I’m planning to submit the following ADM to amend Chapter 8, Section 1 of the rulebook to define what is meant by an established rating, a provisional rating and a published rating.
In Chapter 8, The USCF Rating System, Section 1, Rated games, add the following as the second paragraph of this section:
An established rating is a rating based on at least 26 games. A provisional rating is a rating based on fewer than 26 games. In order to be published in a rating supplement, a rating must be based on at least four games.
RATIONALE: The terms “established rating”, “provisional rating” and “published rating” aren’t currently defined in the rulebook. The proposed new paragraph defines these terms which are used in the match rules in section 2.
Question for rating system experts: if a player has played more than 25 games but has won all of those games or lost all of those games, is the player’s rating considered to be established or provisional? As I read the rating system document, the player has an established rating but the special rating formula will be used calculate his new rating. Since the player has an established rating he can get a rating floor, for example. Is that right?
Chapter 8 of the Official Rules of Chess is non-normative. The document that contains the normative definition of the rating system is the paper maintained by Dr. Mark Glickman, chair of the ratings committee.
If memory serves me correctly, the version of chapter 8 in the sixth edition is the rulebook editor’s best attempt to provide a chapter on the rating system. I believe the ratings committee was asked to provide the content for chapter 8 for the sixth edition, but the committee could not do so in the time frame provided by the tight schedule for the sixth edition. (Bear in mind that the sixth edition really is the fifth edition with the rules changes document applied thereto.)
The rating system is not under control of the delegates. The Executive Board approves changes to the rating system, and the Executive Board are no longer under the control of the delegates. I believe ADMs to modify chapter 8 are misplaced at best. Personally, I think chapter 8 really doesn’t even belong in the rulebook. I’m guessing it’s there because it was there in previous editions. It may have made sense to have it in previous editions, back in a time when “go to the USCF web site and download the PDF of the rating system specification” would have sounded like meaningless gibberish.
(Edit: Fix Prof. Glickman’s name. I apologize for the error.)
Given that Chapter 8 is part of the rulebook, I think the information it contains should be accurate, i.e. it should agree with what’s posted in Professor Glickman’s rating system document. I’m not trying to change the way the rating system works.
As chapter 8 is not a normative reference for the USCF rating system, and as the version of chapter 8 in the sixth edition was essentially created by the rulebook editor without any delegate approval, I still do not understand the need for an ADM to effect changes to the chapter.
Perhaps I will introduce a motion to replace the entire contents of chapter 8 with a reference to Dr. Glickman’s paper.
(Edit: Fix Prof. Glickman’s name. I apologize for the error.)
That may be a difference between the description of the rating system in Prof. Glickman’s rating system document and the implementation. According to the rating system document, an established rating is any rating based on over 25 games. If all the games were won or lost the games are rated using the special rating formula.
That may indeed be true. However, with the warning that what I know about statistics would barely fill a thimble, it seems to me that an established rating based on all wins or all losses is, per se, absurd. The only information the rating system has is that the player in question is either stronger (all wins) or weaker (all losses) than all the opponents. There is absolutely nothing that gives the rating system any idea how much stronger or weaker the player is. This seems to directly contradict the notion of an established rating indicating that the rating system has a certain level of confidence in the player’s rating.
Of course, it has been quite some time since provisional ratings were calculated as average of opponents +/- 400 for each win/loss. Once the player has enough games to be off the special formula, the only effect of a low number of games is to raise the value of the K factor.
Why do we even need the terms “established” and “provisional”?
There are two good answers.
One, as a statement by the rating system as to how much confidence it (the rating system) has in the player’s rating, and as a factor in determining which formula (regular or special) to use.
Two, as a criterion for organizers to use, if they wish, in determining tournament eligibility, section eligibility, prize eligibility, etc.
For purpose one, it makes sense to declare a player provisional until he has played 26 games and has at least 1 win and 1 loss, or at least 1 draw.
For purpose two, it makes more sense to declare a player provisional until he has played 26 games, period.
So, have it both ways. Go ahead and list a player as P86 if he has lost 86 games. The organizer, seeing the P86, can still treat him as established for eligibility purposes.