Author |
Topic: 2009 World Championship (Read 9048 times) |
|
naveed
Forum Full Member
Arimaa player #110
Gender: 
Posts: 11
|
 |
Re: 2009 World Championship
« Reply #90 on: Feb 4th, 2009, 5:11pm » |
 Quote  Modify
|
Since this tournament is so close I wouldn't mind playing another round (tiebreaker) so that the finalist may be decided by the performance and not so much by SOS
Naveed
|
|
 IP Logged |
|
|
|
Tuks
Forum Guru
Arimaa player #2626
Gender:
Posts: 203
|
|
Re: 2009 World Championship
« Reply #91 on: Feb 4th, 2009, 8:04pm » |
Quote Modify
|
i would agree to that
|
|
IP Logged |
|
|
|
99of9
Forum Guru
Gnobby's creator (player #314)
 
Gender:
Posts: 1413
|
|
Re: 2009 World Championship
« Reply #92 on: Feb 4th, 2009, 8:27pm » |
Quote Modify
|
It doesn't make sense to change the tournament format at this stage. It's not even clear to me that anything has gone wrong, or that one more round would fix the problem. It's almost guaranteed that another round would lead to plenty of players tied on 3/6, and the line would need to be drawn through them.
If this is about woh's opportunity to place in the top 8, it seems clear to me that his results to date do not warrant that place. In particular, he lost to omar, who is himself only on the borderline between making and not making the 8. Jeh will also be out if he loses to omar. The fact that his entire tournament result was basically decided by that one game is disappointing, but others have also had make or break games.
|
| « Last Edit: Feb 4th, 2009, 8:33pm by 99of9 » |
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: 2009 World Championship
« Reply #93 on: Feb 5th, 2009, 7:55am » |
Quote Modify
|
on Feb 4th, 2009, 5:11pm, naveed wrote:| Since this tournament is so close I wouldn't mind playing another round (tiebreaker) so that the finalist may be decided by the performance and not so much by SOS |
|
It's generous of you to offer a playoff game, since it appears that woh is the only player who will have three wins and be eliminated on tiebreakers, whereas you may be the player who makes it 8th place on tiebreakers and could be eliminated by a playoff.
But are you talking about everybody playing another round, or just the #8 and #9 players playing another round? If it is just the #8 and #9, then it will be tiebreaker points determining which of the 3-2 players have to play that extra game, while #6 and #7 who are also 3-2 don't have to risk anything because they have better tiebreak points. This seems to defeat the purpose of trying to rely on W-L only while taking SoS out of it.
As 99of9 pointed out, if instead everybody has to play another round, then some players who have lost three times will be eliminated while other players who have lost three times will qualify on tiebreakers, again defeating the purpose.
If we want a perfectly fair way to discriminate 8th place from 9th place, we are almost forced to play a round-robin before the finals have even begun. The way we structure the preliminaries now is definitely a nod to practicality. Yes, there is a fair bit of luck in the current format, enough to move people around several seeds from their objective strength. At the same time, however, everyone is control of his own destiny and everyone has a reasonably fair shot of making the finals.
|
|
IP Logged |
|
|
|
RonWeasley
Forum Guru
Harry's friend (Arimaa player #441)
Gender:
Posts: 882
|
|
Re: 2009 World Championship
« Reply #94 on: Feb 5th, 2009, 10:13am » |
Quote Modify
|
Just a reminder that this format change discussion really applies to future WC tournaments. We're committed to the current rules unless something fundamentally unfair is discovered.
This year it seems we've simply discovered that seeds 7-10 have to play well and get some luck to make the finals. We knew that before we started.
|
|
IP Logged |
|
|
|
naveed
Forum Full Member
Arimaa player #110
Gender:
Posts: 11
|
|
Re: 2009 World Championship
« Reply #95 on: Feb 5th, 2009, 4:03pm » |
Quote Modify
|
I meant only #8 and #9 players playing another round.
Naveed
|
|
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: 2009 World Championship
« Reply #96 on: Feb 8th, 2009, 2:31pm » |
Quote Modify
|
on Feb 1st, 2009, 2:25pm, Fritzlein wrote:| During The_Jeh vs. arimaa_master, I counted a peak of 22 people logged in and 11 people in chat. The Arimaa community is thriving! |
|
The_Jeh sure knows how to draw a crowd. During his game with Omar (the last game of the preliminaries) there were 24 people in the game room and 16 people in chat.
Critical mass, here we come!
|
|
IP Logged |
|
|
|
woh
Forum Guru
Arimaa player #2128
Gender:
Posts: 254
|
|
Re: 2009 World Championship
« Reply #97 on: Feb 9th, 2009, 4:18am » |
Quote Modify
|
Congratulations to the 8 qualifying players!
Best of luck in the finals.
|
|
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: 2009 World Championship
« Reply #98 on: Feb 9th, 2009, 10:01am » |
Quote Modify
|
Congrats especially to the newcomers in the finals: The_Jeh, Tuks, and camelback!
It's nice that three of the pairings in the first round of the finals didn't happen in the preliminaries. It's just luck that it worked out that way, but since it was camelback vs. Fritzlein and Tuks vs. chessandgo in the preliminaries, I am glad that it is Tuks vs. Fritzlein and camelback vs. chessandgo in the finals. Also Adanac vs. 99of9 should happen at least once per World Championship.
|
| « Last Edit: Feb 9th, 2009, 1:30pm by Fritzlein » |
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: 2009 World Championship
« Reply #99 on: Feb 9th, 2009, 11:01am » |
Quote Modify
|
on Jan 27th, 2009, 9:15am, chessandgo wrote:Maybe for next year there is an issue of players floating down. I seem to recall a WC (floating double elim) where the top seed kept floating down, and in this swiss tourney, camelback kept floating down just as well. Wouldn't it be "more fair" if a player who has already floated down would be defavorized to float down in the next rounds with respect to those who haven't? This is admitedly a very minor issue  |
|
Camelback certainly benefited from "floating down" in Round 2 and Round 3 this year, because he got to play a weaker schedule than omar, who was just one seed higher. On the other hand, camelback did get wins against Tuks and naveed. If things had worked out just slightly differently, camelback might have ended up with three wins but in ninth place on tiebreakers instead of eighth, and might have felt mistreated by the pairing system which kept denying him a chance to prove himself. Note that in Round 5 it was woh who floated down, and this cost woh a chance at a vital tiebreaker point. Note also that when camelback floated down in Round 4 it didn't get him a win but did cost him a tiebreak point.
I think this is basically an issue with how much luck there is in the order of finish in the middle of the pack of a five-round tournament. This year camelback was lucky while omar was unlucky, but next year the person who floats down may feel unlucky. What if camelback had lost to Tuks in Round 2? And even this year woh has a reason to feel unlucky for floating down.
Similarly floating up can cut both ways. Tuks can feel unlucky that he floated up against camelback in Round 2, but 99of9 probably felt lucky to float up against camelback in Round 4, because all the 2-1 players were higher seeds than 3-0 camelback was. If there had been an upset (or forfeit/disconnection) in Round 1 like there was in Round 3, then floating up in Round 2 would have been lucky for Tuks too.
Ordering #8 and #9 within a five-round tournament is a hard problem when we can't rely on game-room ratings. My own feeling is that the luck introduced by having a short tournament is truly a problem, but that problem is offset by having the outcome depend almost entirely on in-tournament results, and very little on game room ratings.
Yes, naveed was seeded #7 compared to camelback at #9 and Tuks at #10. But he lost to both of them. In that sense the tournament did a fantastic job of deciding who should be above the line and who should be below it. If you look at the game histories, naveed's rating comes almost entirely from BombBlitz and BombLightning. I would rather not use that source of information to decide whether or not he is on the bubble. Meanwhile woh was seeded #11, but he had a chance comparable to the chance Tuks and camelback had; if woh had beaten omar he would have gotten stronger pairings the last two rounds, with a better chance to prove himself.
Maybe some day we will have some form of reliable ratings, and we can use those ratings to directly seed the finals, doing away with the preliminary. Until that day, however, there is an egalitarian element to the open Swiss that should not be discarded lightly. Certainly not everyone had an equal path to the finals, but everyone had at least a reasonable path.
|
| « Last Edit: Feb 9th, 2009, 1:33pm by Fritzlein » |
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: 2009 World Championship
« Reply #100 on: Feb 9th, 2009, 1:29pm » |
Quote Modify
|
I have thought more about the strength of schedule tiebreaker problem. I believe the root of evil is that differences in winning percentages don't correlate to difference in in SoS points. It is just like Kenneth Massey described.
Let's consider six players A through F, who I will give imaginary ratings and tournament results:
A, 2500, 5-0
B, 2260, 4-1
C, 2020, 3-2
D, 1780, 2-3
E, 1540, 1-4
F, 1300, 0-5
Let's use these players as a benchmark for strength of schedule. If I play against A & F, I get 5 total tiebreak points, and if I play against C & D, I also get five total tiebreak points. So it should be the same to me which pair I play against, right? Wrong.
If my own rating is 2500, I have a 50% chance to beat player A and a 99.9% chance to beat player F for a total of 1.499 expexted wins. Meanwhile I have a 94.1% chance to beat player C and a 98.4% chance to beat player D, for a total of 1.925 expected wins. For me, playing C&D is a much easier schedule than playing A&F.
By the same token a player rated 1300 could expect 0.501 wins against A&F, while they could only hope for 0.075 wins against C&D, so for that player A&F is much easier schedule for them than C&D. That player has exactly the opposite preference that I do! Neither of us think that five SoS points is equal to five SoS points in the current system.
The solution is to have the strength of schedule depend on the number of wins of the player for whom you are calculating schedule strength. This is fair because SoS is only a tiebreaker between players with equal wins. Players who are 3-2 will use the same formula as each other, even though it is a different formula than the one the 4-1 players are using.
Let R be the number of rounds played and W be the wins of a player under consideration, and O be the number of wins of the opponent whose strength we are trying to calculate. Then calculate the strength S of that opponent incorporating the Elo formula as:
S = 1/(1+10^((W-O)*3/R))
Let me apply this formula to the tiebreakers for 3-2 players in the current tournament. The actual tiebreakers were
99of9: 15 = 1 + 2 + 3 + 4 + 5
Adanac: 13 = 0 + 2 + 2 + 4 + 5
The_Jeh: 13 = 0 + 2 + 2 + 4 + 5
Tuks: 13 = 1 + 2 + 2 + 3 + 5
camelback: 12 = 0 + 2 + 3 + 3 + 4
woh: 11 = 1 + 2 + 2 + 2 + 4
By my proposed conversion formula, for a 3-2 player
5 -> 0.94
4 -> 0.8
3 -> 0.5
2 -> 0.2
1 -> 0.06
0 -> 0.02
So the SoS converts to
99of9: 2.5
Adanac: 2.16
The_Jeh: 2.16
Tuks: 1.9
camelback: 2.02
woh: 1.46
I think my formula is generally pretty fair. In the specific case of our tournament it has a nice side benefit of mitigating a first-round win over a player that withdraws. Adanac, The_Jeh, and camelback were the three players who took a hit to strength of schedule when their first-round opponent withdrew. Tuks, on the other hand, got helped by playing chessandgo, which is a "sure loss" for any 3-2 player, but helped his SoS disproportionately. So it makes sense to me that Tuks falls out of a tie with Adanac and The_Jeh for SoS, and actually falls behind camelback.
In my formula playing someone who withdraws is still punished relative to playing someone who goes on to win one game, but the difference in SoS for someone who finishes 3-2 is negligible. Indeed, for someone who finishes 3-2, playing one 3-2 player plus one 0-5 player is a tougher SoS than playing two 2-3 players.
What do people think? Is this a reasonable way to keep pre-tournament ratings out of the picture, while still relying on strength of schedule for the tiebreaker? Does it mitigate the first-round-withdrawal unfairness?
|
| « Last Edit: Feb 10th, 2009, 10:16am by Fritzlein » |
IP Logged |
|
|
|
chessandgo
Forum Guru
Arimaa player #1889
Gender:
Posts: 1244
|
|
Re: 2009 World Championship
« Reply #101 on: Feb 10th, 2009, 1:06am » |
Quote Modify
|
on Feb 9th, 2009, 11:01am, Fritzlein wrote:
Camelback certainly benefited from "floating down" in Round 2 and Round 3 this year, because he got to play a weaker schedule than omar, who was just one seed higher. On the other hand, camelback did get wins against Tuks and naveed. If things had worked out just slightly differently, camelback might have ended up with three wins but in ninth place on tiebreakers instead of eighth, and might have felt mistreated by the pairing system which kept denying him a chance to prove himself. Note that in Round 5 it was woh who floated down, and this cost woh a chance at a vital tiebreaker point. Note also that when camelback floated down in Round 4 it didn't get him a win but did cost him a tiebreak point.
|
|
Not saying whether floating down is good or bad, just pointing out that we might slightly twist the rules to make it so that the same player would have an incentive not to float down twice. Well, or rather that if we can do so, it should make it even "more fair". But no big deal.
|
|
IP Logged |
|
|
|
aaaa
Forum Guru
Arimaa player #958
Posts: 768
|
|
Re: 2009 World Championship
« Reply #102 on: Feb 10th, 2009, 4:02pm » |
Quote Modify
|
Would it be too much of a distortion if the seeds were to be variable in the finals, being based on all the games played so far in both tournaments?
|
|
IP Logged |
|
|
|
Fritzlein
Forum Guru
Arimaa player #706
Gender:
Posts: 5928
|
|
Re: 2009 World Championship
« Reply #103 on: Feb 10th, 2009, 8:05pm » |
Quote Modify
|
on Feb 10th, 2009, 4:02pm, aaaa wrote:| Would it be too much of a distortion if the seeds were to be variable in the finals, being based on all the games played so far in both tournaments? |
|
Can you explain how that would apply?
|
|
IP Logged |
|
|
|
aaaa
Forum Guru
Arimaa player #958
Posts: 768
|
|
Re: 2009 World Championship
« Reply #104 on: Feb 10th, 2009, 8:09pm » |
Quote Modify
|
Simply by continuing to update the standings of the Swiss tournament.
|
|
IP Logged |
|
|
|
|
Home
Help
Search
Members
Login
Register
Reply
Notify of replies
Send Topic
Print