|
Joined on 14 Февруари 2001
Total posts: 1 964
|
Rating system
03 Януари 2005 г. 14:47
We are in the final stages of a new world-wide rating system for dancing couples. We will release Beta version pretty soon. If you are interested in testing and influencing development of it - please contact us (i.e. mailto://webmaster@dancesportinfo.net)
|
|
Joined on 14 Февруари 2001
Total posts: 1 964
|
Rating system
11 Януари 2005 г. 10:39
As you can see rating system is online. It is still in beta state and as such can be modified at any time.
Please send us your opinions or post them on the forum.
|
|
Joined on 12 Август 2003
Total posts: 84
|
Rating system
11 Януари 2005 г. 14:42
Glad to see you've done so much work on this!!
Only one thing I noticed immediately: The system appears to be not be correctly primed, or rather it's not equilibrated yet. Many couples scores increase at every competition.
Another "feature" I noticed is that at large competitions, everyone from QF upwards tends to improve their score. It's hard to believe that they have all done better than expected, similarly most everyone in the lower half seems to lose points, again, it's hard to believe they all did worse than expected. Again, I expect that it is either not correctly primed, or the comparison to many inferior dancers is overrated...
I think a possible solution is to run the current set of competitions in the database several times, as you did when we originally discussed this system. It would disperse results more and thus the features you included (such as only considering a certain range of scores as relevant) will prevent some of these effects...
Other than those two points it appears to be producing pretty good results...
Is there a place to check the whole rankings? Sorry if i'm just blind!!!
cheers
robin
|
|
Joined on 14 Февруари 2001
Total posts: 1 964
|
Rating system
11 Януари 2005 г. 15:13
As for couples gaining too much or loosing too much - yes this is a small problem. I hope it can be achieved by tweaking coefficients. Running several times - did not give expected results, especially with some couples who had 'bounce' effect, i.e. they start to bounce from very low rating to very high one. BTW - have a look at http://www.dancesportinfo.net/DisplayResults.aspx?EventID=5607 - you will see that not every body in first half is going up :)
I will try to make some changes to the coefficients tonight.
The bigger problem we have are partial results. If we received only final for example - 6th couple will be penalized for loosing and will loose points (because according to results) they did not win with anybody, while in fact it could have been huge success to get to final.
This is an issue which is not resolved (yet?)
As for list of top couples/events etc, we have everything ready, but we have to package it nicely into desing :). Write to me directly and I will point you to the test page.
|
|
Joined on 13 Януари 2005
Total posts: 468
|
Rating system
13 Януари 2005 г. 12:13
Continuing my comments in public here, since maybe others can add valuable opinion to this discussion.
Admin, I have a feeling that the arrows, simple as they are now, do not neccessary represent the most interesting figure. Currently all they display is how a couple's rating have changed as a result of this event.
What is more interesting, if the couple have finished above or below its expected position based on its rating before the comp.
What I mean is the following: there should be a calculation of "expected result" based on the "before" rating, and this should be compared to the actual result. If a couple has finished higher/lower/as expected, this can be displayed, as well as if the change is significant (thus pointing out "shooting stars" and "unexpected early exiters").
The two indicators may both be displayed.
The reason I think this would be more intersting, is because by the nature of the ELO calculation, especially on competitions with lot of couples, the couples on top are all expected to increase their rating, while people on the bottom are expected to lose points. Even though this is the result of the calculation, still a starter if beginning to beat 1-2 couples, it is an improvement for him, while it will continually decrease its rating.
Let me knwo what You all think.
|
|
Joined on 07 Март 2003
Total posts: 29
|
Rating system
13 Януари 2005 г. 14:50
Just looked in the rating page at top 50/100 English Juniors and at the momment it makes no sence at all. Lots of the couples either went up from junior ages ago or left dance scene years ago. Also it is not a true reflection at all of whom can/does beat whom with very many dancers placed way to high and others placed way too low. Im not sure how you will be able to get a better reflection of true placings but Im sure anyone who knows the Junior scene well at present would agree it is meaningless at present.
|
|
Joined on 14 Февруари 2001
Total posts: 1 964
|
Rating system
13 Януари 2005 г. 16:15
Rating VERY much depends on quality and amount of results which we get as it is calculated from them rather than assigned manually. If you know couples which stopped dancing or started Youth - please let us know and we will change their status. If we have only one or two results of some couple - they will only have very approximated rating - sometimes too high, sometimes too low.
As you may agree top junior couples are more less in right order.
We would appreciate also if you could show us which ratings in your opinion are too high (or too low) and why. It will help us improve our calculation engine.
We have now remove couples with only few results (less than 5)
If we have only partial results (for example only final, or missing placing from semi) calculations are also a bit unfair. For example if we have only final - system does not know that there were another couples and assumes that couple on the 6th place lost with all couples. If your child (or you) qualified (for example) to the sei - please make sure that we receive at least quarter final - your rating will improve!
Try to take results from the competition by copying scrutineers sheets or try to convince organisers to send us results directly.
We cannot generate sensible report from data we have never had :)
|
|
Joined on 12 Август 2003
Total posts: 84
|
Rating system
13 Януари 2005 г. 17:01
quote:
Originally posted by peter
Continuing my comments in public here, since maybe others can add valuable opinion to this discussion.
Admin, I have a feeling that the arrows, simple as they are now, do not neccessary represent the most interesting figure. Currently all they display is how a couple's rating have changed as a result of this event.
What is more interesting, if the couple have finished above or below its expected position based on its rating before the comp.
What I mean is the following: there should be a calculation of "expected result" based on the "before" rating, and this should be compared to the actual result. If a couple has finished higher/lower/as expected, this can be displayed, as well as if the change is significant (thus pointing out "shooting stars" and "unexpected early exiters").
The two indicators may both be displayed.
The reason I think this would be more intersting, is because by the nature of the ELO calculation, especially on competitions with lot of couples, the couples on top are all expected to increase their rating, while people on the bottom are expected to lose points. Even though this is the result of the calculation, still a starter if beginning to beat 1-2 couples, it is an improvement for him, while it will continually decrease its rating.
Let me knwo what You all think.
|
|
Joined on 07 Март 2003
Total posts: 29
|
Rating system
14 Януари 2005 г. 02:00
I have taken another look at Junior Ratings this morning after you have taken couples out with less than 5 results recorded. List does make more sence now and Im sure where couples are placed too high or low, it will rectify itself in the next few results if we all try and send in more results. I would also maybe suggest you think about taking couples out who have not had any results for say over a year as there still seem to be a few listed that I have not seen compete for a very long time.
You also ask for info on who has moved up to Youth etc. Maybe some others on this forum who also know Junior scene could confirm what I have heard as some of these couples definately have moved up, and others I have been told have moved up by others. These are - Jack Beale & Joanne Shearer, David Brown & Lucy Verrechia, Tom Parkes & Kerry Dennehy. Split Partnerships - Phillip Skinner & Nikita Vince.
This couple Im not sure about and maybe someone else could comfirm that they have only just gone up - Karl Drew & Joanne Barnett.
Also I had heard Leon Garner & Rebecca Jackson are no longer together, but again someone else would need to confirm this as Im not sure about it.
If others from forum help to sort out lists in other sections they dance in I think it will be quite interesting once it all gets up and running with more results sent in. ![Smile [:)]](/emoticons/emotion-1.gif)
|
|
Joined on 28 Януари 2003
Total posts: 128
|
Rating system
14 Януари 2005 г. 02:49
I think the problem here for Admin is that unless anyone informs him that a couple have split up or somebody has ceased dancing is that he probably dosn't know about the split until one of these people appear in the results of a competition dancing with a different partner. I am not really sure if or how this problem could be completely overcome. It might help to have an area on the site where dancers could update their details regarding splits, new partnerships etc but I doubt that everyone would bother to do it. Again perhaps the problem of couples moving from one grade to another because of age could be overcome by the system if there was an area where dancers could enter their dates of birth then perhaps the system could automatically pick up when a dancer reaches the maximum age of one group and moves to the next. I could still foresee this causing problems though because other countries have different age groups to England and seniors works the opposite way around in that the youngest half of the partnership must be aged 35 rather than the oldest.
Mo is right in that some of the couples which she mentions have changed status but not all of them. For example Jack Beale is not yet 16 and his partner Joanne is younger than him so I don't think they will have gone up to youth yet. So of course any information for changes really has to come from the dancers themselves and not via third parties. If admin accepted information from anyone there would be a great deal of room for error and also for mischief making by the odd one or two morons who would think it was funny to invent a split partnership or two.
|