| |
BGonline.org Forums
Alternative to byes
Posted By: Henrik Bukkjaer In Response To: Alternative to byes (Bob Koca)
Date: Wednesday, 4 May 2011, at 11:23 p.m.
This post is just stuck into the thread at this point addressing many of the other posts :-)
---
1) Auctioning byes etc. works fine in "non-title" or "non-masterpoint" events, such as one-off tournaments with focus on fun and/or money. (could be a jackpot sideevent, etc.). It does not work for "sporting" events where you try to find the best player of the lot.
2) Looking at Swiss style events (at least to just thin the field prior to a cup format finale) makes sense. But as shown in this thread, there are a number of challenges with Swiss, mainly tiebreaking.
Throughout 20+ years, DBgF gained a lot of experience in running variations of the Swiss tournament format, mostly the "Monrad" one, which is somewhat close to what Bob's suggesting.
I'll just add some comments/suggestions here:
Tom's idea of calculating performance ratings is fine but you should not "adjust" anything with the number calculated "on the way", rather than using the rating calculated at the end. Otherwise the draw get influence on the result since the strong players will not be recognized until they actually win a few matches more than they loose and thus are not that fun to meet in the first couple of rounds. Contrary to the week players, whom you'd prefer to meet early in the tournament.
If instead you "adjust" (by adjust I mean do the calculation for deciding tiebreaks) with the end result of your opponents, you get a very usable result. Tom suggested to include only wins in the calculation that is the wrong idea. You want to do the tiebreak like this: Player A and B won the same number of matches, the one who did so against the toughest opponents performed best! And then it doesn't matter if you won or lost to them it's your entire population of opponents who has to be considered. Byes in the Swiss should always be given to the players with the worst records and never two byes to the same player. In a tiebreak situation, a bye should not count for anything, is actually just half a point! I'd go as far as stating that in a tiebreak between a player with a bye and one without, the latter should simply win... But of course it depends on the number of rounds, etc.
Now, using opponents records to adjust with, is bad in some ways:
Some players leave the tournament before the last round (say they have 4 losses after 4 rounds in a 6 round Swiss). This will affect the results through adjustment calculation - it's not great to see your 1st round opponents leave the tournament 3 rounds later, messing up your adjustment... Other players with bad records might stay and play but not to the best of their ability, and maybe even while drinking heavily.
Now, in Monrad tiebreaking you remove the worst opponents from the calculation eg. the worst two. If you end up still a tie, you remove one more, etc. That works great in chess, but it doesn't make as much sense in backgammon. In a distributed chess field, the difference between best and worst is very significant compared to backgammon. In chess it really doesn't matter if you play someone rated say 300 points below you or 400 points below you - you will simply win. Why it makes sense to remove these "bad" players from the adjustment calculation between two top players in a field, looking at the better of their opponents. You have no "sure" points in backgammon so why remove the worst opponents from the calculation? Well if not for doing it for some theoretical reason, then at least you get rid of the messed up numbers from players leaving the tournament prematurely!
Now, we've discussed how to avoid that people leave prematurely. We've considered having players pay a deposit of some kind, to be payed back at the end of the tournament. But that's not the right idea! So we've typically gone with other incentives such as "Sprint prices" for best performances in the last 3 rounds, etc. It doesn't solve the problem it minimizes it.
If you use tournament result of opponents to calculate adjustments (tiebreaks) you get another effect: A cup-like effect due to the pairings method used in Swiss. Early round losers will find it very difficult to get a good adjustment, since they meet players that lost as well, and more often will end up in the button half of the results table. It's somewhat fine, because these players (all other things being equal) will also be the not so strong ones ;-) But it poses one problem: Adjourned games (that end up like wins). If you have one or two adjourned games and subsequently win them, your draws will be affected negatively, and your chances of getting a good adjustment are lowered. TDs should give half points for adjourned games in order to minimize this problem. In fact TDs should enforce clock play in all Swiss formats, to eliminate the adjourned games entirely.
So we looked at rating adjustment instead since we have a very good rating system and since all our players are rated. This was used with success in some events recently! One great thing about using ratings instead of tournament result, is that adjustments is not affected by players leaving the tournament prematurely! And they can be calculated and displayed throughout the tournament (because the adjustment value of each opponent is know at the beginning of the event) and thus be used as input to the tournament match pairings ;-)
Anyway, forcing top players to play games with no chance of winning anything is not the way to go. Imagine a cup tournament where you ask loosing players to play each other "just for the fun of it". How many would sit and play?
Swiss is also time consuming since you have to wait for all matches in a round to finish, before you can calculate the next round pairings. It's a little time-inefficient (and we are dealing with a game where luck can be minimized more or less only by plying a matchformat that uses as much time as possible.
But it's GREAT for beginner and novice events. And it's GREAT for doubles events (team of two players playing 2 individual games, not the consulting version). It's also great for "thinning" a field on a multiple day event.
But as alternatives we've been looking at other formats. Including but not limited to, multiple-elimination events. Swiss style formats where you're out after X losses, etc.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.