year ; it would feel strange not to have longer thinking time for a WC game, even a preliminary round. Would 1 mn per move be ok for everyone ?
|
|
IP Logged |
|
|
|
|
Home
Help
Search
Members
Login
Register
Reply
Notify of replies
Send Topic
Print