Page 2 of 4 FirstFirst 1234 LastLast
Results 16 to 30 of 50
  1. #16
    CC International Master
    Join Date
    Jul 2005
    Posts
    2,156
    One could allocate him with his old rating before the tournament started and win on both accounts - speed and accuracy.

  2. #17
    CC FIDE Master Hobbes's Avatar
    Join Date
    Dec 2008
    Posts
    951
    Wow, hopefully nobody is paying money to use this!
    Leave the gun. Take the cannoli.

  3. #18
    CC Grandmaster
    Join Date
    Apr 2008
    Posts
    4,860
    Quote Originally Posted by Kevin Bonham
    IM Robert Jamieson, listed on the ACF master list with an expired rating of 2438x, plays in a seven game blitz tournament at Glen Eira.

    Jamieson scores 7/7 defeating the following (I give their ChessKids ratings first with their ACF in brackets):
    Daniel Poberezovsky 948 (357)
    Michael Gershenzon 1028 (850?)
    Dejan Krmpotic 1512 (1512)
    Carl Gorka 2173 (2165!!)
    Max Phillips 1132 (1171!)
    Samuel Low 1984 (1984?)
    David Cordover 1969 (1969x)

    If this was rated on an ACF list (and yes the ACF will be rating blitz soon, but I'll let Bill announce the details of that!) then Jamieson's new rating on the list including blitz would be probably something in the mid-2500s with a ?? after it. Probably it would be a bit too high but since he wouldn't be listed on any Top lists and since it would correct quickly if he played some more it wouldn't matter. Especially since it isn't his main rating anyway.

    How does the Tornelo system reward IM Jamieson for his 7/7? Answer: with a new rating of 1935, a "loss" of 503 points. And this isn't just on some rapid list either; it's his new rating!

    This is possible because of two bugs in Tornelo that I've mentioned time and time again. The first is the use of batched averages in rating a player who is being given their first rating on the system. Although jammo's opponents have a wide spread of ratings, Tornelo just averages and treats it as if every player he played is rated 1535. To indicate what a difference it makes, the TPR for scoring 6.5/7 against seven 1535s is 1981. But the TPR for scoring 6.5/7 against the field jammo actually met using the system's ratings for them before the event is 2343.

    The second is Tornelo's unbelievably stupid treatment of a perfect score as +400 against the average of opponent ratings, no matter what the spread of ratings or the length of tournament. So in this case, 1535+400=1935. That is even though, as I have noted above, true performance ratings show that dropping half a point against that field would have been good for a performance 808 points above the straight average.
    I think that the first bug is the real problem. It's easily fixed: if the difference between the rating of a player and that of his opponent is more than 400/less than -400, just replace the rating of the player's opponent by that of the player minus 400/plus 400.

    In this case, the average rating of the field would then be slightly larger than 2438-400 = 2038 (2060 to be precise), so the TPR would then be (using the 400 addition for a perfect score) slightly more than 2440 (2460 to be exact).

    Effectively Vlad is correct - his rating won't change.

  4. #19
    Monster of the deep Kevin Bonham's Avatar
    Join Date
    Jan 2004
    Posts
    37,722
    Quote Originally Posted by Patrick Byrom
    I think that the first bug is the real problem. It's easily fixed: if the difference between the rating of a player and that of his opponent is more than 400/less than -400, just replace the rating of the player's opponent by that of the player minus 400/plus 400.
    Yes something like this is a common solution in ELO systems that use batching. It does have its own minor drawbacks but nothing compared to this fiasco.

  5. #20
    CC Grandmaster road runner's Avatar
    Join Date
    Apr 2006
    Location
    on the skin of the pale blue dot
    Posts
    12,482
    It's not a bug, it's a feature.
    meep meep

  6. #21
    CC FIDE Master
    Join Date
    Jan 2004
    Posts
    934
    Fear not dear Chess Chatters. It is heartening to see such concern over my rating but the Chess Guru has been very kind and has told me that I can have whatever rating I like. I have chosen 2455 for sentimental reasons.

    Fortunately Tornelo is very accommodating in such circumstances whereas I'm told that the ACF rating system can be most unkind to returning players such as myself.
    Still searching for Bobby Fischer....
    and fighting against those humourless bureaucrats who are forever lost in the minutiae.

  7. #22
    Monster of the deep Kevin Bonham's Avatar
    Join Date
    Jan 2004
    Posts
    37,722
    Quote Originally Posted by jammo
    Fortunately Tornelo is very accommodating in such circumstances whereas I'm told that the ACF rating system can be most unkind to returning players such as myself.
    The ACF system is only supposedly "unkind" to returning players whose existing "ratings" are outdated and who then perform way below them. Docking a player 503 points for scoring 7/7 is something else entirely!

    At this stage your Tornelo "rating" remains at 1935 but if you are serious about being able to choose a rating and CG implements that, then it is not the rating program that is being accommodating but rather its owner. After all he won't allow just everyone to have the rating they would like.

  8. #23
    CC FIDE Master Hobbes's Avatar
    Join Date
    Dec 2008
    Posts
    951
    Quote Originally Posted by jammo
    the Chess Guru has been very kind and has told me that I can have whatever rating I like.
    Ah, now I understand why people are paying money to use tornelo!

    Still...
    Leave the gun. Take the cannoli.

  9. #24
    Monster of the deep Kevin Bonham's Avatar
    Join Date
    Jan 2004
    Posts
    37,722
    Perhaps it will come with a premium rating function in which for $25/year you can change the result of a game of your choice, remove any three of your losses or draws from the system and select your own k-factor!

  10. #25
    CC Grandmaster
    Join Date
    Jan 2004
    Posts
    3,444
    I am not sure rating blitz is a good idea, it is not even real chess.

  11. #26
    Reader in Slood Dynamics Rincewind's Avatar
    Join Date
    Jan 2004
    Location
    The multiverse
    Posts
    21,570
    Quote Originally Posted by oepty
    I am not sure rating blitz is a good idea, it is not even real chess.
    Rating blitz is fine provided it is a separate rating to the classic rating and probably best to separate it from rapid ratings as well.
    So einfach wie möglich, aber nicht einfacher - Albert Einstein

  12. #27
    CC FIDE Master James Peirce's Avatar
    Join Date
    Jul 2012
    Location
    The Isle of Tasmania
    Posts
    607
    I am concerned with the issues raised by some Chesschatters. It seems a bit arbitary to "give" someone a rating based on their personal preference. Even more suprising is that it is possible to claim someone played in a tournament and have that effect their rating even though they didn't participate. Particulary in chesskids tournament people will sometimes have more than 1 account and therefore end up with an inaccuruate rating. I know some whose first account has a rating of about 900 and the other account has a rating of over 1180.
    Luck has the peculiar habit of favouring those who don't rely on it!-Anonymous
    Great minds discuss ideas, average minds discuss events, small mind discuss people.-Anonymous
    He who dares, wins.-SAS motto

  13. #28
    CC FIDE Master James Peirce's Avatar
    Join Date
    Jul 2012
    Location
    The Isle of Tasmania
    Posts
    607
    I have also discovered something else with the Glen Eira tournament that seems irregular. Michael Gershenzon had a tornelo rating of 979 before the tournament. He came 5th with 4/7 and a performance rating of 1491 and yet his rating only increased by 11 points which seems ridicolous that there is a 500 point gap between his rating before the tournament and his performance rating. You would think that his rating would go up by more than 11 rating points.
    Luck has the peculiar habit of favouring those who don't rely on it!-Anonymous
    Great minds discuss ideas, average minds discuss events, small mind discuss people.-Anonymous
    He who dares, wins.-SAS motto

  14. #29
    CC FIDE Master James Peirce's Avatar
    Join Date
    Jul 2012
    Location
    The Isle of Tasmania
    Posts
    607
    Quote Originally Posted by jammo
    I have chosen 2455 for sentimental reasons. .
    Tornelo never had 2455 as Robert Jamieson's rating, it states his external rating is 2444 and his tornelo rating was never 2455 at any point.
    Luck has the peculiar habit of favouring those who don't rely on it!-Anonymous
    Great minds discuss ideas, average minds discuss events, small mind discuss people.-Anonymous
    He who dares, wins.-SAS motto

  15. #30
    Monster of the deep Kevin Bonham's Avatar
    Join Date
    Jan 2004
    Posts
    37,722
    Quote Originally Posted by James Peirce
    I have also discovered something else with the Glen Eira tournament that seems irregular. Michael Gershenzon had a tornelo rating of 979 before the tournament. He came 5th with 4/7 and a performance rating of 1491 and yet his rating only increased by 11 points which seems ridicolous that there is a 500 point gap between his rating before the tournament and his performance rating. You would think that his rating would go up by more than 11 rating points.
    I think it was because it was a blitz tournament and therefore rated with a very low k-factor of 5.

    So Tornelo are happy to accept that a tournament being blitz is a legitimate reason to allow it to have only a very low impact on a player's rating. But they are also happy for it to be effectively the tournament that resets a player's rating for a player with inadequate prior data, which in ELO can do enormous damage because the slow rate of change means you take forever to catch up again.

    jammo wasn't the only victim; David Cordover copped it as well, docked 194 points for a blitz tournament because he has few games on the system. But at least in David's case the performance he was docked for was actually bad!

    When will Tornelo realise that their ratings product is rubbish?

Thread Information

Users Browsing this Thread

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

Similar Threads

  1. Replies: 9
    Last Post: 31-08-2009, 12:16 PM
  2. whats happening in the markets
    By Davidflude in forum Non-Chess
    Replies: 2
    Last Post: 19-08-2007, 11:15 AM
  3. Regarding the behaviour of some BB members
    By Alan Shore in forum Non-Chess
    Replies: 324
    Last Post: 14-08-2006, 09:27 PM
  4. Best posts of 2003
    By paulb in forum Australian Chess
    Replies: 289
    Last Post: 29-03-2004, 10:54 AM
  5. ACF Dec 2003 V FIDE Jan 2004
    By Bill Gletsos in forum Ratings Arena
    Replies: 77
    Last Post: 26-02-2004, 12:21 AM

Posting Permissions

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