| |
BGonline.org Forums
What is the best (fairest and most efficient) way to tie-break 4 player round robins?
Posted By: sebalotek In Response To: What is the best (fairest and most efficient) way to tie-break 4 player round robins? (Rick Janowski)
Date: Friday, 30 September 2016, at 7:21 a.m.
RJ >> your mind was already 95% made up and you were looking for people to agree with you.
I now feel obliged to state that the second half of the above is simply not true and your estimation of 95% is quite an exaggeration.
I did however, clearly put a caveat in my very first post indicating that unless there was a clear overall benefit or reason for changing, we would be reluctant to do so.
As far as this thread is concerned, and after considering each alternative suggestion, I could not discern any broadly superior system with a clear overall benefit (apart from playing additional matches to tie-break). Having said that, Neil Robins' suggestion of matches in the double round robin having different lengths sounded interesting. It sounds slightly more complex to track and manage though.
>> Assuming all the players have established Elo Ratings, a reasonable tie-break might be to consider the average rating of opponents played
Rating based tie-breaks do not give as much weight to immediate performance: on the days of the tournament, an Expert player may play like an Intermediate and an Intermediate may play like an Expert. IMO the person who plays better at the tournament itself, should be the one who progresses.
In addition, as well as the referenced Elo-based system currently underestimating some player's abilities, it can also over-estimate them. For example, I (a mediocre player) am currently ranked higher than 10 of the UK's 13 Backgammon Masters and higher than one Grandmaster in terms of Elo. N.B. After approx. 3 years of data, I have approaching 1000 experience points recorded and in some cases the mentioned Masters have between similar to a few thousand.
I am under no illusion as to my abilities and believe that if I were to play a series of matches with these Masters, score difference would be a better reflection of skill level than our current Elo ratings. i.e. I would most likely get beaten with a larger score difference overall, whereas the suggested Elo rating based tie-break would have these Masters eliminated, or placed lower. Therefore, in our situation, I think score difference is currently a much fairer (and simpler) decider than Elo rating.
If we had a more matured system with each player being judged on, say, 10,000 experience points, I would have more support for your proposal, but, of course, this increase in available data will take a number of additional years to come about.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.