Page 2 of 2 FirstFirst 12
Results 16 to 29 of 29
  1. #16
    Monster of the deep Kevin Bonham's Avatar
    Join Date
    Jan 2004
    Posts
    34,996
    Quote Originally Posted by SquishyDoggy View Post
    Thats really silly. It means I will just forfeit a tournament after my 17th game of the month (40x17=680) to avoid losing all the gains from my K=40 games.
    And you will (hopefully) then be disciplined for withdrawing from a tournament without good reason. But it will be interesting to see if anyone actually does this. It could be a big issue. It would be a big problem if some scumbag withdrew from a round robin to protect their rating gains and killed somebody else's norm.

    There is a potential problem that for a player who has done exceptionally well in their first batch of games, they need to maintain that performance to maintain the same rating gain. If they are a fast-improving junior they will probably do this, but anyone else will most likely regress to the mean and effectively lose points for performing at their old rating.

    Meanwhile, if I do badly for the month I can just keep playing so all my losses become K=10 so I lose much less.
    That's true but that bit is not a problem. In that case you'll be playing more chess, which is good, but also in order to reduce the losses you will have to play closer to your rating, which will show that you don't deserve to lose that many points.

    And also hardly anyone actually plays 70 FIDE rated games in a month. These issues where a player has several dozen games processed at once are usually caused by tournaments being processed late.
    Last edited by Kevin Bonham; 27-03-2017 at 01:10 PM.

  2. #17
    CC Candidate Master
    Join Date
    Dec 2014
    Posts
    208
    Quote Originally Posted by Kevin Bonham View Post
    And you will (hopefully) then be disciplined for withdrawing from a tournament without good reason. But it will be interesting to see if anyone actually does this. It could be a big issue. It would be a big problem if some scumbag withdrew from a round robin to protect their rating gains and killed somebody else's norm.

    There is a potential problem that for a player who has done exceptionally well in their first batch of games, they need to maintain that performance to maintain the same rating gain. If they are a fast-improving junior they will probably do this, but anyone else will most likely regress to the mean and effectively lose points for performing at their old rating.
    I think it will be more that players simply avoid playing in more than 2 rated tournaments in a month. The prospect of players not being able to realistically gain any further points in subsequent tournaments after having an exceptional start to the month, even if they play well above their rating, will make it hard to sell the new rule.

    As a kind of middle ground, my first thought would be to use a constraint of the form K*sqrt(n) <= constant. Say, for example, K*sqrt(n) <= 160. Then a junior who has not reached 2300 will have K=40 until they play more than 16 games in the month. If they play 64 games, their K factor decreases to 20 (as compared to K=11 under the new rule). But K*n is unbounded above, from which it should follow that the player will continue to gain some points from additional games played, if they perform above their rating throughout the month.

    Quote Originally Posted by Kevin Bonham View Post
    And also hardly anyone actually plays 70 FIDE rated games in a month. These issues where a player has several dozen games processed at once are usually caused by tournaments being processed late.
    Agreed. My K factor is 20, and I am not likely to be able to play 35 rated games in a month any time soon. In fact, I can't remember ever playing 17 FIDE-rated games in a month.
    Southern Suburbs Chess Club (Perth)
    www.southernsuburbschessclub.org.au

  3. #18
    CC Rookie
    Join Date
    Dec 2016
    Posts
    20
    Quote Originally Posted by Kevin Bonham View Post
    And you will (hopefully) then be disciplined for withdrawing from a tournament without good reason. But it will be interesting to see if anyone actually does this. It could be a big issue. It would be a big problem if some scumbag withdrew from a round robin to protect their rating gains and killed somebody else's norm.
    Easy, just play at open tournaments where there is no norm, or take the bye/forfeit early in the tournament.

    Secondly, its really FIDE's fault. How the hell do they come up with 17 games exactly?!?! Knowing a tournament is usually 9 rounds, if anything 18 games would be PERFECT but they had to come up with 17 games for K=40 which is very very dumb. Basically they are telling juniors, "oh don't play 2 tournaments a month or more." This is more a clerical error that they don't know 17 games would be just under two tournaments. It's probably because they are lazy and like whole numbers (700 lol).

  4. #19
    CC Candidate Master
    Join Date
    Dec 2014
    Posts
    208
    Quote Originally Posted by SquishyDoggy View Post
    Easy, just play at open tournaments where there is no norm, or take the bye/forfeit early in the tournament.

    Secondly, its really FIDE's fault. How the hell do they come up with 17 games exactly?!?! Knowing a tournament is usually 9 rounds, if anything 18 games would be PERFECT but they had to come up with 17 games for K=40 which is very very dumb. Basically they are telling juniors, "oh don't play 2 tournaments a month or more." This is more a clerical error that they don't know 17 games would be just under two tournaments. It's probably because they are lazy and like whole numbers (700 lol).
    The `700 rule' presumably is inspired by Article 12.3 of the FIDE Rating Regulations:

    Code:
    12.3
    K is used as a stabilising influence in the system. When K = 10, the rating turns over in approximately 70 games; K = 20, it is 35 games; K = 40, it is 18 games.
    If you play 18 FIDE-rated games in a month, then your K factor for that month will be 700/18 ~ 39, as opposed to 40.

    I don't think this should really discourage anyone from playing. My K factor is 20, and I still play. Strong players have K=10, and are still playing.
    Last edited by Andrew Hardegen; 27-03-2017 at 09:39 PM.
    Southern Suburbs Chess Club (Perth)
    www.southernsuburbschessclub.org.au

  5. #20
    CC Rookie
    Join Date
    Dec 2016
    Posts
    20
    Wait andrew I don't understand, I thought Kevin is saying the moment you play your 18th game, your entire K-factor for this whole month, drops from 40->10

    Quote Originally Posted by Kevin Bonham View Post
    From mid this year a change is expected to come in in which when a player has a large number of games processed in a month their k-factor is reduced for all those games (but not permanently). The idea is that k for that period is reduced such that k*n does not exceed 700, where n is the number of games. So if a player with normally k=40 has 70 games processed in a period their k is 10 for that period. Not sure if this has been finally approved yet or if current PB will do so.
    Oh I think I MISREAD this completely, I thought k*n can't be greater than 700, or else it drops back to 10 instantly....

    So let say an example of 21 games, the k factor of a junior becomes 700/21 = 33 instead of k=40

  6. #21
    CC Candidate Master
    Join Date
    Dec 2014
    Posts
    208
    Quote Originally Posted by SquishyDoggy View Post
    So let say an example of 21 games, the k factor of a junior becomes 700/21 = 33 instead of k=40
    Yes I think that's it.

    As should be apparent from my comments, I partly agree with your concerns about the rule.

    Regarding the arguments that were actually put forward at the FIDE proceedings where the new rule was brought into place: I hope I am not being too unkind in asking whether they were largely centred around the following piece of wisdom?

    Code:
    12.3
    K is used as a stabilising influence in the system. When K = 10, the rating turns over in approximately 70 games; K = 20, it is 35 games; K = 40, it is 18 games.
    Last edited by Andrew Hardegen; 27-03-2017 at 10:52 PM.
    Southern Suburbs Chess Club (Perth)
    www.southernsuburbschessclub.org.au

  7. #22
    CC International Master
    Join Date
    Jan 2004
    Location
    Wynyard,Tas
    Posts
    1,984
    Quote Originally Posted by Andrew Hardegen View Post
    Regarding the arguments that were actually put forward at the FIDE proceedings where the new rule was brought into place: I hope I am not being too unkind in asking whether they were largely centred around the following piece of wisdom?
    The short answer is probably Yes, and you're not being unkind. The ratings are, after all, supposed to be an (approximately) accurate representation of performance, not a treat for the kiddies.

    Consideration of typical cases suggests that the cutoff figure of 700, or 17.5 games, for scaling back K is not arbitrary. For example if you play 17 games against opposition your own rating and score 12 points then that's 3.5 above your expected score of about 50% or 8.5 and a rating increase of 3.5 x 40 = 140. 12 out of 17 is about 70% which equates on the Elo table to a difference of 149. So that's fine, but if K stayed at 40 and you kept scoring at the same level you would race away from a sensible result at the same rate as you initially approached it.

  8. #23
    CC Rookie
    Join Date
    Dec 2016
    Posts
    20
    What does "turn over" mean, does it mean you can't earn anymore points?

    Let say an underrated 2000 junior beats 2000 player consecutively non stop next month.

    17 games in a row K=40, thats 20 points a game, he earns 340 points
    18 games in a row K=38.88 thats 19.4 points a game, he earns 350 points
    19 games in a row K=36.8 thats 18.4 a game, he earns 350 points
    at n games in a row K=700/n thats 350/n a game, he earns 350 points

    So I assume the effect someone stands with other kinds of results too, but losing a game etc would make things more complex...

  9. #24
    CC International Master
    Join Date
    Jan 2004
    Location
    Wynyard,Tas
    Posts
    1,984
    Quote Originally Posted by SquishyDoggy View Post
    Let say an underrated 2000 junior beats 2000 player consecutively non stop next month.
    Then his name is probably Magnus. Or at least Sergey. And even in that extreme case his rating will be right after another couple of months. If you model a realistic scenario you'll get a better idea of what's happening.

  10. #25
    CC Rookie
    Join Date
    Dec 2016
    Posts
    20
    Quote Originally Posted by Ian Rout View Post
    Then his name is probably Magnus. Or at least Sergey. And even in that extreme case his rating will be right after another couple of months. If you model a realistic scenario you'll get a better idea of what's happening.
    No this is plenty realistic, underrated juniors beating up U2000 fields for perfected near perfect score happens all the time.

    E.g. just look at this player https://ratings.fide.com/card.phtml?event=4902661

    She went like 27/27 against 1800 level club players and gained like 440 rating in a month.

    Going near perfect score against U2000 fields is pretty realistic for 2100-2200 level players even, especially when they are underrated at 1800-1900 and playing other 1800-1900s.

  11. #26
    Monster of the deep Kevin Bonham's Avatar
    Join Date
    Jan 2004
    Posts
    34,996
    Quote Originally Posted by SquishyDoggy View Post
    What does "turn over" mean, does it mean you can't earn anymore points?

    Let say an underrated 2000 junior beats 2000 player consecutively non stop next month.

    17 games in a row K=40, thats 20 points a game, he earns 340 points
    18 games in a row K=38.88 thats 19.4 points a game, he earns 350 points
    19 games in a row K=36.8 thats 18.4 a game, he earns 350 points
    at n games in a row K=700/n thats 350/n a game, he earns 350 points
    The modified ks are rounded down to the next whole number, thus 40, 38, 36, 35.

    If the player is really much better than 2350 then having them go to 2350 first won't do much harm as they will soon rack up many more points.

    I note that the QC report has been approved with modifications but I doubt this one was modified. There was much more contentious stuff in it than that.

  12. #27
    CC International Master
    Join Date
    Jan 2004
    Location
    Wynyard,Tas
    Posts
    1,984
    Quote Originally Posted by SquishyDoggy View Post
    No this is plenty realistic, underrated juniors beating up U2000 fields for perfected near perfect score happens all the time.

    E.g. just look at this player https://ratings.fide.com/card.phtml?event=4902661

    She went like 27/27 against 1800 level club players and gained like 440 rating in a month.

    Going near perfect score against U2000 fields is pretty realistic for 2100-2200 level players even, especially when they are underrated at 1800-1900 and playing other 1800-1900s.
    I'm not saying it could never happen but that it's somewhat of an anomaly if it does, and even then usually not a genuine improvement, and hardly a basis for policy. It's more useful to consider less fantastic cases.

    While we can't ascribe a performance rating to a 100% score, for someone to score 17.5/18 or 97% they would be performing, on the Elo table, at 538 points above their opposition. So a 2000 player scoring 100% against that many 2000 players is playing somewhere above 2538, putting them in the company of the world's best 500 players while having only earned a rating of 2000. For perspective, David Smerdon is 2533. If it happens at all it would have an interesting story behind it.

  13. #28
    CC Rookie
    Join Date
    Dec 2016
    Posts
    20
    Quote Originally Posted by Ian Rout View Post
    I'm not saying it could never happen but that it's somewhat of an anomaly if it does, and even then usually not a genuine improvement, and hardly a basis for policy. It's more useful to consider less fantastic cases.

    While we can't ascribe a performance rating to a 100% score, for someone to score 17.5/18 or 97% they would be performing, on the Elo table, at 538 points above their opposition. So a 2000 player scoring 100% against that many 2000 players is playing somewhere above 2538, putting them in the company of the world's best 500 players while having only earned a rating of 2000. For perspective, David Smerdon is 2533. If it happens at all it would have an interesting story behind it.
    Yea, but rating doesn't really work linearly like that. Even tho thats what the mathematical model try to encapsulate. The sample is never perfect.

    There are plenty of players that play weak fields to boister their rating. An extreme example is Afromeev (although it's suspect in his case). You can get to a point where a player of certain rating would have near 0% against you because your style/repertoire counters the weakness at their rating ranges.

    I just think low K-factors is especially unfavourable to people from far off countries like Australia with level access and high financial cost to compete regularly in tournaments. I hear plenty of reviews about how european players seem overrated but really it's most ppl here being underrated (small pool). In fact, I would say that underrated players is way more of a problem than overrated players, especially in a region like this one. Having underrated players prevent growth.
    Last edited by SquishyDoggy; 28-03-2017 at 11:09 PM.

  14. #29
    Monster of the deep Kevin Bonham's Avatar
    Join Date
    Jan 2004
    Posts
    34,996
    The protection against overshooting has been added to the 2017 ratings regs:

    If the number of games (n) for a player on any list for a rating period multiplied by K (as defined above) exceeds 700, then K shall be the largest whole number such that K x n does not exceed 700.

    It comes in from July 1.

    https://www.fide.com/fide/handbook.h...7&view=article

Thread Information

Users Browsing this Thread

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

Similar Threads

  1. FIDE K factor - stuck at 30?
    By SHump in forum Ratings Arena
    Replies: 5
    Last Post: 08-04-2014, 10:34 AM
  2. The mongrel factor
    By PHAT in forum General Chess Chat
    Replies: 45
    Last Post: 24-08-2004, 11:29 PM
  3. chicken factor
    By Garvinator in forum Completed Tournaments
    Replies: 4
    Last Post: 06-08-2004, 09:39 PM
  4. Dynamite Factor
    By Blitzer in forum Completed Tournaments
    Replies: 16
    Last Post: 07-01-2004, 08:11 AM
  5. Name Change
    By skip to my lou in forum News / Announcements
    Replies: 0
    Last Post: 03-10-2003, 05:02 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
  •