The Standings Listings |
|
On the Web |
In the Newsletter |
| Ties are listed in "tie-break" order; i.e., the winner of the tie is listed first. |
|
| Where a tie cannot be broken, the total score is listed in red. | I don't have a mechanism yet to show unbroken ties. |
| To be invited to The Championships, you must place in the top three and have at least two successful (non-zero scoring) pulls. |
|
| Qualifying dogs in the top three positions have their total score back grounded by yellow. | I don't have a mechanism yet to show qualifying invitees. |
| To qualify for a medal, you must place in the top three and have at least one successful (non-zero scoring) pull. |
|
| Qualifying dogs in the top three positions have the dog's name back grounded by the closest color to the medal: | I don't have a mechanism yet to distinguish medallists. |
| Gold for 1st (alright, it's yellow) | |
| Silver for 2nd | |
| Bronze for 3rd (really more of a brown) | |
|
The tie-break programming and listing
indicators are being hurried in at the last minute and are subject to errors at this time. The tie-break line-up has the same accuracy for both the newsletter and the web listing (I believe it is correct - except where 3 or more dogs tie). The markings for medallist and Championship invitation should be very close on the web listing. The final Medallist list and Championship invitees will be hand checked. |
|
On my 1st year as statistician, I turned over the final standings and someone else made up the Medallist and Championship Invitee lists. On my 2nd year I was tasked with the lists. I asked how ties were resolved and was told: 1st - compare the dogs where they competed against each other and the one winning the most would be the winner; 2nd - if still tied, compare their scores and the highest low wins. At least, that is what I think I was told back in 1996. However, I didn't quite pick-up on the highest-low; but instead implemented what might be termed the highest-high. What I did that year was presented to The Board and they approved it and wrote it into the By-Laws.
The following year, I picked-up on my misinterpretation and did some comparisons of the two methods for that year's finalists. Neither method stood out as superior. I saw cases where each method picked what I felt was the less deserving dog. I presented these to The Board that year, but it did not generate enough concern for any action or study to be taken.
The method has been questioned again this year. In my opinion, neither method is very good. I plan to do some analysis and present the findings to The Board again this year to see if they want to re-evaluate the method.
Meanwhile, below are the details of how I am resolving (or failing to resolve) ties. I have done this by hand the last few years. This year, I have programmed it into the database at the last minute. I believe the programming is correct where 2 dogs tie; but has flaws where 3 or more tie. Regardless, the top three places will be hand checked before the Medallist and Invitee list are made final. BUT, the standing published on The Web and in The Newsletter could have errors.
Head - to - Head |
| Each pull in which the tied dogs competed against each other (same pull,
same class) is singled out. A tally of how many times each dog beat the other is
recorded. The dog winning the most number of times is declared the winner. If
it is still a tie, or they did not compete against each other, we move on to the second
part of the algorithm. Note: I am counting "No Pulls" (zero score) in this comparison. In the case of three or more dogs tied, I am looking only at pulls where they all competed together. If a winner or loser pops out, the remaining dogs are then compared. With three of more dogs, we could drop into the second part of the algorithm to get a winner or loser; then come back to this part for the remaining dogs. I am not doing that properly at this time. |
| Remarks: I tend to agree with this step of the algorithm. Of course, you can always find exceptions; but this step is the most clear-cut of the entire method. |
Highest - High |
| If "Head-to-Head" fails to resolve the tie, I then compare
scores. I exclude pulls that were already used in the
"Head-to-Head" comparison, and "No-Pulls" (zero scores). The scores are listed in descending order and I march down the list until the scores differ. The one with the highest score wins (or lowest score loses if more than 2 dogs). This could then iterate back with the "Head-to-Head" if we were working with more than two dogs (but I don't have that working correctly). If the tie is still not resolved, we continue on the the "Any More" step. |
| Remarks: If I were to implement
"Highest-Low", I would do the same as above, except the scores would be listed
in ascending order. One could argue that the "Head-to-Head" pulls should not be excluded. In many cases, it would resolve a deadlock. My choice would be to do some statistics on the scores (or number of dogs beat) rather than to compare scores. I would tend to look at averages and deviations from the average. However, there are many cases where there is insufficient data to make any meaningful statistical conclusion. |
Any More |
| If the above two steps do not resolve the tie, I look to see if a dog has additional non-zero pulls that have not been used in the comparison. If so, he is declared the winner. Otherwise, it is called a draw. |
| Remarks: Of course, one can argue this is unfair. |
In viewing the Standings, your highest 5 pulls are listed. If all pulls for the tied dogs are listed, you can verify following the steps above. However, if there are more than two Out-of-Region, or exceed the 5 total, you may not see all the pulls used in evaluating the Tie-Break.
For some examples, see 9899 ties and 9798 ties and 9697 ties.
Ron Bowser