PDA

View Full Version : The ACF and ratings - has it lost its way?



Brian_Jones
16-01-2010, 09:02 AM
Now that the 2010 Australian Championships have finished I will make a few comments.

1. Garner and Hu should have played in the Championship to make the field up to 30.

2. FIDE ratings should have been used for the selection of players in the Championship and Major.

3. FIDE ratings should have been used for the seedings in both the Championship and the Major.

4. The ACF Rating system (Glicko) should be overhauled. Accuracy should be sacrificed. It needs an audit trail and improved presentation.

5. The ACF Selection Panel needs a minimum of one member who has FIDE qualifications.

6. I hereby announce that I have now given up playing rated one hour chess. As many players already know, the current ACF rating system (glicko) encourages all chess players (even life long players such as myself) to give up playing fast one hour chess. The ACF has won its battle - from now on I will play only FIDE-rated events inside and outside of Australia!


IO Brian Jones

Kevin Bonham
16-01-2010, 10:20 AM
Now that the 2010 Australian Championships have finished I will make a few comments.

1. Garner and Hu should have played in the Championship to make the field up to 30.

Garner was not eligible as has already been explained. Jason Hu was under the threshhold and chose to play in the Major. He was not a rejected applicant for the Champs. Whether the field for the Champs is 28 or 30 or 32 makes virtually no difference despite whatever pseudostatistical urban myths people might advance otherwise.


2. FIDE ratings should have been used for the selection of players in the Championship and Major.

3. FIDE ratings should have been used for the seedings in both the Championship and the Major.

For players in the rating range around the cutoff FIDE ratings can be inflated and insufficiently responsive and are often based on less data than ACF ratings. Why should we use an inferior system?


4. The ACF Rating system (Glicko) should be overhauled. Accuracy should be sacrificed. It needs an audit trail and improved presentation.

I have no problem with your views about improved presentation but that is a separate issue to that of which system you use.

As for an "audit trail", what exactly do you propose and why, why do you think it is worth sacrificing accuracy for, and how many people really care? I'll answer the last one for you: no more than under the old system where everyone could check their rating for themselves but the ratings officers got about the same amount of griping.


5. The ACF Selection Panel needs a minimum of one member who has FIDE qualifications.

Firstly, by ACF Selection Panel, what exactly do you mean? Do you mean the panel for the Championship specifically? Properly, the ACF Selection Panel is the list of 20+ people from whom selection groups of five are selected for picking Aus representatives for events. The committee for the Aus Champs admission decisions is a subcommittee of Council.

Secondly, what qualifications are you suggesting are relevant to selections that are awarded by FIDE, and why? Feel free to amuse the audience by saying that an FM title is relevant however it was obtained. :lol:


6. I hereby announce that I have now given up playing rated one hour chess.

That is up to you.


As many players already know, the current ACF rating system (glicko) encourages all chess players (even life long players such as myself) to give up playing fast one hour chess.

Not true. Any rating system rewards players who prefer time controls that they are good at. In the case of one hour chess, the rating of this time control on the main list might encourage older players to play longer time control games, but might also encourage juniors to play that time control. One thing is for sure, if you took one hour games off the main list net interest in them would collapse.

Garvinator
16-01-2010, 11:22 AM
Not true. Any rating system rewards players who prefer time controls that they are good at. In the case of one hour chess, the rating of this time control on the main list might encourage older players to play longer time control games, but might also encourage juniors to play that time control. One thing is for sure, if you took one hour games off the main list net interest in them would collapse.
I must have missed something in your post here Kevin. It seems like you are saying that having one hour chess in the normal system is good because it will encourage juniors and may discourage adults. If this is true, I accuse you of pro junior bias.

It also seems from your last sentence that you are saying that the only reason people play 60 + 10 is that it is normal rated? Would having 60 + 10 taken off the normal list be so bad. Personally I would love to see it.

Kevin Bonham
16-01-2010, 11:37 AM
I must have missed something in your post here Kevin. It seems like you are saying that having one hour chess in the normal system is good because it will encourage juniors and may discourage adults. If this is true, I accuse you of pro junior bias.

No it is not even a remotely accurate interpretation and indeed it is so ludicrous an interpretation that I accuse you of having a deeply ingrained belief that any system that does not require all juniors to complete a two hour obstacle course followed by a swim across a shark infested river at the commencement of each round (including blitz) is pro-junior biased. :lol:

All I am saying is that under any rating system people who think they are poor at certain time controls within the rating range have the option of preferentially entering the others if they want to protect their rating that desperately. I am not saying it is good that including one-hour games in the normal list could discourage a small number of adults from playing them but I am extremely confident that rating one hour games on the normal list (i) increases predictiveness across all time controls in general and (ii) encourages more chess than it discourages.

From an Aus Champs selection perspective it is fairly irrelevant anyway because if a player could demonstrate an ability to consistently and reasonably frequently perform at 2150 average standard in the slow events then their bad results in faster events would be overlooked under the equiv proficiency rule and they would in my view get in.


It also seems from your last sentence that you are saying that the only reason people play 60 + 10 is that it is normal rated? Would having 60 + 10 taken off the normal list be so bad. Personally I would love to see it.

No but I think that if it was not normal rated it would be significantly less popular to a degree that would be extremely detrimental.

Garvinator
16-01-2010, 12:04 PM
No it is not even a remotely accurate interpretation and indeed it is so ludicrous an interpretation that I accuse you of having a deeply ingrained belief that any system that does not require all juniors to complete a two hour obstacle course followed by a swim across a shark infested river at the commencement of each round (including blitz) is pro-junior biased. :lol: I did not think my assessment was accurate, but it did seem to read that way.

As for the obstacle course, not at all. All I want to see if juniors treated exactly the same as adults when playing in open competitions. Which from my experience does not happen most of the time.


All I am saying is that under any rating system people who think they are poor at certain time controls within the rating range have the option of preferentially entering the others if they want to protect their rating that desperately. I am not saying it is good that including one-hour games in the normal list could discourage a small number of adults from playing them but I am extremely confident that rating one hour games on the normal list (i) increases predictiveness across all time controls in general and (ii) encourages more chess than it discourages.This is an item I would certainly like to explore in more depth, so I might create a different thread for it.

No but I think that if it was not normal rated it would be significantly less popular to a degree that would be extremely detrimental.
Answered just above with idea to maybe create a different thread.

My initial blush is that I disagree with both the conclusion and the premise, but I do see where you are coming from and why you might think that.


From an Aus Champs selection perspective it is fairly irrelevant anyway because if a player could demonstrate an ability to consistently and reasonably frequently perform at 2150 average standard in the slow events then their bad results in faster events would be overlooked under the equiv proficiency rule and they would in my view get in.I would think that the equal proficiency to 2150 rule is created exactly for players who can demonstrate that they are 2150 or better at 90 + 30 or longer, but have poor results at 60 + 10.

Kevin Bonham
16-01-2010, 12:46 PM
I did not think my assessment was accurate, but it did seem to read that way.

No, it only even hinted at reading that way if you were absolutely hellbent on looking for things that weren't there.


As for the obstacle course, not at all. All I want to see if juniors treated exactly the same as adults when playing in open competitions. Which from my experience does not happen most of the time.

Without even getting into whether that is true generally it is certainly not true in this case. Anyone can play at whatever main-list-rateable time control they like and the system "favours" neither seniors nor juniors, although given players can pick and choose whether they want to play more of one kind of event than another.

To remove G60/+10 from main list while leaving long/classic type time controls in would be giving adults concerned about loss of ratings points preferential treatment over juniors for no objectively valid reason.


I would think that the equal proficiency to 2150 rule is created exactly for players who can demonstrate that they are 2150 or better at 90 + 30 or longer, but have poor results at 60 + 10.

It is not there for any one kind of case. There are very many possible reasons to admit players under this rule. Other cases it can cover, for instance (and this list is not exhaustive):

* Players whose ACF rating is sub-2150 but based on very few games compared to a very much higher and more established FIDE rating
* Players who bob around 2150 above and below from list to list and are currently a bit below.
* Players who are usually 2150+ but have recently lost a big pile of points from bad form or one extremely bad event
* Players who are on the way up and yet to reach 2150 but have recent much stronger performances
* Players who appear to be genuinely underrated for whatever reason

and so on.

CameronD
16-01-2010, 01:16 PM
It would be interesting (though impossible) to see what the rating lists look like if only 90+30 or longer are included, Id say they would be entirely different with a lot of swings of 300-500 points.

The problem is their is only 1-2 long control tournaments a year locally and all events are 60+10 (around 90% id guess).

Kevin Bonham
16-01-2010, 01:28 PM
It would be interesting (though impossible) to see what the rating lists look like if only 90+30 or longer are included, Id say they would be entirely different with a lot of swings of 300-500 points.

I don't think this is as impossible as you think and I doubt the swings would be anything like that large for 95+% of players. Even the extant differences between normal and rapid are sometimes in that range, but not habitually.

Sonas did some stuff that even suggested that rating rapids and even blitz on the main list can improve predictiveness if the games carry lower weightings. I'm not regarding that as an established fact since his pool was of very strong players and his methods were post hoc conclusion as I recall rather than a standard hypothesis-test-conclusion method.

Denis_Jessop
16-01-2010, 02:40 PM
1. Garner and Hu should have played in the Championship to make the field up to 30.

This has been fully addressed by Kevin.


2. FIDE ratings should have been used for the selection of players in the Championship and Major.

This goes to the issue of selection criteria for the Championship. There is no selection for the Major - it is open to all who don't qualify for the Championship.

The selection criteria for the Championship need to be revisited (though not necessarily revised) and this will be done in time for the next Championship though sole reliance on FIDE ratings is questionable as the criterion unless some good arguments are put to the ACF. They can be put to me by e-mail as I am the convenor of the ACF Constitutional Subcommittee which will be putting proposals to the Council.

Use of FIDE ratings in the Major has already been the subject of postings recently. Briefly, except for the strongest players, there is generally a discrepancy between FIDE and ACF ratings. The FIDE ratings are higher; sometimes ludicrously so by several hundred points. Those ratings bear no relation to the player's actual playing strength which is the fundamental aim of a rating system.


3. FIDE ratings should have been used for the seedings in both the Championship and the Major.

See the final paragraph in response to point 2.


4. The ACF Rating system (Glicko) should be overhauled. Accuracy should be sacrificed. It needs an audit trail and improved presentation.

I have no idea what Brian is getting at here. Perhaps if he could explain exactly what he means and why, we would all be wiser.


5. The ACF Selection Panel needs a minimum of one member who has FIDE qualifications.

I find this odd. What sort of FIDE qualifications? (International Organiser? :) ) And how would they be of benefit in applying ACF conditions. FIDE has trouble interpreting its own rules and not only because they are poorly written. FIDE has various qualifications - the playing , organiser, arbiter and trainer titles. What is the alleged relevance of these to deciding qualification for entry to the Australian Championship by applying ACF rules?


6. I hereby announce that I have now given up playing rated one hour chess. As many players already know, the current ACF rating system (glicko) encourages all chess players (even life long players such as myself) to give up playing fast one hour chess. The ACF has won its battle - from now on I will play only FIDE-rated events inside and outside of Australia!

I give up here too! I have no idea what Brian thinks the ACF rating system has done to encourage players to give up fast one hour chess. The rating of games of one hour as normal is not part of the Glicko system. It is an ACF Council decision. I'm sure your old club in England will be saddened by your decision not to play in any of its non-FIDE-rated tournaments when you visit. :D

DJ

Brian_Jones
17-01-2010, 10:52 AM
Well it is always good to get a response from two of the three members of the ACF Australian Championship sub-committee (or whatever it is called).

ACF Vice-President Kevin Bonham defends the current system. ACF Vice-President Denis Jessop pleads ignorance. ACF Deputy President Bill Gletsos is silent.

All three members know that the ACF By-laws are guidelines. Everything and anything can be changed subject to ACF Council approval.

David Garner should have played in the 2010 Australian Championship. Although he is English, he is married and living in Melbourne (as is FM Erik Teichmann and Carl Gorka). David has played at least six seasons of 4NCL for Cambridge.

Jason Hu should have played in the 2010 Australian Championship. Jason played in the 2008 Championship scoring 4.5/11 and finished 5th in the 2009 Australian Open with 8/11. Jason entered the 2010 Major because he could. He wanted an easy life. But the ACF sub-committee should have talked to him and offered him a place in the 2010 Championship.

GM Ian Rogers, IA Charles Zworestine or FA Shaun Press should have been appointed to the ACF sub-committee. The current members (Bonham, Jessop, Gletsos) appear to lack experience. They also have some anti-FIDE bias.

What do I mean by a Ratings audit trail?

Well FIDE now has a system where you can now see FIDE rating changes game by game and tournament by tournament.

The ACF rating system is currently a black hole which nobody can see or understand. It must be opened up to inspection and to change.

And finally, where does this ACF 2150 cut-off number come from? Why not 2100 or 2200? What is the reasoning behind this number?

Brian_Jones
17-01-2010, 11:18 AM
I must have missed something in your post here Kevin. It seems like you are saying that having one hour chess in the normal system is good because it will encourage juniors and may discourage adults. If this is true, I accuse you of pro junior bias. .

Kevin has already proven that he is ageist! ;)


It also seems from your last sentence that you are saying that the only reason people play 60 + 10 is that it is normal rated? Would having 60 + 10 taken off the normal list be so bad. Personally I would love to see it.

Personally, I would propose four seperate ACF rating lists:

1. FIDE Standard (90/40 +30 + 30 second increments)
2. ACF Standard (Minimum 60 min)
3. Rapid (Minimum 20 min)
4. Fast Chess (Less than 20 minutes including Blitz)

Denis_Jessop
17-01-2010, 11:58 AM
Well it is always good to get a response from two of the three members of the ACF Australian Championship sub-committee (or whatever it is called).

ACF Vice-President Kevin Bonham defends the current system. ACF Vice-President Denis Jessop pleads ignorance. ACF Deputy President Bill Gletsos is silent.

All three members know that the ACF By-laws are guidelines. Everything and anything can be changed subject to ACF Council approval.

David Garner should have played in the 2010 Australian Championship. Although he is English, he is married and living in Melbourne (as is FM Erik Teichmann and Carl Gorka). David has played at least six seasons of 4NCL for Cambridge.
Jason Hu should have played in the 2010 Australian Championship. Jason played in the 2008 Championship scoring 4.5/11 and finished 5th in the 2009 Australian Open with 8/11. Jason entered the 2010 Major because he could. He wanted an easy life. But the ACF sub-committee should have talked to him and offered him a place in the 2010 Championship.

GM Ian Rogers, IA Charles Zworestine or FA Shaun Press should have been appointed to the ACF sub-committee. The current members (Bonham, Jessop, Gletsos) appear to lack experience. They also have some anti-FIDE bias.

What do I mean by a Ratings audit trail?

Well FIDE now has a system where you can now see FIDE rating changes game by game and tournament by tournament.

The ACF rating system is currently a black hole which nobody can see or understand. It must be opened up to inspection and to change.

And finally, where does this ACF 2150 cut-off number come from? Why not 2100 or 2200? What is the reasoning behind this number?

FA Shaun Press on the Selection Committee? - that's a candidate for the best joke of the century :D

Anyway, the last time that Shaun and I spoke on this topic, Shaun said that he would not do anything more for the ACF until it had an individual membership scheme - not a rational standpoint but one nevertheless. So Shaun has disqualified himself from ACF service for the foreseeable future.

I didn't plead ignorance of anything except the substance of your oyster-like arguments. ;)

None of the three of us is inexperienced as we have all been involved in this for quite a few years, not that experience is necessarily relevant to this matter.

The alleged anti-FIDE bias is not real and, moreover, doesn't come into it. A belief that Glicko 2 is a better system than ELO is not anti-FIDE bias. A view that FIDE drafters are not the world's best is fact, not opinion.

DJ

Bill Gletsos
17-01-2010, 12:55 PM
Well it is always good to get a response from two of the three members of the ACF Australian Championship sub-committee (or whatever it is called).

ACF Vice-President Kevin Bonham defends the current system. ACF Vice-President Denis Jessop pleads ignorance. ACF Deputy President Bill Gletsos is silent.I remained quiet because Kevin & Denis addressed your comments.

All three members know that the ACF By-laws are guidelines. Everything and anything can be changed subject to ACF Council approval.There was no need to change them for the current group of players.

David Garner should have played in the 2010 Australian Championship. Although he is English, he is married and living in Melbourne (as is FM Erik Teichmann and Carl Gorka). David has played at least six seasons of 4NCL for Cambridge.David had not played any ACF rated games prior to the closing date for selections in the championship.
He had not played any FIDE rated games since April 2008.
In the last 2 years he had played 4 FIDE rated games.
In the last 3 years he had played 7 FIDE rated games.
In the last 7 years he had played 20 FIDE rated games.

Jason Hu should have played in the 2010 Australian Championship. Jason played in the 2008 Championship scoring 4.5/11 and finished 5th in the 2009 Australian Open with 8/11. Jason entered the 2010 Major because he could. He wanted an easy life. But the ACF sub-committee should have talked to him and offered him a place in the 2010 Championship. It is not up to the ACF sub-committee to force players to play in the Championship if they wish to play in the Major.

GM Ian Rogers, IA Charles Zworestine or FA Shaun Press should have been appointed to the ACF sub-committee. The current members (Bonham, Jessop, Gletsos) appear to lack experience. They also have some anti-FIDE bias.No doubt we lack experience because we rejected Brian Jones application for the Australian Championship.
As for Shaun Press being on any sub-committee why would the ACF be appointing anyone who is an official of another Federation.

What do I mean by a Ratings audit trail?

Well FIDE now has a system where you can now see FIDE rating changes game by game and tournament by tournament.

The ACF rating system is currently a black hole which nobody can see or understand.Speak for yourself when it comes to understanding.
There are numerous people with an understanding of the Glicko system.

It must be opened up to inspection and to change.So you are another one who wants to sacrifice accuracy for ease of calculation.

And finally, where does this ACF 2150 cut-off number come from? Why not 2100 or 2200? What is the reasoning behind this number?It had been 2100 throughout the 90's. Clearly it was set to this to maintain a minimum standard. It was changed to 2150 in 2004 after the 70 point uplift in March 2004.

Kevin Bonham
17-01-2010, 09:47 PM
David Garner should have played in the 2010 Australian Championship. Although he is English, he is married and living in Melbourne (as is FM Erik Teichmann and Carl Gorka). David has played at least six seasons of 4NCL for Cambridge.

What is your suggestion for a wording of a by-law that would allow for players in such cases to be included? Or should Council just declare exceptions ad hoc?


Jason Hu should have played in the 2010 Australian Championship. Jason played in the 2008 Championship scoring 4.5/11 and finished 5th in the 2009 Australian Open with 8/11. Jason entered the 2010 Major because he could. He wanted an easy life. But the ACF sub-committee should have talked to him and offered him a place in the 2010 Championship.

The whole idea that the ACF sub-committee should have the role of trying to cajole players to play in the Championship when they want to play in the Major is utterly silly. I'm sure if Jason wanted to play in the Champs he would have applied.


GM Ian Rogers, IA Charles Zworestine or FA Shaun Press should have been appointed to the ACF sub-committee. The current members (Bonham, Jessop, Gletsos) appear to lack experience.

Brian, to the extent as your comment reflects on me it is an absolute joke and I hope you will retract it. As you should be well aware I have been involved in selections of various kinds both as a selector and as Selections Director for about eight of your earth years, as well as working on selection committees like this one for ACF Council.

Rogers and Zworestine are excellent and reliable selectors who are among several used frequently in selections administered through the selections systems for Australian representatives and juniors.

The system used for the Aus Champs is somewhat different. The selection is made by Council (not through the ACF Selections Panel) and Council typically does this by appointing a subcommittee of their number. The selections issues that arise when asking a question about "equivalent proficiency" (to 2150) or whether a player is "improving", as well as other Aus Champs field selections issues, are somewhat different ones. It is probably not accidental that the committee is the way it is because it combines experience in (i) ratings (Bill) (ii) interpretation of by-laws (Denis) (iii) selections generally (me). Direct expertise in the first two areas are very important to forming the field for the Aus Champs in a way in which they are much less critical in selecting an Olympiad team, for instance, in which other factors and experiences come into play.


They also have some anti-FIDE bias.

As Denis says it is a matter of fact that there are many things FIDE do badly. But a person's views on FIDE generally have nothing to do with their ability to interpret and apply ACF by-laws re selection for an event.


The ACF rating system is currently a black hole which nobody can see or understand. It must be opened up to inspection and to change.

It is open to change and is changed where a need for fine-tuning is demonstrated based on actual evidence rather than unsound claims or random unspecified whinging that there might be something wrong, but any time a change is made that improves it we have moaners complaining that that is tinkering that somehow shows the system isn't working.


Kevin has already proven that he is ageist!

Twaddle.