PDA

View Full Version : historical consistency sf. 2010 chess goals



Santa
09-01-2010, 06:27 AM
Your April 1996 rating was 1397. You received a 100 increase as part of the April 2000 uplift.


How does that help historical consistency?

I was talking to to Bill Jordan a couple of years ago, discussing players ratings way back then, and now, and how players ratings' were higher now than formerly. It would have been well explained by these adjustments, but Bill didn't mention it, and I was completely unaware of it.

Bill Gletsos
09-01-2010, 05:35 PM
How does that help historical consistency?There was no historical consistency to start with.

BTW what happened to the Compaq computer?

Santa
09-01-2010, 10:25 PM
There was no historical consistency to start with.


The question is how your actions help historical consistency.

Bill Gletsos
09-01-2010, 11:04 PM
The question is how your actions help historical consistency.My answer remains the same.
There was no historical consistency to start with.

BTW what happened to the Compaq computer?

Kevin Bonham
09-01-2010, 11:24 PM
What is the point of historical consistency if the previous ratings had (i) become seriously flawed as the result of use of insufficiently dynamic systems and other problems and (ii) significantly lost parity with FIDE ratings at the pointy end?

Why strive to be consistent with suboptimal figures?

Santa
16-01-2010, 09:25 PM
What is the point of historical consistency if the previous ratings had (i) become seriously flawed as the result of use of insufficiently dynamic systems and other problems and (ii) significantly lost parity with FIDE ratings at the pointy end?

Why strive to be consistent with suboptimal figures?


I made some points about the earlier system some years ago, and proposed a new system which was ready for testing when the ACF decided it was happy with the present system.

Whatever was wrong with the old system (and I'd like your reasoned views on that), it does not justify making random changes to current ratings, changes that are not related to players' performance.

If the current maths is working well, why are regular changes needed?

Why is parity with FIDE important?

Kevin Bonham
16-01-2010, 10:04 PM
Whatever was wrong with the old system (and I'd like your reasoned views on that), it does not justify making random changes to current ratings, changes that are not related to players' performance.

What "random changes" supposedly unrelated to player performance did you have in mind?


If the current maths is working well, why are regular changes needed?

They aren't. Since the Glicko system was started there have been two major changes, the 150 point uplift which was to correct problems caused by the prior systems and one 70 point uplift essentially for FIDE parity.

There have been a number of minor improvements but many of these are along the lines of taking a good thing and making it work even better (for instance re-runs of past ratings with more complex formulae to implement intermediate ratings). Yet for some reason there is this perverse line of argument that says that if you take a good system and fine tune it to make it even better, then that means it must be suspect compared to a hypothetical inferior system that just sits there being inferior. I don't get this.

A crucial difference between ratings now and 20+ years ago is that a ratings system can now make far more complicated use of the capacities of computers. It is no surprise that better and more complex ways of modelling and predicting player performance have come about, just as they have in many other similar fields of endeavour. I sometimes work in population modelling for threatened species, for example. You can do stuff now you just couldn't do 30+ years ago because of the advances in computing.


Why is parity with FIDE important?

It's no big deal, but "historical consistency" with previous more flawed systems is even less of a bigger deal. As for the case in favour, it allows for the use of FIDE ratings of players arriving from overseas to seed their ACF rating and also allows for easy seeding of such players in tournaments. True, the ACF could instead decree official adjustements but anyone who is concerned about historical consistency can make those themselves to past ACF ratings by adding whatever uplifts they consider represent a genuine adjustment to the base rather than just correction for deflation.

As for the previous system my major issues with it were insufficiently dynamic response to two classes of players: improving players (esp juniors) and inactive players who play rarely. At times fixes such as bonus points were tried for the first but because the model was simplistic it was prone to overshoot.

Santa
16-01-2010, 10:27 PM
My answer remains the same.
There was no historical consistency to start with.


A key objective of Elo's recommendations was for ratings to be consistent over time, and he tested his system with data spanning 125 years and found, with the FIDE implementation, they were. See 2.8 "The test of time" in The rating of Chessplayers Past and Present.

Whatever the faults of the previous system, it wasn't given to enormous jumps overnight, and whatever its faults, the new system should have addressed them.

The question hasn't been addressed, but needs to be.

NorthsCC (Axiom)
16-01-2010, 11:02 PM
A key objective of Elo's recommendations was for ratings to be consistent over time, and he tested his system with data spanning 125 years and found, with the FIDE implementation, they were. See 2.8 "The test of time" in The rating of Chessplayers Past and Present.

Whatever the faults of the previous system, it wasn't given to enormous jumps overnight, and whatever its faults, the new system should have addressed them.

The question hasn't been addressed, but needs to be.
But what happened to the Compaq computer?

Santa
16-01-2010, 11:31 PM
What "random changes" supposedly unrelated to player performance did you have in mind?



They aren't. Since the Glicko system was started there have been two major changes, the 150 point uplift which was to correct problems caused by the prior systems and one 70 point uplift essentially for FIDE parity.

The new ratings system should be capable of handling those itself given some time, and doing so smoothly.

Elo discusses rating a pool comprising only unrated players (3.3), a suggestion is to enter them all into the system with the same ratings and large K (up to 50). After 30 games, he says, the players' relative ratings will be consistent.

I might do that and calculate their performance ratings, then start again using their performance ratings as their initial ratings.



There have been a number of minor improvements but many of these are along the lines of taking a good thing and making it work even better (for instance re-runs of past ratings with more complex formulae to implement intermediate ratings). Yet for some reason there is this perverse line of argument that says that if you take a good system and fine tune it to make it even better, then that means it must be suspect compared to a hypothetical inferior system that just sits there being inferior. I don't get this.


I'm not the only person asking about how the new system works. In Shaun Press's time, the calculations and formulae were published in the ratings book. When I made changes, those were documented.

Now, all I've been able to find is that the Ratings Officers were asked to develop a new system like, but not exactly the same, the Glicko2 system.

I've been reading old tournament results in old magazines. They publish estimated ratings changes attributable to that tournament. I can see that Geoff Saw's performance rating in his first tournament was 1525 (obviously I did something right! and another Waverley junior performed at 1076 - better than I did!. That tournament (unofficially, of course) cost me 8 points.



A crucial difference between ratings now and 20+ years ago is that a ratings system can now make far more complicated use of the capacities of computers. It is no surprise that better and more complex ways of modelling and predicting player performance have come about, just as they have in many other similar fields of endeavour. I sometimes work in population modelling for threatened species, for example. You can do stuff now you just couldn't do 30+ years ago because of the advances in computing.


I've been using computers for over 40 years, I well understand the changes that have occurred in the time frames you talk about. The ACF has enjoyed the use of computers for 20 years or so.

If the ACF published the formulae and algorithms used to calculate ratings, then others could implement them themselves, cast critical eyes over the results and, maybe, question discrepancies and suggest improvements.


It's no big deal, but "historical consistency" with previous more flawed systems is even less of a bigger deal. As for the case in favour, it allows for the use of FIDE ratings of players arriving from overseas to seed their ACF rating and also allows for easy seeding of such players in tournaments. True, the ACF could instead decree official adjustements but anyone who is concerned about historical consistency can make those themselves to past ACF ratings by adding whatever uplifts they consider represent a genuine adjustment to the base rather than just correction for deflation.

For a very long time, the ACF has entered foreign FIDE-rated players into Australian tournaments with ACF=FIDE.

There has never been any need to do this, it would be permissible to make a more accurate conversion. If the evidence is there that FIDE ratings are 200 points above Australian ratings, the transformation is easy: enter a 2300 player at 2100. Or, rank players with FIDE ratings and assume Australians without FIDE ratings have FIDE=ACF+200.

For ratings calculations, those players should be handled as unrated players, and their ACF-equivalent ratings not stored.



As for the previous system my major issues with it were insufficiently dynamic response to two classes of players: improving players (esp juniors) and inactive players who play rarely. At times fixes such as bonus points were tried for the first but because the model was simplistic it was prone to overshoot.

The reason that it was prone to overshoot was easily fixed. Ratings changes for a period were calculated with players' ratings unchanging for the period. if my rating was 1300 on a ratings list, all my ratings changes (and those of all my opponents) were calculated with that rating.

Instead, tournaments should have been ordered by date (I suggest the date of the middle round to cater better for tournaments running for some weeks), and ratings updated at the end of each tournament. Then, use the end-of-tournament rating for the players' next event.

If I earned a rating increase for my first tournament of a period of 100 points, my next tournament would be rated with my rating set to 1400, not 1300.

That would pretty much kill the overshoot. However, the ACF would not agree to that change.

Elo has K pretty much stable for a player, but Elo was principally targetting established, strong players. Really, there is no reason it should not be adjusted from time to time for any number of reasons such as
A prolonged absence from the game
Recent results inconsistent with a long-established rating.
Illness affecting a players' skill (a stroke maybe) or ability to play (impending blindness?).

There's no real need to keep players ratings for ever, I'd be happy to drop players ratings three (but you choose a number!) years after their last game. Entering them again, if they return, should not be a big deal, and there's no good reason to suppose their old rating has any meaning.

Bill Gletsos
16-01-2010, 11:35 PM
I made some points about the earlier system some years ago, and proposed a new system which was ready for testing when the ACF decided it was happy with the present system.Yes, you were the one who deleted over 8,000 names from the ACF master file in 1993 without ACF Council approval.
You continually refused to re-add them.
As such you were replaced as ACF ratings Officer in mid 1996.

BTW what happened to the Compaq computer?

Bill Gletsos
16-01-2010, 11:55 PM
A key objective of Elo's recommendations was for ratings to be consistent over time, and he tested his system with data spanning 125 years and found, with the FIDE implementation, they were. See 2.8 "The test of time" in The rating of Chessplayers Past and Present.The point is he may have said that but it has not been the case.
most people who have studied the FIDE ratings believes that the FIDE Elo system has been inflating virtually every year.


Whatever the faults of the previous system, it wasn't given to enormous jumps overnight, and whatever its faults, the new system should have addressed them.As usual you dont understand. This has nothing to do with Glicko Vs Elo.

The problem was with the ACF Elo system of the 1980's & 1990's.
Because of the 336 rule players at the top of the rating list continued to gain points they did not deserve and players at the bottom of the list continued to lose points that they should not have lost. Also the bonus point scheme in place generally caused massive overshoots in one list that lead to loss of points by those players in the next list.
All the while the ratings of players were deflating compared to their FIDE ratings.

The ACF Council decided that it would apply an uplift to the December 1999 ACF ratings to try and combat this long standing deflation of the ACF ratings and inflation of the FIDE ratings.

BTW what happened to the Compaq computer?

Bill Gletsos
16-01-2010, 11:56 PM
But what happened to the Compaq computer?Yes he seems to be deliberately ignoring this question.

Bill Gletsos
16-01-2010, 11:58 PM
Since the Glicko system was started there have been two major changes, the 150 point uplift which was to correct problems caused by the prior systems and one 70 point uplift essentially for FIDE parity.
Actually the 150 point uplift occurred in April 2000 under the ACF Elo system. It had nothing to do with the Glicko system.

Kevin Bonham
17-01-2010, 04:43 AM
The new ratings system should be capable of handling those itself given some time, and doing so smoothly.

How is any system supposed to fix a system-wide deflation caused by the failures of a previous system? (Bill notes correctly that the uplift was pre-Glicko; I had thought it came just after Glicko started, but in any case the point was that the old system created the problems that led to that uplift, not Glicko.)


Elo discusses rating a pool comprising only unrated players (3.3), a suggestion is to enter them all into the system with the same ratings and large K (up to 50). After 30 games, he says, the players' relative ratings will be consistent.

I don't see how this is relevant to anything whatsoever since it has no bearing on a system that consists largely of players with established ratings being continued from one system to the next.

Furthermore even if the players' relative ratings are consistent with each other a closed system proves nothing. It is how a system responds to situations caused by players entering and leaving the pool at various strengths and trajectories that is a bigger test.

I might do that and calculate their performance ratings, then start again using their performance ratings as their initial ratings.


I'm not the only person asking about how the new system works. In Shaun Press's time, the calculations and formulae were published in the ratings book. When I made changes, those were documented.

Now, all I've been able to find is that the Ratings Officers were asked to develop a new system like, but not exactly the same, the Glicko2 system.

I've been reading old tournament results in old magazines. They publish estimated ratings changes attributable to that tournament. I can see that Geoff Saw's performance rating in his first tournament was 1525 (obviously I did something right! and another Waverley junior performed at 1076 - better than I did!. That tournament (unofficially, of course) cost me 8 points.

I don't see how any of this is relevant to the quote from me it was written in reply to. You asked why regular changes are needed. I disposed of the argument that (supposedly) regular fine tuning demonstrates inferiority in a system (in fact it can often demonstrate that a system is well maintained). You then go talking about the difficulty in finding details of the system but that is not the same thing as evidence that the system itself is not working.


If the ACF published the formulae and algorithms used to calculate ratings, then others could implement them themselves, cast critical eyes over the results and, maybe, question discrepancies and suggest improvements.

People can do that anyway. You don't need to know the formulae used to calculate ratings to have a reasonable idea of whether those ratings are proving predictively accurate (or not) in practice.


For ratings calculations, those players should be handled as unrated players, and their ACF-equivalent ratings not stored.

That is simply throwing away information that whatever the flaws in FIDE ratings is nonetheless still predictively useful.


There has never been any need to do this, it would be permissible to make a more accurate conversion.

An adjustment for parity makes no difference to the accuracy of a conversion except that it removes the need for one.


That would pretty much kill the overshoot. However, the ACF would not agree to that change.

Ordering tournaments by middle date is not practical since a weekender late in a ratings period could get rated before a weekly tournament with an earlier middle date that extends into the next period, even though most games in the latter event would be earlier.


Elo has K pretty much stable for a player, but Elo was principally targetting established, strong players. Really, there is no reason it should not be adjusted from time to time for any number of reasons such as
A prolonged absence from the game
Recent results inconsistent with a long-established rating.
Illness affecting a players' skill (a stroke maybe) or ability to play (impending blindness?).

What Glicko does is effectively adjust k based on the evidence and the time a player has had off the game so Glicko is several steps ahead of you there. Of course any system that required adjustment for illness would be a complete nightmare to police; ratings officers are ratings officers, not health monitors of their player base.


There's no real need to keep players ratings for ever, I'd be happy to drop players ratings three (but you choose a number!) years after their last game. Entering them again, if they return, should not be a big deal, and there's no good reason to suppose their old rating has any meaning.

You seem to like throwing away information that may still be quite predictively useful. Keeping the old rating but allowing its influence compared to new results to wane over time is what Glicko does.

The only problem we have at the moment is some very inactive players who know they are overrated as hell and don't want to lose ratings points, trying to pretend their ancient rating still means anything.

Desmond
17-01-2010, 08:14 AM
I have to admit I'm getting intrigued about this Compaq computer. Was this the old ratings officer machine?

Bill Gletsos
17-01-2010, 09:13 AM
For a very long time, the ACF has entered foreign FIDE-rated players into Australian tournaments with ACF=FIDE.

There has never been any need to do this, it would be permissible to make a more accurate conversion. If the evidence is there that FIDE ratings are 200 points above Australian ratings, the transformation is easy: enter a 2300 player at 2100. Or, rank players with FIDE ratings and assume Australians without FIDE ratings have FIDE=ACF+200.

For ratings calculations, those players should be handled as unrated players, and their ACF-equivalent ratings not stored.Yes I recall that when you were the ACF National Ratings Officer overseas GM's and IM's were treated as unrated when they played in Australia even though it was pointed out to you that this was plainly stupid.

BTW what happened to the Compaq computer?

Brian_Jones
17-01-2010, 10:07 AM
The only problem we have at the moment is some very inactive players who know they are overrated as hell and don't want to lose ratings points, trying to pretend their ancient rating still means anything.

A very commital statement Kevin! So you think this is the only problem with the current ACF ratings?

You must have your head in the sand.

Just look at the rating of the players in the recent Solomon Islands International in Honiara. Indeed why was this ACF rated at all?

Maybe ACF is trying to make the ACF Rating system more important to compete with FIDE Ratings which now have a good audit trail (which the ACF Rating system has not!)

BTW who cares a **** about the Compaq computer?

Bill Gletsos
17-01-2010, 01:00 PM
Just look at the rating of the players in the recent Solomon Islands International in Honiara. Indeed why was this ACF rated at all?It was rated with the approval of the ACF Council at the request of Oceania Zone President Gary Bekker.

Given one of its aims was to generate FIDE ratings for the Solomon Island players it is strange that it does not appear to have been rated by FIDE.

Kevin Bonham
17-01-2010, 10:25 PM
A very commital statement Kevin! So you think this is the only problem with the current ACF ratings?

You have removed my comment from its context which was specifically that it was a response to santa's comment about inactive players. I was saying it was the only problem we have that arises as a result of keeping the ratings of inactive players as an alternative to chucking them out.

As for any other supposed problems feel free to raise some that are real and actually matter for my inspection any time.

By the way Tassie was quite badly affected by the early-90s chuckout of master list ratings mentioned above; some players who were not all that inactive (a few years off during a period where there was very little chess here) ended up having their ratings scrubbed and coming back to chess to find they were suddenly unrated. Might explain why none of them stayed back very long, so thanks for nothing to whoever was responsible for that decision.


You must have your head in the sand.

Hmmm, I could say where yours is on rating issues but I won't. :lol:


Just look at the rating of the players in the recent Solomon Islands International in Honiara. Indeed why was this ACF rated at all?

Are you saying that these players are inaccurately rated? If so why?


Maybe ACF is trying to make the ACF Rating system more important to compete with FIDE Ratings which now have a good audit trail (which the ACF Rating system has not!)

Audit trail is an extremely minor issue unless there are actually suspicious ratings, especially when a player can approximately monitor their own likely ratings changes.

Here's a little "audit trail" for you, even if not quite of the kind you suggest. :lol: In your last thirteen tournaments you have performed below 2100 strength in 12 (the exception was an event in WA), including eight below 2000 strength (some well below). Are you going to tell us all those were one hour events?

You have not performed above 2150 across any ratings period since September 2006 which was also the last time your rating was that high. In this time you have played 135 games in ten ratings periods.

You have implied that your problem is playing in fast tournaments and said that you will quit them. You had ample opportunity to prove this point at the recent Major and failed. Your result was not at all bad, but even including the win over Garner you still performed at 2009, a figure differing by less than 30 points (hence not significantly) from your supposedly insulting ACF rating.

Having almost invariably failed to prove over the last few years in any kind of tournament you've played in that you are still a 2150+ strength player you have the temerity to launch all kinds of silly and spurious personal public attacks on the selection panel that didn't select you for the Championships, and do so after a tournament in which your result, while good, only further confirmed that that decision was correct under the stated criteria.

I would probably not be so explicit in spelling out all this stuff to anyone else in a similar situation but you are the one demanding transparency so as far as I'm concerned you can have what you want in spades. :lol:

Santa
18-01-2010, 02:31 AM
This personal attack is a little unseemly.:(


You have removed my comment from its context which was specifically that it was a response to santa's comment about inactive players. I was saying it was the only problem we have that arises as a result of keeping the ratings of inactive players as an alternative to chucking them out.


I don't remember actually doing that, if I did there are several Popes and Zoltan Kovacs should have been dropped too.

I've no doubt it would have been part of the new system though. A ratings system should be a system for calculating and publishing chessplayers' ratings.

It's not an archive.

People who do not play chess shouldn't have ratings. Chessplayers who stop playing chess are no longer chessplayers, they're just people who used to play chess. If they become chessplayers again, there's no reason to suppose that their skills were as they were before.

So far as the ratings system is concerned, Zoltan Kovacs could have been in the Balkans for the past 20 years and become a GM. Or, he could be an even weaker player than before, never having played a game of any sort in the meantime.

Should he return to chess, how to rate games against him?
The choices are these:
Enter him at his old rating. Really, is this any better than entering him at any other number in the range 800 to 2500?

If he scores zero points, clearly he's out of his league and wins against him don't mean much. Ignore his results. Probably, a player who has such a result will improve quickly and earn a real rating, drop out altogether or, as I need to do, find a field where he's competitive.

If he wins the tournament, winning all his games, then clearly he's a pretty big fish in this pond, and maybe we should rate player's games against him. Let's pretend the player actually drew one game. We can calculate a performance rating on that basis, and while noting he''s probably better than that, we have a credible number to begin with.

In other cases, calculate a performance rating and enter the player at this rating. Whether you then calculate ratings changes for the new player or leave then rating at that is a matter of taste, there are arguments both ways. The important points are that the player has a rating that can be used to calculate ratings changes against him, and that that rating is consistent with his results in that pool.

Myself, I would be very happy for my old rating to be dropped. I expect that if I play a couple of ordinary club tournaments that I will win some, lose some and get a rating consistent with others playing in Western Australia.





Audit trail is an extremely minor issue unless there are actually suspicious ratings, especially when a player can approximately monitor their own likely ratings changes.


Au contraire, he says and so exhausting his French vocabulary, an audit trail is critically important. It proves the evidence needed to establish that things are working as they should.

Brian_Jones
18-01-2010, 07:38 AM
I was not unhappy with my exclusion from the 2010 Australian Championship. (I would not have qualified even if all my proposed changes were accepted).

I was disappointed that my ACF Rating went down from 2020 to 1980 after three events (Honiara, Blayney and Rooty Hill) in which I scored 7/9, 4/6 and 4.5/6 respectively. Please can I see the rating chages of these three events in detail?

So, how do individual chess players see their own rating changes? Where is the audit trail? This is not a minor point as many players are suspicious of Glicko and want to see how their rating has changed.

Why is the ACF Rating system so secretive (no K factors published) ?

Why do ACF officers seem to defend the ACF (and the ACF Rating System) on this bulletin board. Why do they not take an even-handed approach.

CameronD
18-01-2010, 07:50 AM
One thing I would like to see is all rated tournament crosstables or games posted on the ACF website to be viewed by players. Any incorrect results (eg, clerical error or other) currently go unnoticed as players dont know the results sent to the rating officer(even FIDE report all game results).

I had a game not rated once because the tournament officer didnt want it rated, after much confusion and accusations the matter was dealt with. But I only noticed it due to the games columns number. What if the result is wrong, no one would ever know, and surely this most have happened under the current system??.

Bill Gletsos
18-01-2010, 09:36 AM
This personal attack is a little unseemly.:(

Originally Posted by Kevin Bonham
You have removed my comment from its context which was specifically that it was a response to santa's comment about inactive players. I was saying it was the only problem we have that arises as a result of keeping the ratings of inactive players as an alternative to chucking them out.
I don't remember actually doing that, if I did there are several Popes and Zoltan Kovacs should have been dropped too.Kevin's quote that you quoted was directed at Brian Jones not you.

Adamski
18-01-2010, 09:56 AM
BTW what happened to the Compaq computer?I saw somewhere that Santa's real Christian name is John. So it is therefore worth pointing out that Santa is not ACF National Ratings Officer Graham Saint (with whom I once worked at IBM GSA, and whom I most recently saw at the Chess Discount Sales shop in central Sydney).

I am pointing this out because one could assume (wrongly) that the Ratings Officer had lost / misplaced a certain Compaq PC...

Update: I see that Santa has now got his name in his sig line.

Bill Gletsos
18-01-2010, 10:18 AM
I saw somewhere that Santa's real Christian name is John. So it is therfore worth pointing out that Santa is not ACF National Ratings Officer Graham Saint (with whom I once worked at IBM GSA, and whom I most recently saw at the Chess Discount Sales shop in central Sydney).Santa is most definitely not Graham Saint.

I am pointing this out because one could assume (wrongly) that the Ratings Officer had lost / misplaced a certain Compaq PC...Replace the words lost/misplaced with kept and you would be closer to the mark.

Update: I see that Santa has now got his name in his sig line.Santa was the ACF National Ratings Officer from 1993 to mid 1996.

Kevin Bonham
18-01-2010, 01:09 PM
This personal attack is a little unseemly.:(

The bit you quoted was not a personal attack, it was simply a reply to what you actually said which was "I'd be happy to drop players ratings three (but you choose a number!) years after their last game. " I was saying this is unnecessary and what we have is in most regards a better alternative.

Whether you were actually responsible for the chucking of ratings from the master list is no concern to me; I am just saying that (i) whoever did it, it was silly and detrimental (ii) you are saying you have no problem with the concept even though in my practical experience it is a very bad idea.


I've no doubt it would have been part of the new system though. A ratings system should be a system for calculating and publishing chessplayers' ratings.

It's not an archive.

It can be both. Even old ratings have substantial predictive value especially if the player is an adult of middling years. We have a player here who took 13 years off and has basically picked up where he left off (lost a pile of points in his first two tournaments, got them back in the next two).


People who do not play chess shouldn't have ratings. Chessplayers who stop playing chess are no longer chessplayers, they're just people who used to play chess. If they become chessplayers again, there's no reason to suppose that their skills were as they were before.

No but there is reason to assume they could well be somewhere around that mark and that is what the current system does, but it adjusts them rapidly if evidence shows otherwise.

The fact is that if someone was once a GM or even a 2000+ they are not going to drop to sub-1000 standard unless they are over 100 years old or suffer major brain injury. An old rating might be hundreds of points out but is still some use.

There is something I sort-of agree about with you though - people who have been inactive for years should not be able to say they still "have" their old rating. It should be called an expired rating and marked as such in the master list.


Au contraire, he says and so exhausting his French vocabulary, an audit trail is critically important. It proves the evidence needed to establish that things are working as they should.

It only provides the evidence to test that the system is being applied as stated, not the evidence that it is a good system. The evidence to test that a system is good can be found by examining its predictive success. If a system is not being applied as it should then its predictive success will also suffer.

Brian_Jones
18-01-2010, 02:03 PM
It was rated with the approval of the ACF Council at the request of Oceania Zone President Gary Bekker.

This is a big porkie Bill.

Gary Bekker only gave agreement for the ACF Rating of the Solomon Islands International after you suggested it to him!

Out of ten players, there were only three Australian players and one PNG player. The other six were from the Solomon Islands.

Bill Gletsos
18-01-2010, 02:32 PM
This is a big porkie Bill.Please retract this false claim.


Gary Bekker only gave agreement for the ACF Rating of the Solomon Islands International after you suggested it to him!Before accusing people of telling lies you need to check the facts.

Gary Bekker in an email dated 9th August 2009 to the Solomon Islands Chess Federation suggested that if they were to submit all their major events to the ACF for rating the ACF would be happy to do it. You were copied on that email as was Gary Wastell.

That was the first I had heard about it.

I subsequently spoke to Gary Bekker about it and informed him that it was up to the ACF Council but that I had no problems in doing it..
The matter was raised as an item out of Gary Wastell's Presidents report to the October ACF Council meeting.
the ACF Council approved the rating of the Solomon Islands events.

Brian_Jones
18-01-2010, 03:00 PM
Gary Bekker in an email dated 9th August 2009 to the Solomon Islands Chess Federation suggested that if they were to submit all their major events to the ACF for rating the ACF would be happy to do it. .

This refers to the possible ACF Rating of Solomon Island events generally (eg their inter-club competition). This would avoid them having to create their own rating system. Have the SICF made this formal request to ACF? If so what is the date of this request?

My understanding is that the email did not specifically include International events held under the auspices of FIDE, particularly the first ever Solomon Islands International held in Spring 2009, which you spoke to Gary Bekker about specifically and received tacit approval.

So the Australians got nine (9) rated games - how did you treat the unrated players? I would have expected no more than four rated games.

Where is the audit trail?

Brian_Jones
18-01-2010, 03:02 PM
One thing I would like to see is all rated tournament crosstables or games posted on the ACF website to be viewed by players. Any incorrect results (eg, clerical error or other) currently go unnoticed as players dont know the results sent to the rating officer(even FIDE report all game results).

I had a game not rated once because the tournament officer didnt want it rated, after much confusion and accusations the matter was dealt with. But I only noticed it due to the games columns number. What if the result is wrong, no one would ever know, and surely this most have happened under the current system??.

We would all like to see the Audit Trail.

Maybe there are hundreds of mistakes not picked up! :evil:

Denis_Jessop
18-01-2010, 03:14 PM
This is a big porkie Bill.

Gary Bekker only gave agreement for the ACF Rating of the Solomon Islands International after you suggested it to him!

Out of ten players, there were only three Australian players and one PNG player. The other six were from the Solomon Islands.

Talking about porkies you are rather disingenuous about the "PNG" player who appears on the ACF rating list as an ACT player and hasn't lived in PNG for aeons.

DJ

Kevin Bonham
18-01-2010, 03:19 PM
This is not a minor point as many players are suspicious of Glicko and want to see how their rating has changed.

No more than were suspicious even when they could check it for themselves.


Why is the ACF Rating system so secretive (no K factors published) ?

Glicko2 does not use k-factors.


Why do ACF officers seem to defend the ACF (and the ACF Rating System) on this bulletin board. Why do they not take an even-handed approach.

Because this bulletin board is a place where we are subject to often unfactual, misinformed or otherwise silly attacks which we choose to rebut.

If an even-handed approach involves accepting half of every pile of proposed ordure that is shovelled our way then I can't see the merits of it personally. :lol:

As for whether you would have qualified under your own proposed changes, one thing you do not make clear is whether you think the equivalent proficiency rule should stay.

The problem with applying the equivalent proficiency rule to FIDE ratings would be that a FIDE 2150 player might only be an ACF 1900s player with a FIDE rating acheived from a few tournaments some time ago when the floor was higher or they might be that good on an ACF ratings basis as well. So what does equivalent proficiency to FIDE 2150 even mean?

Bill Gletsos
18-01-2010, 03:22 PM
This refers to the possible ACF Rating of Solomon Island events generally (eg their inter-club competition).The email specifically referred to major events.
Are you suggesting the Solomon Islands International was not a major event?

My understanding is that the email did not specifically include International events held under the auspices of FIDE, particularly the first ever Solomon Islands International held in Spring 2009, which you spoke to Gary Bekker about specifically and received tacit approval.Gary Bekker's email specifically mentioned major events.
The Solomon Islands International was a major SICF event.
Naturally when I discussed the rating of SICF events in general with Gary it included the rating of the SI International.

So the Australians got nine (9) rated games - how did you treat the unrated players?The same way all unrated players are treated.
It has been explained on here in the past that games against unrated players are rated if the system can assign a rating to the player based on games they played in the rating period.

I would have expected no more than four rated games.Then your expectation is wrong.

Bill Gletsos
18-01-2010, 03:32 PM
I don't remember actually doing that, if I did there are several Popes and Zoltan Kovacs should have been dropped too.

I've no doubt it would have been part of the new system though.No it was part of the ACF system that was actually in place, not part of any new system.
You deleted over 8,000 names from the ACF master list between December 1992 and April 1993.

A ratings system should be a system for calculating and publishing chessplayers' ratings.

It's not an archive.

People who do not play chess shouldn't have ratings. Chessplayers who stop playing chess are no longer chessplayers, they're just people who used to play chess. If they become chessplayers again, there's no reason to suppose that their skills were as they were before.The ACF Council back in the mid 90's disagreed with you.

So far as the ratings system is concerned, Zoltan Kovacs could have been in the Balkans for the past 20 years and become a GM. Or, he could be an even weaker player than before, never having played a game of any sort in the meantime.

Should he return to chess, how to rate games against him?Glicko/Glicko2 easily handles this.

The choices are these:
Enter him at his old rating. Really, is this any better than entering him at any other number in the range 800 to 2500?

If he scores zero points, clearly he's out of his league and wins against him don't mean much. Ignore his results. Probably, a player who has such a result will improve quickly and earn a real rating, drop out altogether or, as I need to do, find a field where he's competitive.

If he wins the tournament, winning all his games, then clearly he's a pretty big fish in this pond, and maybe we should rate player's games against him. Let's pretend the player actually drew one game. We can calculate a performance rating on that basis, and while noting he''s probably better than that, we have a credible number to begin with.

In other cases, calculate a performance rating and enter the player at this rating. Whether you then calculate ratings changes for the new player or leave then rating at that is a matter of taste, there are arguments both ways. The important points are that the player has a rating that can be used to calculate ratings changes against him, and that that rating is consistent with his results in that pool.

Myself, I would be very happy for my old rating to be dropped. I expect that if I play a couple of ordinary club tournaments that I will win some, lose some and get a rating consistent with others playing in Western Australia.Glicko/Glicko2 handles this scenario.

Au contraire, he says and so exhausting his French vocabulary, an audit trail is critically important. It proves the evidence needed to establish that things are working as they should.When you were ACF National Ratings Officer it was impossible for a player who played either a provisionally rated player or played a player listed as 'xg' to be able to calculate their new ratings even though the Elo formula was published. So where was the audit trail then.
Back under the old ACF Elo system the ratings of provisionally rated players and 'xg' rated players would fluctuate during a single rating period such that if rated player A played provisionally rated player C in tournament #1 and rated player B played provisionally rated player C in tournament #2 the rating used for provisionally rated player C was different for player A and player B.
In fact if Player C was an unrated player new to the rating list it was even possible for a players A and B to play player C in the same tournament and for player C to be treated as unrated against player A and with a rating of r1 against B if a rating for C was processed in that tournament after A and before B. If player C then played players D and A in a subsequently rated tournament in the same rating period it was possible for player C to be treated with a rating of r1 against D and a rating of r2 against A if the new rating for C in this tournament was calculated after D but before A.

This made an audit trail impossible.

Basil
18-01-2010, 03:48 PM
I'm no expert, but it appears this split should have gone straight to 'Bozo Debunking Corner'.

bockman
18-01-2010, 05:30 PM
here are the bare results of the SI international:




1 2 3 4 5 6 7 8 9 0
1 Press,Shaun 2076 PNG * 1 1 1 1 1 1 7.5
2 Jones,Brian 2065 AUS * 1 1 1 1 1 7.0
3 Jones,Lee 2117 AUS 0 * 1 1 0 1 1 1 6.0
4 Aguilar,Fernando SOL 0 * 1 0 1 1 1 1 6.0
5 Stead,Kerry 2087 AUS 0 0 * 1 1 1 1 1 6.0
6 Tangaibasa,Brandon SOL 0 0 1 1 0 * 1 0 1 1 5.0
7 Tesua,Scarden SOL 0 0 0 0 0 * 1 0 1 2.5
8 Tepuke,Price SOL 0 0 0 0 0 1 0 * 0 1 2.0
9 Tuata,Takika SOL 0 0 0 0 0 0 1 1 * 0 2.0
10 Maruia,Budds SOL 0 0 0 0 0 0 0 0 1 * 1.0

Source: http://www.auschess.org.au/oceania/solomons/invitational.htm



note. the ratings are fide ratings. i don't have the acf ratings on hand.

if 9 games were acf-rated from this tournament
then what ratings were used for players 4, 6-10??

fernando had played in australia before and did ok.
http://chess-results.com/tnr23103.aspx?art=9&lan=1&fed=SOL&m=-1&wi=1000&snr=73

bill can you verify what ratings were given to 4, 6-10??

please...

Garvinator
18-01-2010, 05:37 PM
I had a game not rated once because the tournament officer didnt want it rated, after much confusion and accusations the matter was dealt with. But I only noticed it due to the games columns number. What if the result is wrong, no one would ever know, and surely this most have happened under the current system??.Please give more information on this matter?

Patrick Byrom
18-01-2010, 06:14 PM
One thing I would like to see is all rated tournament crosstables or games posted on the ACF website to be viewed by players. Any incorrect results (eg, clerical error or other) currently go unnoticed as players dont know the results sent to the rating officer(even FIDE report all game results).

Queenslanders can check their rating changes for the last period here:
http://www.southsidejuniorchessclub.org/Sept-Dec%20_09_Unofficial_ACF_Rating_Changes.htm

And there should be a more official site available for this ratings period (when I have some events to rate).

I would have thought that all tournament crosstables would be available on the net, either on a state or club website.

bockman
18-01-2010, 06:32 PM
here are the bare results of the SI international:




1 2 3 4 5 6 7 8 9 0
1 Press,Shaun 2076 PNG * 1 1 1 1 1 1 7.5
2 Jones,Brian 2065 AUS * 1 1 1 1 1 7.0
3 Jones,Lee 2117 AUS 0 * 1 1 0 1 1 1 6.0
4 Aguilar,Fernando SOL 0 * 1 0 1 1 1 1 6.0
5 Stead,Kerry 2087 AUS 0 0 * 1 1 1 1 1 6.0
6 Tangaibasa,Brandon SOL 0 0 1 1 0 * 1 0 1 1 5.0
7 Tesua,Scarden SOL 0 0 0 0 0 * 1 0 1 2.5
8 Tepuke,Price SOL 0 0 0 0 0 1 0 * 0 1 2.0
9 Tuata,Takika SOL 0 0 0 0 0 0 1 1 * 0 2.0
10 Maruia,Budds SOL 0 0 0 0 0 0 0 0 1 * 1.0

Source: http://www.auschess.org.au/oceania/solomons/invitational.htm



note. the ratings are fide ratings. i don't have the acf ratings on hand.

if 9 games were acf-rated from this tournament
then what ratings were used for players 4, 6-10??

fernando had played in australia before and did ok.
http://chess-results.com/tnr23103.aspx?art=9&lan=1&fed=SOL&m=-1&wi=1000&snr=73

bill can you verify what ratings were given to 4, 6-10??

please...

the decision by the system in the example above
cost some players to lose 50+ points.

isn't it entirely reasonable to ask for a simple print out of the calculation just for this event?

Bill Gletsos
18-01-2010, 07:37 PM
here are the bare results of the SI international:




1 2 3 4 5 6 7 8 9 0
1 Press,Shaun 2076 PNG * 1 1 1 1 1 1 7.5
2 Jones,Brian 2065 AUS * 1 1 1 1 1 7.0
3 Jones,Lee 2117 AUS 0 * 1 1 0 1 1 1 6.0
4 Aguilar,Fernando SOL 0 * 1 0 1 1 1 1 6.0
5 Stead,Kerry 2087 AUS 0 0 * 1 1 1 1 1 6.0
6 Tangaibasa,Brandon SOL 0 0 1 1 0 * 1 0 1 1 5.0
7 Tesua,Scarden SOL 0 0 0 0 0 * 1 0 1 2.5
8 Tepuke,Price SOL 0 0 0 0 0 1 0 * 0 1 2.0
9 Tuata,Takika SOL 0 0 0 0 0 0 1 1 * 0 2.0
10 Maruia,Budds SOL 0 0 0 0 0 0 0 0 1 * 1.0

Source: http://www.auschess.org.au/oceania/solomons/invitational.htm



note. the ratings are fide ratings. i don't have the acf ratings on hand.

if 9 games were acf-rated from this tournament
then what ratings were used for players 4, 6-10??Internal ratings calculated by the system based on the players results against their 9 opponents.

fernando had played in australia before and did ok.
http://chess-results.com/tnr23103.aspx?art=9&lan=1&fed=SOL&m=-1&wi=1000&snr=73His performance rating in his 8 games at the Zonal was around 1580.

bill can you verify what ratings were given to 4, 6-10??Lee, in the case of those players you can get a very good indication by looking at their ratings on the ACF December 2009 list.

Denis_Jessop
18-01-2010, 07:37 PM
It doesn't need much nous to realise why some players might have lost points in the event. By the simple expedient of looking up the ACF ratings on the ACF website one finds the following current ratings of the 4 Australians:

Press 1790
Jones, B 1981
Jones, L 1897
Stead 1981.

Then look at the results of the Tournament.

Note also the vast overrating of those players on the FIDE list.

DJ

Bill Gletsos
18-01-2010, 07:43 PM
the decision by the system in the example above
cost some players to lose 50+ points.

isn't it entirely reasonable to ask for a simple print out of the calculation just for this event?It isnt possible since tournaments are not rated on a tournament by tournament or even game by game basis.
The Glicko formula uses the results of all games played by a player in the rating period.

Lee your performance rating in the Solomon Islands International was around 1870 whilst your performance in the Blayney Open was a dismal 1480.
Your dads performance rating in the Solomon Islands International was around 1960 whilst his performance in the Blayney Open was a dismal 1490.

Garvinator
18-01-2010, 07:59 PM
I really do not think it matters how these questions are answered. I think Brian's only agenda here is to criticise any ACF decision. It would not matter what decision the ACF made on any of these matters. Brian would be criticising the decisions regardless.

Kevin Bonham
18-01-2010, 08:27 PM
It doesn't need much nous to realise why some players might have lost points in the event. By the simple expedient of looking up the ACF ratings on the ACF website one finds the following current ratings of the 4 Australians:

Press 1790
Jones, B 1981
Jones, L 1897
Stead 1981.

And furthermore at the time of the event, Shaun Press's rating had crashed to 1686. He actually gained 104 points in the December ratings period and appeared in the Top Improvers list.

Brian drew with Press who at the time was rated way below him. He drew with L Jones and Stead who at the time had ratings almost identical to his. So from the rated players he clearly deserved to lose points.

Everybody else he played was unrated. He drew with Aguilar who broke even against the Australian players, but the Australian players included Press who is weaker than B Jones.

Brian got 5/5 against the other locals but given that Stead and Press also clean-sheeted them with only L Jones dropping 1.5 points against them, he shouldn't expect too much out of that.

Overall Brian can't complain if he should lose ratings points for this tournament because he drew with a player rated over 300 ACF points below him and did nothing else in the event that goes anywhere near cancelling that sort of points loss out.

Hey Garvin, I see you sigfiled the Compaq computer question; I was thinking just last night that someone would probably do that.

Basil
18-01-2010, 08:34 PM
Overall Brian can't complain if he should lose ratings points for this tournament because he drew with a player rated over 300 ACF points below him and did nothing else in the event that goes anywhere near cancelling that sort of points loss out.
Well der - this is so elementary, even an idiot like Captain Underpants understands it. I demand that this thread be added to the 'Bozo Debunking Corner' thread.


Hey Garvin, I see you sigfiled the Compaq computer question; I was thinking just last night that someone would probably do that.
But mine is in red!

Garvinator
18-01-2010, 08:46 PM
Hey Garvin, I see you sigfiled the Compaq computer question; I was thinking just last night that someone would probably do that.I did it a few days ago and is working well. Looks good too. It will stay there until the question is actually answered on this board.

It has done me no harm and seems to be working out quite well, so I recommend more people give it a whirl to. Remember, it is only kinky the first time ;).

Captain Underpants has decided to be a bit different and gone for the red shade.

Patrick Byrom
18-01-2010, 10:42 PM
the decision by the system in the example above
cost some players to lose 50+ points.

isn't it entirely reasonable to ask for a simple print out of the calculation just for this event?

Running my Glicko1 program for this event, and replacing the ratings for the previously unrated players with their performance ratings for a few cycles, gave the following results:


Name New Rating Old Rating Change
Press, Shaun P 1803 1686 117
Jones, Brian A 2019 2021 -2
Jones, Lee R 1971 2007 -36
Aguilar, Fernando 1913 1861 52
Stead, Kerry 1998 2018 -20
Tangalbasa, Brandon 1766 1860 -94
Tesua, Scarden 1421 1423 -2
Tepuke, Price 1336 1544 -208
Tuata, Takika 1500 1544 -44
Maruia, Budds unr unrated

Some players (such as Aguilar) have other results in the system, but the new ratings for most of the previously unrated players seem fairly close to the ones on the next rating list.

The ELO system would give very similar results (I suspect), as long as you use individual K-factors, equal to one third of the individual's RD.

CameronD
18-01-2010, 11:08 PM
Please give more information on this matter?

This was a long time ago and there is no point naming names and giving a more detailed account.

CameronD
18-01-2010, 11:17 PM
Queenslanders can check their rating changes for the last period here:
http://www.southsidejuniorchessclub.org/Sept-Dec%20_09_Unofficial_ACF_Rating_Changes.htm

And there should be a more official site available for this ratings period (when I have some events to rate).

I would have thought that all tournament crosstables would be available on the net, either on a state or club website.

Id like to see the records the ACF received instead of the orgalnizer website, though your result format would satisfy me. Putting this on the CAQ website as an official CAQ document would impress most players.

Brian_Jones
19-01-2010, 08:25 AM
It isnt possible since tournaments are not rated on a tournament by tournament or even game by game basis.
The Glicko formula uses the results of all games played by a player in the rating period.

This is the real problem with the ACF Rating System.

I know what happened to the Compaq computer but I don't know how the ACF Rating System works because it uses Glicko and has no audit trail.

Also, I still don't know how many points I lost at Blayney (60+10). But I do know it is easy to lose to lowly-rated juniors at 60+10 but hard to find opponents with compensating high ratings. So maybe Glicko puts people off from playing chess and they become inactive. This is an good example of negative marketing!

Bill Gletsos
19-01-2010, 09:13 AM
This is the real problem with the ACF Rating System.

I know what happened to the Compaq computer...Do you.
Then why not tell us what happened.

but I don't know how the ACF Rating System works because it uses Glicko and has no audit trail.The lack of an audit trail should not stop you from understanding how it works.

Also, I still don't know how many points I lost at Blayney (60+10). But I do know it is easy to lose to lowly-rated juniors at 60+10 but hard to find opponents with compensating high ratings.This just looks like an excuse from a person who played badly. Although you were the second seed there were a number of other strong players such as Vladimir Smirnov, Paul Broekhuyse, Lee Jones and Angelito Camer.

So maybe Glicko puts people off from playing chess and they become inactive. This is an good example of negative marketing!Actually it is a good example of you sprouting rubbish.

Spiny Norman
19-01-2010, 11:15 AM
One thing I would like to see is all rated tournament crosstables or games posted on the ACF website to be viewed by players. Any incorrect results (eg, clerical error or other) currently go unnoticed as players dont know the results sent to the rating officer(even FIDE report all game results).
I had a game not rated once because the tournament officer didnt want it rated, after much confusion and accusations the matter was dealt with. But I only noticed it due to the games columns number. What if the result is wrong, no one would ever know, and surely this most have happened under the current system??.
Probably lots of players are relatively sympathetic to this idea ... BUT (and this is a BIG but) ... who is going to volunteer to:

1) learn how the rating systems works (in detail); and
2) volunteer to field the inevitable flood of questions that such 'audit trail disclosure' might generate?

I like the idea in principle, but I am dead against loading up the current ratings officers with a whole lot more work fielding (mostly, inevitably) misguided objections and questions.

Brian_Jones
19-01-2010, 12:03 PM
Only a snail would think about a manual system! :)

Take a look at the Australian GP website to see the GP scoring.
It uses SP files as input (same as the ACF Rating system).
See http://www.chessaustralia.com.au/index.cfm?site=grandprix

Also, take a look at the FIDE website to see the FIDE Rating audit trail.
For Australian events it uses the files submitted by Greg Canfell.
http://ratings.fide.com/tournament_list.phtml?moder=ev_code&country=AUS

All open for public inspection! :eek:

Spiny Norman
19-01-2010, 12:14 PM
Only a snail would think about a manual system! :)
<snip>
All open for public inspection! :eek:
How is the response to questions about this audit trail automated?

Kevin Bonham
19-01-2010, 02:04 PM
Also, I still don't know how many points I lost at Blayney (60+10). But I do know it is easy to lose to lowly-rated juniors at 60+10 but hard to find opponents with compensating high ratings. So maybe Glicko puts people off from playing chess and they become inactive. This is an good example of negative marketing!

This has nothing to do with Glicko. It is easy to lose to low-rated juniors at 60+10 in any system no matter what the rating system, but in ELO the low-rated juniors are even lower rated and furthermore the system does not cut you any slack by flagging their ratings as less reliable if they have not played much.

The problem of it being hard to find higher-rated players is also something that has little to do with what system you use.

Brian, you should try living down here where there are no compensating high ratings at all. That might give you a bit of perspective re your complaint about difficulty finding compensating higher-rated players. You have it easy compared to those in many parts of Australia.

Desmond
19-01-2010, 02:23 PM
Perhaps we should only rate games versus juniors when they lose.

Brian_Jones
19-01-2010, 03:02 PM
How is the response to questions about this audit trail automated?

It is self-service - the player can see each game in each event in each rating period with the individual gain or loss (and the cumulative gain/loss).

Questions then become fewer because players can see the answers themselves! :D

I now get very few GP questions. FIDE only answer FIDE Rating questions from the authorised person ie Greg Canfell.

No more black holes!

Spiny Norman
19-01-2010, 03:11 PM
It is self-service - the player can see each game in each event in each rating period with the individual gain or loss (and the cumulative gain/loss).
Questions then become fewer because players can see the answers themselves! :D
What you seem to mean is this: that if more information is provided, there will be less questions asked that have to be answered ad-hoc by someone.

That may be true, but I don't share your optimism, as in other areas where I have worked I have observed that the provision of additional information can be very double-edged and can actually generate more questions. In technical fields, writing and publishing Frequently Asked Questions (FAQs) and their answers is commonplace and I highly recommend this.

However this scenario presumably requires completely opening the kimono and publishing all the detail of specific events, specific games, ratings used in calculations (including interim ratings used), how rapidly-improving players results are calculated asymmetrically when playing stably-rated players, the formulas involved in ACF-modified-Glicko-2, and so on. And I think that all of THAT will possibly generate HEAPS of questions directed to existing ratings officers ... and in turn, THAT is what I think ought to be avoided

Brian_Jones
19-01-2010, 03:13 PM
Brian, you should try living down here where there are no compensating high ratings at all. That might give you a bit of perspective re your complaint about difficulty finding compensating higher-rated players. You have it easy compared to those in many parts of Australia.

Yes, Tony Dowden used to be a 2300 player when I met him in Christchurch. I am surprised he is still motivated to play locally!

Brian_Jones
19-01-2010, 03:19 PM
What you seem to mean is this: that if more information is provided, there will be less questions asked that have to be answered ad-hoc by someone.

That may be true, but I don't share your optimism, as in other areas where I have worked I have observed that the provision of additional information can be very double-edged and can actually generate more questions. In technical fields, writing and publishing Frequently Asked Questions (FAQs) and their answers is commonplace and I highly recommend this.

However this scenario presumably requires completely opening the kimono and publishing all the detail of specific events, specific games, ratings used in calculations (including interim ratings used), how rapidly-improving players results are calculated asymmetrically when playing stably-rated players, the formulas involved in ACF-modified-Glicko-2, and so on. And I think that all of THAT will possibly generate HEAPS of questions directed to existing ratings officers ... and in turn, THAT is what I think ought to be avoided

Which is exactly why I am recommending we go back to a less mathematically accurate system such as Elo. We do not need this degree of accuracy - we need a system that is open and understood by everybody. We do not need a black hole.

Did you know the English Chess Federation (ECF) still uses the simple formula of plus/minus 50 points and the Correspondence Chess Association of Australia (CCLA) uses plus/minus 100 points.

Bill Gletsos
19-01-2010, 03:41 PM
Yes, Tony Dowden used to be a 2300 player when I met him in Christchurch. I am surprised he is still motivated to play locally!When was he a 2300 player, given Tony's highest FIDE rating appears to have been 2220 back in 1990.

Bill Gletsos
19-01-2010, 03:44 PM
Did you know the English Chess Federation (ECF) still uses the simple formula of plus/minus 50 points and the Correspondence Chess Association of Australia (CCLA) uses plus/minus 100 points.Did you know that the ECF rates G60 + increment in their standard rating system just like the ACF and USCF.

Basil
19-01-2010, 04:06 PM
What you seem to mean is this: that if more information is provided, there will be less questions asked that have to be answered ad-hoc by someone.

That may be true, but I don't share your optimism, as in other areas where I have worked I have observed that the provision of additional information can be very double-edged and can actually generate more questions. In technical fields, writing and publishing Frequently Asked Questions (FAQs) and their answers is commonplace and I highly recommend this.

However this scenario presumably requires completely opening the kimono and publishing all the detail of specific events, specific games, ratings used in calculations (including interim ratings used), how rapidly-improving players results are calculated asymmetrically when playing stably-rated players, the formulas involved in ACF-modified-Glicko-2, and so on. And I think that all of THAT will possibly generate HEAPS of questions directed to existing ratings officers ... and in turn, THAT is what I think ought to be avoided
Excellent post. Pointies begone!

Denis_Jessop
19-01-2010, 04:46 PM
Which is exactly why I am recommending we go back to a less mathematically accurate system such as Elo. We do not need this degree of accuracy - we need a system that is open and understood by everybody. We do not need a black hole.

Did you know the English Chess Federation (ECF) still uses the simple formula of plus/minus 50 points and the Correspondence Chess Association of Australia (CCLA) uses plus/minus 100 points.

I'm not sure what the ratings officers' view is about mathematical accuracy but I see a big advantage of Glicko over ELO being its responsiveness so that changes in playing strength are picked up quicker including rapidly improving juniors' improvement. The system better reflects players' current playing strength.

Kevin's point also applies in the ACT where we have a few high-rated players, most of whom rarely play, and then bunnies of varying quality. A good example of the problem is Vlad who was hovering around 2000 here but whose rating and, no doubt, his playing, improved dramatically when he moved to Sydney so that he is now of IM strength. None of this has anything to do with the rating system but with the strength of the playing pool and the fact that here the same few players always play each other.

DJ

bockman
19-01-2010, 05:31 PM
[QUOTE=bockman]here are the bare results of the SI international:




1 2 3 4 5 6 7 8 9 0
1 Press,Shaun 2076 PNG * 1 1 1 1 1 1 7.5
2 Jones,Brian 2065 AUS * 1 1 1 1 1 7.0
3 Jones,Lee 2117 AUS 0 * 1 1 0 1 1 1 6.0
4 Aguilar,Fernando SOL 0 * 1 0 1 1 1 1 6.0
5 Stead,Kerry 2087 AUS 0 0 * 1 1 1 1 1 6.0
6 Tangaibasa,Brandon SOL 0 0 1 1 0 * 1 0 1 1 5.0
7 Tesua,Scarden SOL 0 0 0 0 0 * 1 0 1 2.5
8 Tepuke,Price SOL 0 0 0 0 0 1 0 * 0 1 2.0
9 Tuata,Takika SOL 0 0 0 0 0 0 1 1 * 0 2.0
10 Maruia,Budds SOL 0 0 0 0 0 0 0 0 1 * 1.0

Source: http://www.auschess.org.au/oceania/solomons/invitational.htm



for the sake of completeness
here are the values in the acf master list for the missing players.



id name state sex junior_flag rating
0003252 Tangalbasa, Brandon OS m N 1760
0003160 Aguilar, Fernando OS m N 1684
0003230 Tuata, Takika OS m N 1525
0003215 Tesua, Scarden OS m N 1445
0003226 Tepuke, Price OS m N 1340
0003241 Marula, Budds OS m N 1146

2011012 Jones, Brian A NSW m N 1981
2050690 Stead, Kerry VIC m N 1981
2044155 Jones, Lee R NSW m N 1897
1272705 Press, Shaun P ACT m N 1790


btw. the spellings of tangaibasa and maruia are different on the website and in the acf master list.
the original spelling i've seen were tangaibasa and maruia.
maybe there is a mistake with an l but it should be an i??

bockman
19-01-2010, 05:53 PM
Queenslanders can check their rating changes for the last period here:
http://www.southsidejuniorchessclub.org/Sept-Dec%20_09_Unofficial_ACF_Rating_Changes.htm

And there should be a more official site available for this ratings period (when I have some events to rate).

I would have thought that all tournament crosstables would be available on the net, either on a state or club website.

thanks for posting this data.



No Name New Rating Old Rating Change Rated Score Av Opp Approx RD
1. Tony WELLER 1756 1754 2 3.5/4 1245 60
2. Derek J DONOGHUE 1518 1739 -221 2.5/4 1249 300
3. Vasile BUCIU 1344 1336 8 2.0/4 1349 140
4. Jim PETRIE 1115 1117 -2 1.0/4 1404 60
5. Jim ROGERS 804 787 17 1.0/4 1487 60


so donoghue returns after inactivity
draws with weller and loses to buciu.
he loses 200 points for 1 game because his RD is 300.

any system where you lose 200 points in 1 game is interesting.
according to elo, 200 points was his definition of an entire class.
so he drops from class B to C in 1 game.
ok. fair enough.

---

'Class B and higher is generally considered extremely competitive and the USCF establishes a rating floor of each player's peak rating minus 200 rating points. For instance, once someone has reached a rating of 1600, they can never fall below 1400 for rating and competition purposes. This is to protect the integrity of big tournaments and prevent sandbagging. A floor also has the effect of artificially inflating the rating of someone whose abilities have declined more than 200 points.' - Source: http://en.wikipedia.org/wiki/Elo_rating

Does the ACF-rating have any sandbagging protection :)

Patrick Byrom
19-01-2010, 05:54 PM
Which is exactly why I am recommending we go back to a less mathematically accurate system such as Elo. We do not need this degree of accuracy - we need a system that is open and understood by everybody. We do not need a black hole.

Do you mean Elo with a single k-factor, several k-factors (as FIDE use, I believe), or individual k-factors (which would make it similar to Glicko1)?

I have found that my program, which was written using only publicly available information, can produce ratings which agree fairly closely with the official ones. As you can see by comparing the results I posted earlier for the ratings of the unrated Solomon Island players with their December ACF ones. So why do you describe the system as a black hole?

bockman
19-01-2010, 06:27 PM
on his website glickman says:
'the glicko-2 system works best when the number of games in a rating period is moderate to large, say an average of at least 10-15 games per player in a rating period.'

when you feed the dec 2009 active list into excel you get an average of 4 games per person.

does this mean australia is not suited to glicko-2 since
we are not active enough chess players for it to be effective??

the qld veterns event is an example.

Source:
http://www.glicko.net/glicko/glicko2.doc/example.html
http://math.bu.edu/people/mg/ratings/cl-article.pdf

Patrick Byrom
19-01-2010, 06:42 PM
so donoghue returns after inactivity
draws with weller and loses to buciu.
he loses 200 points for 1 game because his RD is 300.
any system where you lose 200 points in 1 game is interesting.
according to elo, 200 points was his definition of an entire class.
so he drops from class B to C in 1 game.
ok. fair enough.

As he drew with players rated about 800 and 1100, his new rating doesn't seem too unreasonable. Even if he had beaten them, but lost to the 1300 player, it would still imply that he wasn't 1700. And he can get the points back if he plays well in his next event.

There are huge class prizes in USCF events, so it has different priorities to the ACF.

A limit on rating drops in one event might be a reasonable change, but this can be done in Glicko as easily as in Elo.


No Name Loc Total 1 2 3 4 5
1 WELLER, Tony 1754 3.5 * D W W W
2 DONOGHUE, Derek J 1739 2.5 D * W D D
3 BUCIU, Vasile 1336 2 L L * W W
4 PETRIE, Jim 1117 1 L D L * D
5 ROGERS, Jim 787 1 L D L D *

CameronD
19-01-2010, 07:21 PM
Probably lots of players are relatively sympathetic to this idea ... BUT (and this is a BIG but) ... who is going to volunteer to:

1) learn how the rating systems works (in detail); and
2) volunteer to field the inevitable flood of questions that such 'audit trail disclosure' might generate?

I like the idea in principle, but I am dead against loading up the current ratings officers with a whole lot more work fielding (mostly, inevitably) misguided objections and questions.

Just to make it clear, I dont care about a ratings audit trail, just a list of rated results so that player can check their relibility.

eg

Chesschat tournament
CameronD 2.5/7
Adamski 5/7

Thats all thats required for each rated tournament. Im sure most players would check their results each quarter. No mention of ratings or ratings change or anything.

Kevin Bonham
19-01-2010, 07:30 PM
'Class B and higher is generally considered extremely competitive and the USCF establishes a rating floor of each player's peak rating minus 200 rating points. For instance, once someone has reached a rating of 1600, they can never fall below 1400 for rating and competition purposes. This is to protect the integrity of big tournaments and prevent sandbagging. A floor also has the effect of artificially inflating the rating of someone whose abilities have declined more than 200 points.' - Source: http://en.wikipedia.org/wiki/Elo_rating

Does the ACF-rating have any sandbagging protection :)

No, because we do not need it, because our ratings prizes are not large enough to justify the effort of sandbagging. The sandbagging problem arises from ridiculous events like the so-called World Open that have ratings prizes in the ballpark of 10,000 dollars.

Bill Gletsos
19-01-2010, 08:29 PM
for the sake of completeness
here are the values in the acf master list for the missing players.Actually it is not that complete because for the non Solomon Island players you are showing their post tournament December ACF ratings instead of their pre-tournament September ACF ratings.

Bill Gletsos
19-01-2010, 08:48 PM
on his website glickman says:
'the glicko-2 system works best when the number of games in a rating period is moderate to large, say an average of at least 10-15 games per player in a rating period.'The number of games that it works best for is dependant on a number of factors. As Glickman said in an email you have to experiment with the various parameters to determine what works best in your circumstances.
This is especially true with the Glicko2 system.

when you feed the dec 2009 active list into excel you get an average of 4 games per person.Your maths are wrong, the average is a little over 5.

Bill Gletsos
19-01-2010, 08:54 PM
thanks for posting this data.



No Name New Rating Old Rating Change Rated Score Av Opp Approx RD
1. Tony WELLER 1756 1754 2 3.5/4 1245 60
2. Derek J DONOGHUE 1518 1739 -221 2.5/4 1249 300
3. Vasile BUCIU 1344 1336 8 2.0/4 1349 140
4. Jim PETRIE 1115 1117 -2 1.0/4 1404 60
5. Jim ROGERS 804 787 17 1.0/4 1487 60


so donoghue returns after inactivity
draws with weller and loses to buciu.
he loses 200 points for 1 game because his RD is 300.

any system where you lose 200 points in 1 game is interesting.
according to elo, 200 points was his definition of an entire class.
so he drops from class B to C in 1 game.
ok. fair enough.Donoghue had not played a rated game since mid 1989, so had been inactive for over 20 years. He had a performance rating under 1400.

Kevin Bonham
19-01-2010, 08:57 PM
on his website glickman says:
'the glicko-2 system works best when the number of games in a rating period is moderate to large, say an average of at least 10-15 games per player in a rating period.'

when you feed the dec 2009 active list into excel you get an average of 4 games per person.

I wonder if the first figure is meant to be recommending 10-15 games per player who actually plays in a given period.

Is the second figure (as corrected by Bill) the average for all players on the active list including the zeros, or the average for all players who have played in this period?

Bill Gletsos
19-01-2010, 09:06 PM
I wonder if the first figure is meant to be recommending 10-15 games per player who actually plays in a given period.

Is the second figure (as corrected by Bill) the average for all players on the active list including the zeros, or the average for all players who have played in this period?The figure I provided is the average for all players who actually played in the December period.
When doing the calculation it is important to take into account players with xg ratings in this period and yg ratings in the last period where x > y.

Denis_Jessop
19-01-2010, 09:30 PM
Donoghue had not played a rated game since mid 1989, so had been inactive for over 20 years. He had a performance rating under 1400.

This is precisely the kind of thing I had in mind (in an extreme example) when referring to the responsiveness advantage of Glicko. If a player rated 2000 takes a 20-year break from chess and returns with a performance rating of 1500 he deserves to lose a heap of rating points as he is no longer anywhere near 2000. It may dent his pride but rating points are based on performance, not ego.

DJ

michael.mcguirk
19-01-2010, 09:46 PM
Except doesn't that mean that those rating points get transferred onto other peoples ratings, giving them an artificially higher rating?

Kevin Bonham
19-01-2010, 09:56 PM
Except doesn't that mean that those rating points get transferred onto other peoples ratings, giving them an artificially higher rating?

Not to a great degree. That potential aspect is cushioned by two things:

(i) The use of intermediate ratings, meaning that the player who has performed way below their old rating is treated as significantly weaker than their old rating indicated.

(ii) The differing RDs of the players. The returning player's rating is very sensitive to results against established players. The established players' ratings are not very sensitive to results against players whose ratings are clearly unreliable.

Glicko is not zero-sum. Just because a player loses points doesn't mean others get those points in equal measure, and vice versa.

Bill Gletsos
19-01-2010, 09:56 PM
Except doesn't that mean that those rating points get transferred onto other peoples ratings, giving them an artificially higher rating?No.
Unlike Elo, Glicko is not zero sum.

michael.mcguirk
20-01-2010, 11:59 AM
Fair enough! Thanks for that :)

whatteaux
20-01-2010, 06:48 PM
This is precisely the kind of thing I had in mind (in an extreme example) when referring to the responsiveness advantage of Glicko. If a player rated 2000 takes a 20-year break from chess and returns with a performance rating of 1500 he deserves to lose a heap of rating points as he is no longer anywhere near 2000. It may dent his pride but rating points are based on performance, not ego.

DJ

I agree entirely. After an 18-year break, my rating went from 1801 to 1507 (performance rating of 1450, I'm reliably informed) after a 5-game tournament and I'm delighted. After that long, I've completely forgotten my opening repertoire and tend to run out of steam after about 25 moves, so 1507 is a much more accurate reflection of my rustiness than 1801.

Brian_Jones
21-01-2010, 08:40 AM
So why do you describe the system as a black hole?

Show me where players can find the detail of their ratings and the changes from the previous rating.

Show me the system documentation.

Patrick Byrom
21-01-2010, 04:36 PM
Show me where players can find the detail of their ratings and the changes from the previous rating.

If you're lucky enough to be a Queenslander, they are here:
http://www.southsidejuniorchessclub.org/Sept-Dec%20_09_Unofficial_ACF_Rating_Changes.htm

These are not the official calculations, of course, but if a player notices a significant discrepancy I'm sure they'll let me know :uhoh: .

And if you send me the SP files, we could provide the same information for all the GP events.


Show me the system documentation.
The details of Glicko are available online, as well as various Glicko calculators. For most players, Glicko1 will give you satisfactory results, and the application of this is also available online. (I can supply some details, if you want.)

Garrett
21-01-2010, 04:49 PM
thanks Pat

The information is interesting for someone who likes to look at numbers !!

One thing, your only entry for Joerg Hackenschmidt-Uecker (the only example I looked at) is :

Joerg Hackenschmidt-Uecker 1801 1556 245 5.5/6 1297 300

whereas his actual Jan (edit Dec 09) rating is only 1668

Why is there a discrepancy ? Is it because I picked someone with a big rating change ?

Cheers
Garrett.

Bill Gletsos
21-01-2010, 05:10 PM
thanks Pat

The information is interesting for someone who likes to look at numbers !!

One thing, your only entry for Joerg Hackenschmidt-Uecker (the only example I looked at) is :

Joerg Hackenschmidt-Uecker 1801 1556 245 5.5/6 1297 300

whereas his actual Jan (edit Dec 09) rating is only 1668

Why is there a discrepancy ? Is it because I picked someone with a big rating change ?

Cheers
Garrett.For starters his RD is too high.
It should be around 200 not 300.

Santa
21-01-2010, 10:00 PM
The point is he may have said that but it has not been the case.
most people who have studied the FIDE ratings believes that the FIDE Elo system has been inflating virtually every year.

As usual you dont understand. This has nothing to do with Glicko Vs Elo.


Please, read what I said. I said nothing about Glicko{,2} vs Elo. I was talking about previous and current ACF implementations.





The problem was with the ACF Elo system of the 1980's & 1990's.
Because of the 336 rule players at the top of the rating list continued to gain points they did not deserve and players at the bottom of the list continued to lose points that they should not have lost. Also the bonus point scheme in place generally caused massive overshoots in one list that lead to loss of points by those players in the next list.
All the while the ratings of players were deflating compared to their FIDE ratings.

I've already explained that. I would have removed the 336 rule, and rebased ratings for each tournament as it was processed.

That would fix both problems.



The ACF Council decided that it would apply an uplift to the December 1999 ACF ratings to try and combat this long standing deflation of the ACF ratings and inflation of the FIDE ratings.


You have expressed the opinion, and said many others agree with that opinion, that the FIDE ratings are inflationary.

I don't know whether they are or not, could you please point to the evidence?

Kevin Bonham
21-01-2010, 10:07 PM
You have expressed the opinion, and said many others agree with that opinion, that the FIDE ratings are inflationary.

I don't know whether they are or not, could you please point to the evidence?

Here's a good one to start with:

http://members.shaw.ca/redwards1/

Santa
21-01-2010, 10:09 PM
Santa is most definitely not Graham Saint.
Replace the words lost/misplaced with kept and you would be closer to the mark.
.

The only sensible rejoined to that would have me banned from the forum.

Patrick Byrom
21-01-2010, 10:14 PM
For starters his RD is too high.
It should be around 200 not 300.

This appears to be the main explanation. I ran the program again, giving him an RD of 200 (which would be indicated by a single "?" in the Masterlist), and his new rating is only 1693, fairly close to his actual December rating of 1668.

State Rating Officers don't have access to exact player RDs, of course.

Santa
21-01-2010, 10:16 PM
Kevin's quote that you quoted was directed at Brian Jones not you.

I'm accustomed to threaded views of mailing lists, I forgot the context wouldn't be apparent.

I was indeed referring to comments directed towards Brian Jones, and which I find unhelpful to harmonious discussion.

Bill Gletsos
21-01-2010, 10:16 PM
Please, read what I said. I said nothing about Glicko{,2} vs Elo. I was talking about previous and current ACF implementations.And therefore you were talking about Elo v Glicko as the previous implementation was Elo and the current is Glicko2.

I've already explained that.Where exactly did you explain this.

I would have removed the 336 rule,..Then why didnt you do so whilst you were ACF National Ratings Officer between 1993 and mid 1996. Except for deleting 8,000 names off the master file between December 1992 and April 1993 you did not make any significant changes.

and rebased ratings for each tournament as it was processed.If you mean you would have recalculated all ratings from when you first took over as National Ratings Officer this would have been impossible as you apparently had not kept all the transaction files as there were transaction files missing when you handed over all records and files you had to Graham Saint in mid 1996.

That would fix both problems.No it would not.

You have expressed the opinion, and said many others agree with that opinion, that the FIDE ratings are inflationary.

I don't know whether they are or not, could you please point to the evidence?I am not your research assistant.

BTW What happened to the Compaq computer.

Bill Gletsos
21-01-2010, 10:19 PM
The only sensible rejoined to that would have me banned from the forum.Well I know the machine was not passed on to Graham Saint (the ACF Ratings Officer that replaced you).

Now you seem to be going out of your way to avoid answering the question.

So what did happen to the Compaq computer?

Kevin Bonham
21-01-2010, 10:46 PM
I was indeed referring to comments directed towards Brian Jones, and which I find unhelpful to harmonious discussion.

Anyone who accuses me of having my head in the sand based on a misinterpretation of my comments can take the first steps towards such harmony that you seek by admitting they were wrong.

Furthermore my comments in response to Brian were also in the context of his comments on this related thread (http://chesschat.org/showthread.php?t=11349).

If Brian's comments there are in your opinion harmonious then that harmony must be extremely overrated.

Kevin Bonham
21-01-2010, 10:49 PM
The only sensible rejoined to that would have me banned from the forum.

I think the most sensible rejoinder would actually be the facts.

If this is ACF property and you know what became of it then it would certainly be useful for the ACF to know that.

If you're not inclined to explain then perhaps you should say why not.

Santa
21-01-2010, 11:03 PM
The bit you quoted was not a personal attack, it was simply a reply to what you actually said which was "I'd be happy to drop players ratings three (but you choose a number!) years after their last game. " I was saying this is unnecessary and what we have is in most regards a better alternative.

Whether you were actually responsible for the chucking of ratings from the master list is no concern to me; I am just saying that (i) whoever did it, it was silly and detrimental (ii) you are saying you have no problem with the concept even though in my practical experience it is a very bad idea.


I did not, do not deny doing it, I simply do not remember.



It can be both. Even old ratings have substantial predictive value especially if the player is an adult of middling years. We have a player here who took 13 years off and has basically picked up where he left off (lost a pile of points in his first two tournaments, got them back in the next two).



Statistics are based on analysis of large numbers, not isolated cases. My feeling is that I didn't play so badly - 15 years ago, I''d have accepted 0.5/5 as a likely outcome, but

I would be very happy for my rating to be annulled and to be calculated afresh. There has never been any reason to suppose I'm really a 1500 player, though I did have one tournament performance well above that.

Once I've scored 1/10 or better, then there will be some basis for assigning me a rating that is somewhat reflective of my skills, and changes in my opponents' ratings will be fully justified. While I keep losing all my games, all that you can say is that I''m a lot weaker than my opponents.



No but there is reason to assume they could well be somewhere around that mark and that is what the current system does, but it adjusts them rapidly if evidence shows otherwise.

The fact is that if someone was once a GM or even a 2000+ they are not going to drop to sub-1000 standard unless they are over 100 years old or suffer major brain injury. An old rating might be hundreds of points out but is still some use.


I don't see a reason to assume anything about someone who hasn't played chess for some years. However, when they resume and start scoring wins, then you have a measure which can be used.



There is something I sort-of agree about with you though - people who have been inactive for years should not be able to say they still "have" their old rating. It should be called an expired rating and marked as such in the master list.

A rating is not a player's property in the sense their money is, it's a properly - a measure - of the player, and it belongs to the ratings system.

Different systems might legitimately measure players differently, and there's no good reason one ratings system should assign the same players the same ratings.





It only provides the evidence to test that the system is being applied as stated, not the evidence that it is a good system. The evidence to test that a system is good can be found by examining its predictive success. If a system is not being applied as it should then its predictive success will also suffer.

You could easily over-rate the predictive ability of a ratings system. Necessarily, ratings measure events in the past, and sometimes significantly in the past. Games in a club tournament that ended in December won't be rated until the March list when some of the results could be five-six months old, and that assumes the results are submitted promptly.

A junior could have improved in skill by 100 and more in that time. Until recently, FIDE didn't have that problem to such an extent.

Santa
21-01-2010, 11:09 PM
No more than were suspicious even when they could check it for themselves.



Glicko2 does not use k-factors.



Because this bulletin board is a place where we are subject to often unfactual, misinformed or otherwise silly attacks which we choose to rebut.


I've seen a lot of denial, but no rebuttal. A rebuttal requires facts, it requires the ACF to reveal exactly how its ratings system works.

And as someone else, was it CameronD?, suggested publication of all tournament results so that people can check, if they wish, the results and calculate for themselves their ratings changes.

Santa
21-01-2010, 11:18 PM
I think the most sensible rejoinder would actually be the facts.

If this is ACF property and you know what became of it then it would certainly be useful for the ACF to know that.

If you're not inclined to explain then perhaps you should say why not.

I have been accused of theft.

This is not a proper forum for discussion of the disposal of ACF property, but I will say this much, but I do not propose to say anything further.

"The ACF officers of the day know, or should know, what became of the ACF's Compaq computer. It is not proper for me to speculate further."

Santa
21-01-2010, 11:23 PM
Update: I see that Santa has now got his name in his sig line.

There are quite a few Victorians here who, I thought, might recall me, and it amused me to tease them for a while.

Once it was abundantly clear that those who were curious enough had figured it out (and I did get some PMs), then I'd had my fun.

As for "Santa," it is appropriate and it's what the kids at school call me. I'm slimmer than the traditional model, and my ho ho hos need work.

Basil
21-01-2010, 11:32 PM
This is not a proper forum for discussion of the disposal of ACF property, but I will say this much, but I do not propose to say anything further.
I'd say otherwise. It is appropriate enough. The ACF is transparent. Regardless ...

"The ACF officers of the day know, or should know, what became of the ACF's Compaq computer. It is not proper for me to speculate further."[/QUOTE]
There's no need to speculate and there's no compulsion to divulge in public - but the question from an ACF official seems reasonable and perhaps you could communicate what you *do* know via email or private message.

Adamski
21-01-2010, 11:46 PM
There are quite a few Victorians here who, I thought, might recall me, and it amused me to tease them for a while.

Once it was abundantly clear that those who were curious enough had figured it out (and I did get some PMs), then I'd had my fun.

As for "Santa," it is appropriate and it's what the kids at school call me. I'm slimmer than the traditional model, and my ho ho hos need work.Have to admit that when the grey beard is long I look a bit look Santa myself, and this has been pointed out to me! Currently, I am pleased to say, the beard is short, as evidenced by some of the photos on the Aus Champs web site! And those chess chatters who saw me there can also vouch for that!
Good night/ morning all! Time for bed...

Santa
22-01-2010, 01:22 AM
I'm not sure what the ratings officers' view is about mathematical accuracy but I see a big advantage of Glicko over ELO being its responsiveness so that changes in playing strength are picked up quicker including rapidly improving juniors' improvement. The system better reflects players' current playing strength.


Done properly, an Elo system can be quite responsive to players's increasing skills.

What the old ACF system did wrong was to
1. Derive K from the event, not from the player's history
2. Rate every game for a period with the players' initial rating for the period.
3. The 336 rule.

The Elo calculation produces a notional change in rating for a game. Then, that value is multiplied by K and the ratings of both players adjusted accordingly. Typically, a junior should have a higher K say 35. An established, strongish player - say Brian Jones - might have K=20 and someone like GM Johansen K=15 or even K=10.

These numbers are illustrative, not recommendations:)

Jones plays Johansen and wins. Jones' rating would go up by almost 20 points while Johansens' would only go down by almost 10.

But since the ACF made K a property of the event, and used K=15 for most events, K=30 for the Australian Championship, if they'd met in the Australian Championship, Brian's rating would have gone up by almost 30 points, Darryl's down by the same amount.

Glicko addresses this with RD, assigning different values do different players. The results should in fact be broadly similar.

I have in my hands an ACF ratings booklet for December 1986 - before Shaun Press, Ian Laurie was the ratings officer back then.

Some of the most active were:
Victor Kildisis 68 V
Malcolm Murrell 46 N
Stephen Solomon 38 Q
Bernard Anton 34 W
Boris Nikoloff 29 S
Gus Korda 16 A
John Steen 13 T

Funnily enough, Victor was also one of the big improvers. His rating went from 1837 to 2018, an increase of 182.

Because of the way the ACF implemented the Elo system, for each game Victor played in that period, his input rating was 1837, even for the last games he played when the rating should have been much higher than 1837, thought probably much lower than 2018: without actual results from the period, I can only speculate.

Quite probably, he also attracted bonus points based on his input rating of 1837.

Most people would agree that games played long ago are less relevant than recent games in measuring players' skill, and more so with juniors and inexperienced players.

Statisticians will tell you you're getting a pretty good measure after 30 measurements, where there is not reason to suppose the subject's skill is changing. Glicko recognised that with lower RD, Elo with lower K for established players.

A prime effect of K is to discard old results, with low values of K, a player's rating changes very slowly. With higher values of K, old results are discounted quickly, as is appropriate for juniors and inexperienced players.

The point here is that, five of the players I named above played enough games in the period that no games before the period has any meaning.

However, the ACF's wrong implementation of the Elo system did retain those games and used them in every calculation of ratings changes for those players and for their opponents.


Should I go on?
yes, I will!

If your ratings system is working well, then processing the same tournament twice won't have a big effect on players' ratings, and if you process it enough times, the ratings will stabilise.

This occurs because it "predicts" the results, and then the results reprocessed confirm those results.

However, since the ACF system did not update ratings between tournaments, but kept on using the same input ratings for every tournament in the period, that would not happen. I'm looking at the Victorian Winter Interclub B grade, and the player of interest is Malcolm Pyke. One of those who just competed in the Australian Champs.

His rating, according to On The move, was 1323 going in, he scored four points (don't know how many games), performed at 1817 and earned a decent 80 points rating increase.

Now, if the ACF processed that tournament ten times the way it did, Malcolm's rating would have kept on going up, 800 after ten times.

Some simple changes are all that were needed to fixed the most serious problems with the ratings system, and then its performance could have been reevaluated.

Of course, it is easy to criticise the old ratings system because its workings and its failings were clearly documented for all to see (though, of course, the failings were not highlighted and only became evident with understanding).

It's not possible to criticise the existing system because it's a closed secret, and nobody's saying how it works.

Santa
22-01-2010, 01:34 AM
As he drew with players rated about 800 and 1100, his new rating doesn't seem too unreasonable. Even if he had beaten them, but lost to the 1300 player, it would still imply that he wasn't 1700. And he can get the points back if he plays well in his next event.

There are huge class prizes in USCF events, so it has different priorities to the ACF.

A limit on rating drops in one event might be a reasonable change, but this can be done in Glicko as easily as in Elo.


I prefer a change to eligibility for prizes rather than a change to the ratings system.

If Thornicraft's rating has been 2200, make him ineligible for class prizes below 2000 by all means, but if he turns in bad results, punish him. If it makes him ineligible for a most attractive event, it's his problem.

These days, space isn't a great concern in publishing ratings, and players' peak ratings can easily be published.

Santa
22-01-2010, 01:35 AM
Have to admit that when the grey beard is long I look a bit look Santa myself, and this has been pointed out to me! Currently, I am pleased to say, the beard is short, as evidenced by some of the photos on the Aus Champs web site! And those chess chatters who saw me there can also vouch for that!
Good night/ morning all! Time for bed...

I'm well covered top and bottom.

Santa
22-01-2010, 01:41 AM
No.
Unlike Elo, Glicko is not zero sum.

Done properly, Elo is not zero sum either.

Santa
22-01-2010, 01:45 AM
This is precisely the kind of thing I had in mind (in an extreme example) when referring to the responsiveness advantage of Glicko. If a player rated 2000 takes a 20-year break from chess and returns with a performance rating of 1500 he deserves to lose a heap of rating points as he is no longer anywhere near 2000. It may dent his pride but rating points are based on performance, not ego.

DJ

Dropping players' ratings completely after a substantial period of inactivity and starting afresh would do the same.

There is room for some debate as to what constitutes a substantial period, whether it's five, ten, fifteen or twenty years.

Santa
22-01-2010, 02:38 AM
Here's a good one to start with:

http://members.shaw.ca/redwards1/

oh, really!!

He starts out, Everyone is aware that the FIDE ratings have become inflated

He assumes it's so, and then proceeds to A test for rating integrity is to check whether an average measure of the top of the rating pool remains consistent.

and then seeks to justify his stance.

Then there's his Edo ratings system which The Edo system uses the simple (but new) idea of treating a single player in different years as separate players, and positing hypothetical drawn self-games between the same player in consecutive years. For example, Staunton-1845 is considered to have played a 30-game match with Staunton-1844 with a resulting score of 15-15.

Seriously, is chess the only sport where performances are not improving?
In some sports, it's easy to measure the improvement. Take running. in May 1954 Roger Bannister was the first man to run a mile in under four minutes. In June, John Landy did it. These days, that seems pretty pedestrian.

I remember when swimmers first completed the 1500 metre event in under 15 minutes. Now, it's a fairly regular event.

In tennis, it's harder, but there are concrete measurements that can be made - speed of serve, accuracy of serve for example, and watching video footage of years ago and today will show differences too. Doubtless tennis coaches have more measures.

I chess, I suppose one could look at the games of the best players of an era, but I suspect few are well-qualified - I read recently of a game where Byrne (playing Fisher) resigned just as a couple of GMs commentating agreed he was winning! I suggest only the best players could make that kind of assessment, but those players have their own reputations to defend.

Perhaps those commentators had a disadvantage, they couldn't be wholly engaged in the game.

Just as we now have much improved training for football, tennis, cricket, so we have better training and knowledge - especially knowledge - available to chessplayers. You want to know IM Lane's favourite openings to prepare against him? It's all on the Internet.

I note that 1985, when Edwards says inflation really set in is also about when personal computers were halfway decent - it was about then I bought my 386 which I subsequently used for VCA and ACF work, and chessplaying programs running on them were giving some of the better players in the ViC Open a hard time. it was becoming easier to write bigger (and better) books on all matters, including chess.

Perhaps those computers, directly and indirectly, helped players of all levels.

And then came, not Jones, but the Internet....

I don't accept that players' ratings being higher than ever, of itself, means that FIDE's ratings are inflating. I would expect players to become stronger and for more players than before to have high ratings, and for pressure at the very top.

To me, it's more surprising that the highest ratings are not higher, I would expect chessplayers to improve just as much as competitors in other established sports have done.

Recently, I watched some games in the World Chess Cup, as they were played. Years ago, that was not possible. Would that help my chess?

Even as a helper at the Australian Championships one year, I didn't have such a good view of any games, and besides, the players were better.

CameronD
22-01-2010, 06:49 AM
Ratings are not meant to compare players of different playing years, its meant to compare players of the same period. So the fact that the top ratings havnt exploded over 3000 doesnt mean anything, as its not comparing abilities from 50 years ago, its comparing the current players to each other




oh, really!!

He starts out, Everyone is aware that the FIDE ratings have become inflated

He assumes it's so, and then proceeds to A test for rating integrity is to check whether an average measure of the top of the rating pool remains consistent.

and then seeks to justify his stance.

Then there's his Edo ratings system which The Edo system uses the simple (but new) idea of treating a single player in different years as separate players, and positing hypothetical drawn self-games between the same player in consecutive years. For example, Staunton-1845 is considered to have played a 30-game match with Staunton-1844 with a resulting score of 15-15.

Seriously, is chess the only sport where performances are not improving?
In some sports, it's easy to measure the improvement. Take running. in May 1954 Roger Bannister was the first man to run a mile in under four minutes. In June, John Landy did it. These days, that seems pretty pedestrian.

I remember when swimmers first completed the 1500 metre event in under 15 minutes. Now, it's a fairly regular event.

In tennis, it's harder, but there are concrete measurements that can be made - speed of serve, accuracy of serve for example, and watching video footage of years ago and today will show differences too. Doubtless tennis coaches have more measures.

I chess, I suppose one could look at the games of the best players of an era, but I suspect few are well-qualified - I read recently of a game where Byrne (playing Fisher) resigned just as a couple of GMs commentating agreed he was winning! I suggest only the best players could make that kind of assessment, but those players have their own reputations to defend.

Perhaps those commentators had a disadvantage, they couldn't be wholly engaged in the game.

Just as we now have much improved training for football, tennis, cricket, so we have better training and knowledge - especially knowledge - available to chessplayers. You want to know IM Lane's favourite openings to prepare against him? It's all on the Internet.

I note that 1985, when Edwards says inflation really set in is also about when personal computers were halfway decent - it was about then I bought my 386 which I subsequently used for VCA and ACF work, and chessplaying programs running on them were giving some of the better players in the ViC Open a hard time. it was becoming easier to write bigger (and better) books on all matters, including chess.

Perhaps those computers, directly and indirectly, helped players of all levels.

And then came, not Jones, but the Internet....

I don't accept that players' ratings being higher than ever, of itself, means that FIDE's ratings are inflating. I would expect players to become stronger and for more players than before to have high ratings, and for pressure at the very top.

To me, it's more surprising that the highest ratings are not higher, I would expect chessplayers to improve just as much as competitors in other established sports have done.

Recently, I watched some games in the World Chess Cup, as they were played. Years ago, that was not possible. Would that help my chess?

Even as a helper at the Australian Championships one year, I didn't have such a good view of any games, and besides, the players were better.

Brian_Jones
22-01-2010, 06:56 AM
If you're lucky enough to be a Queenslander, they are here:
http://www.southsidejuniorchessclub.org/Sept-Dec%20_09_Unofficial_ACF_Rating_Changes.htm

These are not the official calculations, of course, but if a player notices a significant discrepancy I'm sure they'll let me know :uhoh: .

And if you send me the SP files, we could provide the same information for all the GP events.


The details of Glicko are available online, as well as various Glicko calculators. For most players, Glicko1 will give you satisfactory results, and the application of this is also available online. (I can supply some details, if you want.)

Please post the URLs but I would still prefer to see the official calculations.

Why are you, Pat, the Queensland State rating Officer, not given official calculations.

Or is this another example of the Bill Gletsos secret approach to Australian Chess administration?

Brian_Jones
22-01-2010, 07:05 AM
I'd say otherwise. It is appropriate enough. The ACF is transparent. Regardless ...

More bollocks from Howard Duggan! :)

Kevin Bonham
22-01-2010, 08:35 AM
oh, really!!

He starts out, Everyone is aware that the FIDE ratings have become inflated

He assumes it's so, and then proceeds to A test for rating integrity is to check whether an average measure of the top of the rating pool remains consistent.

and then seeks to justify his stance.

He does address the counter-point you raise however. That he asserts from the start that inflation is a known problem is irrelevant since the purpose of his article is to provide evidence in response to those who are sceptical of the bleeding obvious.

That FIDE ratings have inflated at the top end is clearly fact. Whether that inflation is caused by system defect, improvement or a combination of both is then debatable. In my view it's most likely system defect, and I don't find references to the introduction of effective computers a convincing counter. There had been other technological advances prior to that that had increased the ease with which material relevant to the game could be transmitted but Elo's own historical ratings do not show an inflation over time as a result of this.

Spiny Norman
22-01-2010, 09:01 AM
... the fact that the top ratings havnt exploded over 3000 ...
IIRC, Elo was designed to generate ratings between 0 and 3000, so it would be mathemtically impossible to get ratings over 3000.

Bill Gletsos
22-01-2010, 09:10 AM
IIRC, Elo was designed to generate ratings between 0 and 3000, so it would be mathemtically impossible to get ratings over 3000.Elo has no upper bound of 3000, nor a lower bound of 0.

If a player currently existed who could consistently beat a 2800 player 80% of the time then their rating would be over 3000.

Likewise you could have negative ratings but most systems will set a floor of either 0 or 100.

Basil
22-01-2010, 10:19 AM
More bollocks from Howard Duggan! :)
A strong opening. Somewhat lacking in substantiation. Useful for surprise value but definitely unsound.

Spiny Norman
22-01-2010, 10:26 AM
Thanks Bill. My recollection is clearly incorrect! However I do recall reading something about the range 0-3000 at some point, many years ago. How is it (historically speaking) that all players came to be rated between 0 and 3000? Why not some other range?

Bill Gletsos
22-01-2010, 12:02 PM
Thanks Bill. My recollection is clearly incorrect! However I do recall reading something about the range 0-3000 at some point, many years ago. How is it (historically speaking) that all players came to be rated between 0 and 3000? Why not some other range?Elo in his book noted that the rating scale was open ended.
Prior to the Elo system being used in the USA in 1960, the USA used the Harkness System from 1950-1960. In the Harkness system 2600+ was for GM's, 2400+ for senior masters, 2200+ for master and 2000+ for experts and candidate masters whilst below 2000 was for the various amateur class players.
Elo maintained that scale when he introduced his system.

CameronD
22-01-2010, 06:21 PM
At ICC server, some players have a 3500 rating

Bill Gletsos
22-01-2010, 07:02 PM
Done properly, Elo is not zero sum either.
In the vast majority of cases such as in the FIDE system and the old ACF system it was.
However that is immaterial as I was comparing ACF Elo to its replacement by the Glicko system.

Patrick Byrom
23-01-2010, 01:39 AM
Please post the URLs but I would still prefer to see the official calculations.
Why are you, Pat, the Queensland State rating Officer, not given official calculations.
Or is this another example of the Bill Gletsos secret approach to Australian Chess administration?

As usual, Wikipedia is a good starting point:
http://en.wikipedia.org/wiki/Glicko_rating_system

And then there is Barry's Glicko1 calculator:
http://www.bjcox.com/modules.php?name=Glicko_Calc

These are the formulas I use in my program, which gives results fairly close to the official calculations in most cases. If a player notices a significant difference between my results and the official ones, and thinks that there is a problem, I'm sure that Bill would investigate it - as he has already done several times on this thread.

The ACF Rating Program calculates the total ratings changes for each player over all events. So the official calculations would not be very useful to most players, who want to know how their rating changes in each event - which is what I'm calculating.

I have never asked for the official calculations, as they would only be useful to me in the case of a major discrepancy.

Tony Dowden
01-02-2010, 06:44 PM
What is the point of historical consistency if the previous ratings ...
What kind of ratings were these? (Elo? Glicko? - just curious)

Kevin Bonham
01-02-2010, 08:40 PM
What kind of ratings were these? (Elo? Glicko? - just curious)

It was certainly ELO (of various forms) through the second half of the 80s and the 90s with Glicko commencing in 2000. I am not sure how far back it was ELO and have heard that a primitive Glicko-like system was used at some stage in the distant past.

Bill Gletsos
01-02-2010, 09:31 PM
It was certainly ELO (of various forms) through the second half of the 80s and the 90s with Glicko commencing in 2000. I am not sure how far back it was ELO and have heard that a primitive Glicko-like system was used at some stage in the distant past.It was Elo from 1983 till August 2000.
From around 1962 to 1982 it was the Roger Cook system.

Kevin Bonham
01-02-2010, 09:52 PM
It was Elo from 1983 till August 2000.
From around 1962 to 1982 it was the Roger Cook system.

I have some old Tasmanian ratings from Dec 1970 and Nov 1979 that appear to have been done under this system (whatever it was).

On the 1970 list the highest rating showing for Tas is 1640 and the lowest is -290. Some of the players have m, ?, ?? or x after their name. The x can occur in conjunction with one or two ?s or without.

On the 1979 list the highest rating showing for Tas is 1885. Several players have m showing after their name and relatively few have ? or ??, none have x. The lowest rating listed is 745??

I also have some partial listings from 1975 and 1977 and these show the top Tas listings at around 1900-1950.

I suspect the 1975 (and onward) lists have some rough parity with FIDE and the later ACF ELO ratings while the 1970 list has none at all.

Bill Gletsos
01-02-2010, 09:59 PM
I have some old Tasmanian ratings from Dec 1970 and Nov 1979 that appear to have been done under this system (whatever it was).

On the 1970 list the highest rating showing for Tas is 1640 and the lowest is -290. Some of the players have m, ?, ?? or x after their name. The x can occur in conjunction with one or two ?s or without.

On the 1979 list the highest rating showing for Tas is 1885. Several players have m showing after their name and relatively few have ? or ??, none have x. The lowest rating listed is 745??

I also have some partial listings from 1975 and 1977 and these show the top Tas listings at around 1900-1950.

I suspect the 1975 (and onward) lists have some rough parity with FIDE and the later ACF ELO ratings while the 1970 list has none at all.Up until 1972 the Cook system was on a different scale to the Elo scale. From January 1973 it was to the Elo scale.

Adamski
01-02-2010, 10:17 PM
It was Elo from 1983 till August 2000.
From around 1962 to 1982 it was the Roger Cook system.Is this the same Roger Cook who currently plays at norths?

Bill Gletsos
01-02-2010, 10:36 PM
Is this the same Roger Cook who currently plays at norths?Yes.

Tony Dowden
02-02-2010, 08:20 PM
It was certainly ELO (of various forms) through the second half of the 80s and the 90s with Glicko commencing in 2000. I am not sure how far back it was ELO and have heard that a primitive Glicko-like system was used at some stage in the distant past.
Thanks KB

Santa
27-11-2011, 07:51 PM
Santa is most definitely not Graham Saint.
Replace the words lost/misplaced with kept and you would be closer to the mark.
Santa was the ACF National Ratings Officer from 1993 to mid 1996.

I have thought about this for some time, and decided I will make an explanation.

In the late 80s I bought an Ipex 386 from an outlet in St Kilda Rd called Trinix or something like.

Both my system and the ACF's had similar processors, but I had much more disk space and, I think, more RAM. My system is what I already used to publish On The Move, I had the VCA's Ventura Publisher on it, and the VCA's Laserjet, my own dodgy imitation Laserjet having expired.

The Compaq was of no use to me, and I only had possession of it long enough to transfer relevant materials to my system.