Board
Teamleague Forum
posted at 2013-09-23 23:33 by KiranY
"the fewer forfeits rule is not fair. we already got punished in 2 rounds for our unreliable player. still we managed to get same score, means we are actually better. so why get punished again then?"
I presume this rule was introduced to highlight the importance of having lesser forfeits or having reliable players which might be right in a way. But for all practical purposes wouldn't the team/captain already have learnt their lesson the hard way by losing out on valuable gp/mp? and despite the odds reached the same score, would be hoping for a playoff spot?
If the forfeits had occurred just before the last round or in the last round, then it can be argued that the team might not deserve a place in playoff, due to the reason that they might forfeit one of their boards in the playoff round too. Then it would not be a fair/competitive round to decide the champion. But what if the team having forfeits in the middle rounds had learnt their mistake and replaced with a more reliable player. Then they would only look forward to the playoffs right?
On another note, the next tie breaker is removing the lower boards one by one and calculating gp and mp etc. Is this method fixed or does it change from tourney to tourney. I thought we had a system, where we remove board results from the top and bottom in alternate tourneys. This would give equal importance to the lower boards too. Or even better is to remove b4 and then b1 and b3 etc.
All these tiebreaks occur very rarely, but when it does the rules should be such that they give equal importance to all the factors.
posted at 2013-09-24 06:58 by smallblackcat
There was a rule about swapping between removing boards from the top and bottom in alternating tournaments, which was a holdover from the days when TL was played across both FICS and the ICC. This was removed some years ago, because it seemed needlessly complicated to have the rule change from tournament to tournament.
If there are suggestions for a fairer way of handling such situations, we are open to hearing them.
posted at 2013-09-24 09:21 by anandkvs
For alternate system, why not use a system used in round robin chess tournaments, like the Candidates 2013?
Since wins with black etc. are not relevant in a team competition, the relevant tiebreaker in our case would be either higher match wins (1 match win is better than 2 match draws) or the Sonneborn-Berger.
These scores also have the advantage of being mechanically easily computable and can be displayed in the standings table.
posted at 2013-09-24 09:27 by anandkvs
posted at 2013-09-24 10:49 by smallblackcat
Using a sum of scores-based tiebreaker makes sense, though there is the question of whether they are based on MPs or GPs. If it's based on MPs, this has the problem that the difference between 2-2 rounds and 2.5-1.5 rounds would be elevated even further than they already are. Using GPs would seem contradictory to the overall scoring system.
I would suggest using a sum of progressive MPs. It's simple enough to work out, and it counter-balances the possible advantage of playing an eliminated and disheartened team later in the tourney. Most importantly, it's very unlikely to result in another tie!
posted at 2013-09-24 11:39 by kirany
1. Remove b1 and recompute MP and GP of all rounds (giving importance to lower boards first). And ofcourse match with MP first and then GP.
If its still a tie, remove b4 and calculate again. And finally remove b2. In this method instead of alternating between every tourney we have a fixed system to alternate between the boards.
And yeah I might be arguing for a change again if I am at the receiving end of this :P.
2. Calculate the black board's score in all the rounds. This gives importance to black's results since there is a general tendency to be satisfied with a draw as a black. The downside is, this has more chances of not resolving the tie than the 1st case. But atleast we are honoring black's play. Also if a forfeiting player played black in a round it would be counted again against the team. So that need not be taken into account otherwise it would defeat the purpose. We are counting only played game results.
posted at 2013-09-24 23:30 by PankracyRozumek
This is unlikely to lead to further ties. It is easy to calculate automatically and display in the table if needed.
Sum of progressive scores, if I understand correctly, promotes the team which scored higher earlier in the tournament, right? So a team with scores 1-1-0 in three rounds would rank higher than team scoring 0-1-1? If this is the case, it might discourage some players from fighting till the end.
posted at 2013-09-25 04:37 by smallblackcat
I don't understand your concern about using progressive scores. Since this would only come into play when everything else (MPs, GPs, head-to-head) are already tied, it is not a disincentive to keep fighting till the end of the tourney, because everything else would still be in play (GPs come down to the very last game, as it is).
posted at 2013-09-25 06:02 by anandkvs
So Sonneborn-Berger based on Match Points seems the logical choice (I had meant this earlier, perhaps I was not clear enough).
posted at 2013-09-25 06:39 by anandkvs
a) The second tie-breaker is Game-Points anyway, which works in the opposite direction (against narrow wins). To elaborate. Teams A and B are tied on Match Points with team A winning one (by 2.5-1.5) and losing one and team B drawing both by 2-2. Then the only way they can be tied on game points is if they have lost as narrowly as they have won (1.5-2.5)
b) 2.5-1.5 is a win and 2-2 is a draw, which is in the spirit of the team competition anyway. So the only thing which matters by analogy with single player competition is 1 point for a win and 0.5 for a draw.
posted at 2013-09-25 09:27 by KRMCHESS
As far as a suggestion for an alternative goes I would propose what I call sportsmanship coefficient. Basically what it would mean is that any actions that make running TL harder get a negative while anything that makes it easier gets a positive. Examples could be as follows:
Receiving warning message about first contact in 1 day = -0.1
Missing first contact deadline = -0.3
Forfeit = -1
Captain choosing unavailable player = -0.5 (on top of forfeit if applicable)
Missing game and rescheduling = -0.5 for person who missed game, +0.5 for person who agreed not to insist on forfeit and play match later
Completing game with no issues (i.e. no negatives) = +0.2
There could be a few others and potential TD could award points if people have been particularly helpful or dock some if they've been quite unhelpful or their behaviour has been inappropriate. It's more complicated than forfeits although it should be possible to add it as a variable in standings so a quick easy comparison would be available. Of course it's merely a concept and figures are just ball park figures but such a system could theoretically give an incentive for people to schedule matches promptly and efficiently in such a fashion that it makes TD's jobs easier
posted at 2013-09-25 10:58 by smallblackcat
I suspect that Sonnenborn-Berger has less value in a team competition, since the strength of the opposition is not necessary constant (i.e. sometimes teams field a weaker lineup).
I suspect that the 'sportsmanship coefficient' will appeal to some among the TDs. I'm afraid I have a more cynical viewpoint, which is that I don't like the extra effort it would take to calculate in what is likely to be a limited time between the end of play and the setting of playoff pairings.
I'll also note that the 'fewest forfeits' tiebreaker didn't come into effect this time. Not that this makes the debate any less valid.
posted at 2013-09-25 12:59 by joshuar
posted at 2013-09-25 13:03 by KRMCHESS
Of course cynical viewpoint also works, especially as someone will have to put in a fair amount of work for a new feature that may potentially be used very rarely as fewest forfeit tiebreaker is very rarely used and it's perfectly understandable if it's considered a much better use of coding time to improve things used regularly rather than once in a blue moon
posted at 2013-09-25 15:41 by KiranY
A player loses a RR for not being reliable and now he gets it back by being more accommodating or sportive. This is separate from the fact that it increases if you finish a season without any forfeit. This would also be a form of encouragement to be more sportive and it will affect an individual rather than a team.
Resolving tiebreaks should be done purely by playing with gps and mps accrued so far either by applying the Sonneborn-Berger method or other methods mentioned above, since these points were gained by playing actual games.
posted at 2013-09-26 06:16 by PankracyRozumek
If you think Sonnenborn-Berger is not good because of varying team strengths what about: average of all opponent's TL ratings? Considering different line-ups in subsequent rounds, it is quite likely to be a decisive factor. It is easy to calculate and display automatically. Finally it is easy to understand.
Shortly: I'd like anything but sum of progression scores :). I prefer even a coin flip :).
posted at 2013-09-26 09:58 by smallblackcat
Sonnenborn-Berger has the advantage of simplicity and clarity, and of being used in comparable events elsewhere. Those are good enough reasons for me to sign off on it, though I should point out that I am only speaking as an individual here; I am not representing the views of TL staff as a whole.
posted at 2013-09-28 05:40 by anandkvs
posted at 2013-09-28 11:29 by Twikki
posted at 2013-09-29 01:17 by KiranY
posted at 2013-09-29 06:53 by smallblackcat
posted at 2013-10-02 10:34 by Twikki
1) The Kashdan and Median-Buchholz tiebreaks alter scores for unplayed games;
2) Disadvantaging a team in a later tiebreak round for an unplayed game after they benefited from the unplayed game in at least one earlier tiebreak round doesn't seem that unfair;
3) The teams with more than one or two forfeit losses more than likely won't be in the running by the time this tiebreak is applied;
4) It has a minor benefit of making the round 7 games more meaningful if a potential playoff team is playing against a team that is out of it and no longer cares. For example, if the playoff team goes 2.5 - .5 with 2 unplayed victories in 3 games, then the playoff team needs to play and win the last game to protect their MP. Against a team that doesn't care, they will probably win but it's not a foregone conclusion. I think it's more fair to the other potential playoff teams which had to play the bottom dweller when they were more competitive. And...
5) It's really simple to calculate, and it's something that could be done as the event progressed; unlike systems which require all the results to be in before calculations can be made.
I do see the negatives (and one very specific scenario where it would be possible to game the system that I won't mention here.) But there are going to be tradeoffs with any system. By design tiebreaks turn numerical data into a subjective assessment - after the easy ones like MP & GP (a case could even be made for reversing the order of those two- is 5.0/12.5 a stronger result than 4.0/19), at some level whatever you use is a value judgement.
posted at 2013-12-03 20:59 by smallblackcat
There will be a TD meeting prior to T57, at which this issue will be dealt with. Therefore, if anyone wants to do some more lobbying, now would be a good time!
posted at 2013-12-04 10:30 by LeifPetersen
posted at 2013-12-04 11:27 by KRMCHESS
Personally I don't really mind what is chosen although my preference is for something that is easy to tell at a glance from standings page
posted at 2013-12-06 01:39 by KiranY