Page 1 of 5 123 ... LastLast
Results 1 to 15 of 66
  1. #1
    CC Grandmaster
    Join Date
    Apr 2008
    Posts
    6,764

    Rating Distribution in Different States

    I'm posting this on its own thread, so it doesn't get lost in other discussions. The x-axis is in 100s, of course.
    Attached Images Attached Images

  2. #2
    CC Candidate Master
    Join Date
    Sep 2015
    Posts
    431
    Hi Patrick is this total number of players I can't see the entire image? Would it be possible to work it out based on averages. Of course with NSW population I would expect them to have the most rated players.

  3. #3
    CC Grandmaster
    Join Date
    Apr 2008
    Posts
    6,764
    Quote Originally Posted by Shaun94 View Post
    Hi Patrick is this total number of players I can't see the entire image? Would it be possible to work it out based on averages. Of course with NSW population I would expect them to have the most rated players.
    Those are total numbers, with the horizontal bars going from 0 to 600. I truncated the image, losing the left-hand axes, because I didn't see any point plotting the numbers for the relatively small number of players rated below 600. Although Qld does have a much higher percentage of these than other states.

    I did calculate the average ratings, but I wanted to show that the distributions are fairly similar. Qld has a lower average rating than NSW, but a similar ratings distribution, so the lower average is not due to just having an excess of lower-rated players. But it would be good to standardize the plots.

    The data is from the September 2020 Masterlist, and consists of everyone who had a rating on that list.

  4. #4
    CC Grandmaster
    Join Date
    Apr 2008
    Posts
    6,764
    So these are the standardised plots. I've shifted the Qld plot 200 rating points to the left, so that its peak is now at 1400. And scaled Qld and Victoria so that the area under their curves is the same as NSW.

    I expected that Qld would have relatively fewer higher-rated players than NSW or Victoria, but that doesn't appear to be the case.
    Attached Images Attached Images

  5. #5
    CC Candidate Master
    Join Date
    Sep 2015
    Posts
    431
    Quote Originally Posted by Patrick Byrom View Post
    So these are the standardised plots. I've shifted the Qld plot 200 rating points to the left, so that its peak is now at 1400. And scaled Qld and Victoria so that the area under their curves is the same as NSW.

    I expected that Qld would have relatively fewer higher-rated players than NSW or Victoria, but that doesn't appear to be the case.
    Do you have a theory to go with the data? I think the image itself seems very interesting but would like to know your thoughts.

  6. #6
    CC Grandmaster
    Join Date
    Apr 2008
    Posts
    6,764
    Quote Originally Posted by Shaun94 View Post
    Do you have a theory to go with the data? I think the image itself seems very interesting but would like to know your thoughts.
    The fact that ratings are normally (in a statistical sense) distributed is well known. I think that the difference between Qld and the other two states is the result of two effects. Firstly, there is a relatively large number of under-rated juniors in Qld who are active in ACF-rated events, which has caused a deflation in Qld ratings (relative to NSW and Victoria). Secondly, there is not a lot of mixing between the players in different states, which could have corrected this deflation.

    Both of these effects are described in this paper by Mark Glickman (creator of the Glicko rating system):

    ... But, typically, players who enter the rating pool are assigned low provisional ratings, and players who leave the rating pool are experienced players who have above-average ratings. The net effect of this flux of players is to lower the overall average rating. ...

    ... One of the fundamental problems with using the rating system as a predictor of performance is that it is only accurate on a within-region level. No provisions exist in the rating system to prevent disparities in abilities across different regions of the country for similarly rated players. ...

  7. #7
    CC Candidate Master
    Join Date
    Sep 2015
    Posts
    431
    Quote Originally Posted by Patrick Byrom View Post

    ... One of the fundamental problems with using the rating system as a predictor of performance is that it is only accurate on a within-region level. No provisions exist in the rating system to prevent disparities in abilities across different regions of the country for similarly rated players. ...
    I know that with the Queensland Junior Rating List David McKinnon makes regular adjustments for regions based on the results of state finals or events where there is a large amount of mixing of regional and south east Queensland players. This has been very well done and made a big improvement and ratings more accurate for regions with less events or less higher rated opponents.

    I highly doubt that the ACF ratings officer/s would make such an adjustment for a region such as QLD, this could be considered too hard, possibly unreliable or it may even have something to do with personal ties to various state associations. There could also be a claim that there is not enough clear evidence either way I'm not an expert in this field so can't say.

  8. #8
    CC Grandmaster
    Join Date
    Apr 2008
    Posts
    6,764
    Quote Originally Posted by Shaun94 View Post
    ... I highly doubt that the ACF ratings officer/s would make such an adjustment for a region such as QLD, this could be considered too hard, possibly unreliable or it may even have something to do with personal ties to various state associations. There could also be a claim that there is not enough clear evidence either way I'm not an expert in this field so can't say.
    Actually there was a regional adjustment in ratings made to the Gold Coast (and the ACT) about 20 years ago, as described here by the current ACF rating Officer:
    ... In the case of the ACF V FIDE correction everyone got 70 points hence there is no difference in the Australian pool hence there is no meaningful difference. With regards the ACT and Gold Coast correction the points chnage to the various players was all different, there was no general change. ...

    I think an adjustment to align ratings in Qld with other states would make sense, and could be justified by the fact that the rating distributions seem to be very similar, except that Qld ratings are about 200 points lower across the board. But I'm not sure it would make a significant difference to anyone. And increasing the ratings of everyone rated over 2300 (for example) in Qld by 200 points would obviously create issues - not to mention that it's not so obvious that they are under-rated since they would play more regularly against players from NSW and Victoria.

  9. #9
    CC Candidate Master
    Join Date
    Sep 2015
    Posts
    431
    Quote Originally Posted by Patrick Byrom View Post
    Actually there was a regional adjustment in ratings made to the Gold Coast (and the ACT) about 20 years ago, as described here by the current ACF rating Officer:
    ... In the case of the ACF V FIDE correction everyone got 70 points hence there is no difference in the Australian pool hence there is no meaningful difference. With regards the ACT and Gold Coast correction the points chnage to the various players was all different, there was no general change. ...

    I think an adjustment to align ratings in Qld with other states would make sense, and could be justified by the fact that the rating distributions seem to be very similar, except that Qld ratings are about 200 points lower across the board. But I'm not sure it would make a significant difference to anyone. And increasing the ratings of everyone rated over 2300 (for example) in Qld by 200 points would obviously create issues - not to mention that it's not so obvious that they are under-rated since they would play more regularly against players from NSW and Victoria.
    I'm sure there would be certain benefits potentially making selections and qualifying for events more achievable for Queensland players. Levelling the playing field I guess you could say, these numbers are taken into consideration for many events including Doeberl Cup, Australian Championships and even JETS selections amongst other things.

    I guess NSW having the highest ratings just seems to suit some better than others. I would be interested to see how NSW compare in terms of performance ratings in tournaments such as Aus juniors/ Aus Champs/ Aus open etc. When players mix a little more from state to state.

    I wonder if members of the acf or the ratings officer have an opinion on the data provided here and a potential solution?

    Bill id love to hear from you on this matter, maybe it's best to be raised as a concerned CAQ council member or via the ACF delegate for Queensland.

  10. #10
    Monster of the deep Kevin Bonham's Avatar
    Join Date
    Jan 2004
    Posts
    39,775
    My (see disclaimer) views on this: two ratings pools having a different peak does not necessarily prove that one of them is deflating; need to also look at the age distribution of the pools. If Queensland's active player pool had a much higher proportion of juniors then it would be expected to have a lower average rating.

    Cases where players play players from other states are difficult to use because of confounding effects like age, possible home ground advantage, etc.

    If enough adults are playing then it would be useful to track what has happened on average with the ratings of active adults aged under 50 over the past few years. A slight average decline might or might not mean anything but if they were losing dozens of points a year on average, and there were a reasonable number of them, that would be suggestive.

    The ACF system is very much better at addressing deflation issues than a lot of the Elo based versions. However when player pools consist largely of juniors, and a lot of the juniors are improving very rapidly, and a lot of the juniors stop playing when they leave school, local deflation can still be hard to avoid.
    Moderation Requests: All requests for, comments about, or questions about moderation of any kind including thread changes must be posted in the Help and Feedback section and not on the thread in question. (Or by private message for routine changes or sensitive matters.)

    ACF Newsletter Information - All Australian players and administrators should subscribe and check each issue for relevant notices

    My psephology/politics site (token chess references only) : http://kevinbonham.blogspot.com.au/ Politics twitter feed https://twitter.com/kevinbonham

  11. #11
    CC Grandmaster
    Join Date
    Apr 2008
    Posts
    6,764
    Quote Originally Posted by Kevin Bonham View Post
    My (see disclaimer) views on this: two ratings pools having a different peak does not necessarily prove that one of them is deflating; need to also look at the age distribution of the pools. If Queensland's active player pool had a much higher proportion of juniors then it would be expected to have a lower average rating.
    The plots are based on all the players in the Masterlist, not just currently active players. Wouldn't an excess of lower-rated juniors cause a 'bump' at the bottom of the curve? But the distributions are very similar across the entire ratings spread (once you scale them and adjust the mean).

    Quote Originally Posted by Kevin Bonham View Post
    Cases where players play players from other states are difficult to use because of confounding effects like age, possible home ground advantage, etc.
    True - although this would be the only definitive test.

    Quote Originally Posted by Kevin Bonham View Post
    If enough adults are playing then it would be useful to track what has happened on average with the ratings of active adults aged under 50 over the past few years. A slight average decline might or might not mean anything but if they were losing dozens of points a year on average, and there were a reasonable number of them, that would be suggestive.
    That sounds like a job for the ACF Ratings Officer Although I'm sure that Bill has said somewhere that he is already doing something similar.

    Quote Originally Posted by Kevin Bonham View Post
    The ACF system is very much better at addressing deflation issues than a lot of the Elo based versions. However when player pools consist largely of juniors, and a lot of the juniors are improving very rapidly, and a lot of the juniors stop playing when they leave school, local deflation can still be hard to avoid.
    Agree completely.

  12. #12
    Monster of the deep Kevin Bonham's Avatar
    Join Date
    Jan 2004
    Posts
    39,775
    Quote Originally Posted by Patrick Byrom View Post
    The plots are based on all the players in the Masterlist, not just currently active players.
    If it's just off the master list with no screening for activity level then that introduces other possible explanations. How long has the difference existed? This should be checkable as there are historic versions of the master list back to Dec 1998 on Barry's website.

    Wouldn't an excess of lower-rated juniors cause a 'bump' at the bottom of the curve? But the distributions are very similar across the entire ratings spread (once you scale them and adjust the mean).
    They are strikingly similar. All else being equal, yes, a larger number of juniors would cause a two-peaked distribution or a broader peak. That might not however be the case if it was also accompanied by a different age distribution of adults.
    Moderation Requests: All requests for, comments about, or questions about moderation of any kind including thread changes must be posted in the Help and Feedback section and not on the thread in question. (Or by private message for routine changes or sensitive matters.)

    ACF Newsletter Information - All Australian players and administrators should subscribe and check each issue for relevant notices

    My psephology/politics site (token chess references only) : http://kevinbonham.blogspot.com.au/ Politics twitter feed https://twitter.com/kevinbonham

  13. #13
    CC Grandmaster
    Join Date
    Apr 2008
    Posts
    6,764
    Quote Originally Posted by Kevin Bonham View Post
    If it's just off the master list with no screening for activity level then that introduces other possible explanations. How long has the difference existed? This should be checkable as there are historic versions of the master list back to Dec 1998 on Barry's website.
    I found a similar pattern in the early 90s! But that was all done on paper, unfortunately. However this is from 2003 (with data originally provided by Bill, I believe). It's a bit hard to read, but Qld is the top graph and NSW the bottom one, with the horizontal scale being the same in both.

    2003 Comparison of Ratings.jpg

  14. #14
    CC Rookie
    Join Date
    Oct 2019
    Posts
    4
    It is very interesting to see the discussion regarding to players rating distribution is different between states. Just do a bit work on the current Rooty Hill Chess Tournament (NSW) and Melbourne Chess Club Australia Day Weekend (VIC), plotting with tournaments hold in QLD to observe a similar trend.
    Comparison of Rating Distribution.jpg

  15. #15
    CC Candidate Master
    Join Date
    Sep 2015
    Posts
    431
    Thanks zhuminze, that's a very interesting graph

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Similar Threads

  1. Replies: 24
    Last Post: 22-08-2016, 03:28 PM
  2. Income distribution
    By Johns in forum Politics
    Replies: 3
    Last Post: 17-07-2015, 11:35 AM
  3. prize distribution
    By FM_Bill in forum Arbiters' Corner
    Replies: 9
    Last Post: 24-08-2013, 06:22 PM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •