PDA

View Full Version : rating system



Davidflude
08-04-2006, 10:34 AM
I think that it time to revise the rating system. The big problem is that improving juniors tend to be always und rated and they drive down the adult ratings.

I suggest the following rule for when adults play juniors.

The junior gains or loses rating points according to the result as at present.
The adult neither gains or loses rating points.

Brian_Jones
08-04-2006, 01:06 PM
The adult neither gains or loses rating points.

That sounds like a great idea to me Fludy (seriously!)

Rincewind
08-04-2006, 02:04 PM
Sounds a bit silly to me and would also lead to rorting where adults will throw games to juniors to boost up their rating. Particularly in the bottom boards of club events. Particularly clubs that have a large number of juniors playing.

Cat
08-04-2006, 06:03 PM
That sounds like a great idea to me Fludy (seriously!)

Statistically its sound, tennis does a similar thing (but not exactly the same).

Frank Walker
08-04-2006, 06:38 PM
Thats an absolutely ridiculous idea...

You are just having trouble beating the juniors arent you...
Adults can improve TOO you know!

Cat
08-04-2006, 11:46 PM
Thats an absolutely ridiculous idea...

You are just having trouble beating the juniors arent you...
Adults can improve TOO you know!

No its a sensible idea. Argue the merits, don't insult. If you have a point, raise it - no need to get personal.

PHAT
09-04-2006, 09:11 AM
No its a sensible idea. Argue the merits, don't insult. If you have a point, raise it - no need to get personal.

He has been taking lessons from BG on how to defend the Glicko.

peter_parr
11-04-2006, 03:01 PM
Toby Lewski (rating 1951?) scored 2/7 against an average rating of 1725 in the Doeberl Cup (Easter) 2005. His expected score was 5.53.

He scored 3.5 points less then he should have.
On any normal rating system (FIDE) etc he would expect to lose no more than 50 rating points.

Under the ACF Glicko rating system he lost 337 rating points – down from 1951? To 1614?
Last year I talked to Gletsos (ACF ratings officer) about this.

PP – “under FIDE you lose a maximum of 15 points when you lose to a very low rated player”
BG – “It is a silly system”

PP – “is it a good system to lose 337 rating points when scoring 3.5 points less than expected in a 7 round event”
BG – “Yes”

PP “is there any other country in the world (161 FIDE countries) where the Glicko system is used.
BG “No”

PP – “is there any other city, state or country in the world where a player can lose 337 rating points in a weekend”
BG – “No”

PP – “is the rest of the world wrong and Glicko right?
BG – “yes”

Lewski was in my shop again today in the Sydney CBD telling me once again how absurd it was losing 337 rating points in a weekend.
I fully agree with Lewski a regular Sydney player who I have known for 38 years.
Personally I would like to play a few tournaments from time to time but would never consider playing while the Glicko system is in use.
With a rating of 2227 based on many years I could lose 500 or more rating points in one small event, absurd.

This applies to numerous players who have dropped out of the game for a few years.
I always advise players they are risking being Glicko’d.

BTW on the latest ACF (March 2006) rating list the top 10 players lost 68 rating points, the second 10 lost 45, the third ten lost 70 the fourth 10 lost 164, from the top 40 (top players list) 10 played no games and the other 30 between then lost 347 (average 11.57 rating points per player)

This is only one of the 4 rating lists per year. So overall yearly loss??

Any sensible system like FIDE should be used.
It really is time the ACF over ruled Gletsos and used a normal rating system.

arosar
11-04-2006, 03:08 PM
Shock horror! I am talking to Lewski today. I play chess with him everyday. I get some quotes from him too.

AR

jenni
11-04-2006, 03:11 PM
I thought we weren't allowed to quote stuff, or am I mistaken? There are so many rules now and I am too lazy to keep track of them.......

Vlad
11-04-2006, 03:37 PM
Personally I would like to play a few tournaments from time to time but would never consider playing while the Glicko system is in use.
With a rating of 2227 based on many years I could lose 500 or more rating points in one small event, absurd.


Peter, your rating on playchess is about 2000. That means your ACF should not be higher than 1900-2000. The only reason you have 2227 is because you accumulated 220 points without playing in the last few years, when everybody's rating was raised. I really can't see what you are complaining about. You should be under 2000 and you scared to play because you will loose your overrated rating.

In defence of the current system I can say couple things.

1) It is symmetric. If you did very badly this time and lost 300 points, you can do very well next period and gain 300 points back.
2) I believe most of the top 20 players are overrated in comparison with others. They have received 220 points but they have not played many games recently. This is exactly what your statistics shows. It is saying the system should be even more "Glicko" to get the top players converge to their actual ratings quicker.

WhiteElephant
11-04-2006, 03:53 PM
Peter, your rating on playchess is about 2000. That means your ACF should not be higher than 1900-2000. The only reason you have 2227 is because you accumulated 220 points without playing in the last few years, when everybody's rating was raised. I really can't see what you are complaining about. You should be under 2000 and you scared to play because you will loose your overrated rating.

In defence of the current system I can say couple things.

1) It is symmetric. If you did very badly this time and lost 300 points, you can do very well next period and gain 300 points back.
2) I believe most of the top 20 players are overrated in comparison with others. They have received 220 points but they have not played many games recently. This is exactly what your statistics shows. It is saying the system should be even more "Glicko" to get the top players converge to their actual ratings quicker.

Did some people's rating really go up by 220? I was inactive during the rating uplift periods. I was given a rating increase of 70 or 90 - I don't remember exactly. Then some time later I was given another, similar uplift. However, the second uplift was then taken away supposedly because I continued to be inactive. So my overall rise while inactive was 70-90 in total.

Bill Gletsos
11-04-2006, 03:59 PM
Toby Lewski (rating 1951?) scored 2/7 against an average rating of 1725 in the Doeberl Cup (Easter) 2005. His expected score was 5.53.

He scored 3.5 points less then he should have.
On any normal rating system (FIDE) etc he would expect to lose no more than 50 rating points.

Under the ACF Glicko rating system he lost 337 rating points – down from 1951? To 1614?
Last year I talked to Gletsos (ACF ratings officer) about this.

PP – “under FIDE you lose a maximum of 15 points when you lose to a very low rated player”
BG – “It is a silly system”

PP – “is it a good system to lose 337 rating points when scoring 3.5 points less than expected in a 7 round event”
BG – “Yes”

PP “is there any other country in the world (161 FIDE countries) where the Glicko system is used.
BG “No”

PP – “is there any other city, state or country in the world where a player can lose 337 rating points in a weekend”
BG – “No”

PP – “is the rest of the world wrong and Glicko right?
BG – “yes”I see you are still making up supposed conversations.

Vlad
11-04-2006, 04:14 PM
Did some people's rating really go up by 220? I was inactive during the rating uplift periods. I was given a rating increase of 70 or 90 - I don't remember exactly. Then some time later I was given another, similar uplift. However, the second uplift was then taken away supposedly because I continued to be inactive. So my overall rise while inactive was 70-90 in total.

The first uplift was 150 points. The second one was 70 points. Even if we assume that the second one was taken away, 150 is still huge increase.

If you do not play for a few years, your playing strength will be at least 100 points less. About 10 years ago I had couple competitions with performance around 2300 FIDE. Then I did not play for a few years. I started playing in Australia in 1999. My first rating was 1850. That partly reflects the fact that ACT ratings are highly underrated, but still 450 points drop.

Bill Gletsos
11-04-2006, 04:28 PM
Peter, your rating on playchess is about 2000. That means your ACF should not be higher than 1900-2000. The only reason you have 2227 is because you accumulated 220 points without playing in the last few years, when everybody's rating was raised. I really can't see what you are complaining about. You should be under 2000 and you scared to play because you will loose your overrated rating.Peter last played in 1998. His 2227 rating only includes the 150 point uplift of April 2000. It does not include the 70 point uplift of March 2004 which was only applied to players active between April 2000 and December 2003.

In defence of the current system I can say couple things.

1) It is symmetric. If you did very badly this time and lost 300 points, you can do very well next period and gain 300 points back.
2) I believe most of the top 20 players are overrated in comparison with others. They have received 220 points but they have not played many games recently. This is exactly what your statistics shows. It is saying the system should be even more "Glicko" to get the top players converge to their actual ratings quicker.Yes prior to the August 2002 rating period Lewski had not played a rated game of chess since prior to 1981. In the 7 games he played in the August 2002 period he had a performance rating of around 1854.
In the 2005 Doeberl Cup where he scored 2/7 he results were:
loss 2066, loss 1771, loss 1704, loss 1667, loss 1758, win 1463, win 1650.
That is a rough performance rating of 1566. His true performance rating was even lower at 1526.

Also whilst talking about averages Peter conveinetly ignores a number of issues. Firstly he ignores the fact that players will be losing points to overseas players who are not include in the top list as well as to players further down the list. For the players who were active in the March 2006 rating period who had played at least 30 rated games prior to the March period and had !! at the start of the period and !! at the end of the period the average rating increased by 4.14 points. That represented 44% of rated players who played in the period. If we also include all players who had an ! at the start of the period and either an ! or !! at the end of the period the average increase was 1.11 points. This represented 71% of rated players who played in the period.

ElevatorEscapee
11-04-2006, 04:28 PM
I see you are still making up supposed conversations.
Maybe he was paraphrasing. ;) :lol:



... About 10 years ago I had couple competitions with performance around 2300 FIDE. Then I did not play for a few years. I started playing in Australia in 1999. My first rating was 1850. That partly reflects the fact that ACT ratings are highly underrated, but still 450 points drop.

Or it could just be a difference between FIDE ratings and ACF ratings. :)

Vlad
11-04-2006, 04:46 PM
Or it could just be a difference between FIDE ratings and ACF ratings. :)

The current difference for me is about 12 points, why would you think it should have been 450 points in 1999?

peter_parr
12-04-2006, 02:29 PM
Peter, your rating on playchess is about 2000. That means your ACF should not be higher than 1900-2000. The only reason you have 2227 is because you accumulated 220 points without playing in the last few years, when everybody's rating was raised.

Drug if you are quoting facts please get the facts correct.

My rating on Playchess (Peter Toparrov) is not about 2000.
Playchess has my bullet rating at 2218, last played 10 April 2006 (4239 games) and Playchess has my blitz rating at 2154, last played 9 April 2006 (775 games).

I do enjoy playing chess.
Furthermore I have always been much weaker at quick play games than tournament games.

You claim I accumulated 220 points without playing is also incorrect.
The 150 rating points was added to all active players including myself.
Gletsos claims I last played in 1988 which is of course also untrue.

The last time Gletsos made a rating adjustment to my rating was when he chopped off 70 rating points reducing my rating from 2297 to 2227.

Drug my rating is not relevant.

Answer this:- Lewski’s result if processed by FIDE would lose 50 FIDE ratings points from seven games. Glicko rating the same result loses 337 rating points.

Are you of the opinion FIDE is the better system or Glicko is the better system?

They cannot both be right can they?

There is no place anywhere else in the world when a player can compete in a weekender (7 games) score 2 points and lose 337 rating points.

FIDE does not severely punish players for inactivity.
Why is Australia different to all other nations.

Players losing hundreds of rating points in a few days sometimes stop playing for ever – very sad.

peter_parr
12-04-2006, 02:35 PM
I see you are still making up supposed conversations.

Your litany of lies continues – or is it a similar loss of memory as being experienced by Alexander Downer.

The conversation took place in my shop in the CBD.

Present in the discussion in my shop was ACF Rating Officer Graham Saint, ACF Rating Officer Bill Gletsos and myself. I thought with both ACF ratings officers in my shop at the same time it was a good time to discuss the 337 rating points chopped off of Lewski’s rating from only 7 games (2 points scored).

If you have lost your memory on this and numerous other matters this may explain your overall behaviour in recent months “Peter last played in 1988” – Has your computer also lost its memory??

WhiteElephant
12-04-2006, 02:50 PM
Peter Toparrov - hehe

Vlad
12-04-2006, 03:09 PM
Drug if you are quoting facts please get the facts correct.

My rating on Playchess (Peter Toparrov) is not about 2000.
Playchess has my bullet rating at 2218, last played 10 April 2006 (4239 games) and Playchess has my blitz rating at 2154, last played 9 April 2006 (775 games).


My maximum blitz rating on ICC is about 2900, my current blitz rating on ICC is about 2800, at some point a month ago my rating was about 2500. (I have not recently played on playchess, but when I did a year ago the numbers were comparable.) So tell me what numbers did I get wrong? You are fluctuating from 1900 to 2200 and I picked up the middle of the diapason. If you were to say that my ICC rating is 2600 I would think it is fair.




You claim I accumulated 220 points without playing is also incorrect.
The 150 rating points was added to all active players including myself.



Peter, when you repeat something that has been stated at least 2 times before, you think it will become clearer for others? Yes you did accumulate 150 points without playing plus you have not played for years. If you are to play now your fair rating should be less than 2000. It will be fair if you loose 200 points in the first tournament you play.




Drug my rating is not relevant.



I completely agree with you that it is irrelevant. Why do you start talking about your rating then?





Answer this:- Lewski’s result if processed by FIDE would lose 50 FIDE ratings points from seven games. Glicko rating the same result loses 337 rating points.


Peter, the pool of players who have FIDE ratings is different from the pool of players who have ACF rating. There are a way to many underrated juniors in Australia, who create a huge deflation. The Glicko system is good because it allows juniors to get their real ratings quicker. There are not many juniors among players who have FIDE ratings. In fact, there is a small inflation of FIDE ratings. The FIDE system is good when there is no huge deflation or inflation. It is all technical and I do not expect you to understand. However, I would just like to note that the situation in Australia is quite unique in the sense that very small kids here can get ratings. That certainly was not the case when I was a kid back in Russia. In fact, my first rating was about 1800. Given that the situation is unique I can’t see why you think Glicko is not a solution to this unique situation.



Players losing hundreds of rating points in a few days sometimes stop playing for ever – very sad.

They stopping not because of the Glicko system. They stopping because 1) they care too much about their rating; 2) because there is a deflation; they maximise their ratings by not playing.

Bill Gletsos
12-04-2006, 03:30 PM
Gletsos claims I last played in 1988 which is of course also untrue.That was a typo. It should have been 1998.
I have corrected the original post.

PHAT
12-04-2006, 04:57 PM
I see you are still making up supposed conversations.

Nobody believes you.

Your not worth the stress you cause. You should quit the NSWCA and quit work and get yourself onto the intelectual disability pension. But make sure you don't get too bored living with your relos and a bulletin board. You might go back to work and start earning a quid again while you are still collecting a benefit. That would make you a criminal.

Kevin Bonham
12-04-2006, 07:50 PM
I thought we weren't allowed to quote stuff, or am I mistaken? There are so many rules now and I am too lazy to keep track of them.......

Top post in Moderation Decisions thread lists them all. He is indeed outside the site rules but to this point no complaint has been made.

Kevin Bonham
12-04-2006, 09:00 PM
Toby Lewski (rating 1951?) scored 2/7 against an average rating of 1725 in the Doeberl Cup (Easter) 2005. His expected score was 5.53.

He scored 3.5 points less then he should have.
On any normal rating system (FIDE) etc he would expect to lose no more than 50 rating points.

Under the ACF Glicko rating system he lost 337 rating points – down from 1951? To 1614?

Sounds completely statistically correct to me.

A genuine 1900s player will almost never perform in the 1500s. I have been usually within 50 points of 1950 (or pre-uplift equivalent) for almost a decade and in that time have recorded only one performance that bad in over 70 tournaments. If a player who has not played for several years and was rated in the 1900s returns a 1500s performance then the most likely reason is that their playing strength has decreased, either temporarily due to lack of practice or permanently for whatever reason.



Lewski was in my shop again today in the Sydney CBD telling me once again how absurd it was losing 337 rating points in a weekend.

If he is as strong as his old rating he will quickly get them back.


Personally I would like to play a few tournaments from time to time but would never consider playing while the Glicko system is in use.
With a rating of 2227 based on many years I could lose 500 or more rating points in one small event, absurd.

Do you really think you would ever play at 1600 strength?


BTW on the latest ACF (March 2006) rating list the top 10 players lost 68 rating points, the second 10 lost 45, the third ten lost 70 the fourth 10 lost 164, from the top 40 (top players list) 10 played no games and the other 30 between then lost 347 (average 11.57 rating points per player)

This is only one of the 4 rating lists per year. So overall yearly loss??

Firstly, a minor point but one worth making - taking the top 40 from one list and seeing what happens to them next list is not statistically correct because there is exchange between one top 40 list and the next. So your sample includes Booth (top 40 in December 2005) who then lost 47 points but excludes Ly (top 40 in March 2006) who gained 23 points in entering the top 40. So to be fair you should either exclude Booth from your sample (-300) or include Ly (-324).

In any case, or even leaving it as is, the bulk of the rating loss is due to two players, Caoilli (-127) who had an extremely bad Aus Champs result and Wohl (-80) whose old ACF rating was widely known, including by Wohl himself who apparently found it rather amusing when he was #2, to be inflated. (Wohl happened to be inactive in Australia during a period when there was significant loss of points from the top end of the ratings for reasons much debated on these boards at the time.)

If you examine the net rating difference for the 40 players you have considered in the six months leading up to December 2005 you will notice that the net change is much smaller, not larger as your question asks. By my calculations (which are by hand hence I do not guarantee them to be completely clean of small errors) that group of 40 players dropped 128 ratings points between them between June 2005 and December 2005.

A rate of points loss that low may mean something or may not. It is so prone to the vagaries of individual players that it is impossible to say without examining patterns over a few years very carefully.


Any sensible system like FIDE should be used.
It really is time the ACF over ruled Gletsos and used a normal rating system.

FIDE is way too slow in dealing with improving players.

Bill Gletsos
13-04-2006, 12:45 AM
Your litany of lies continues – or is it a similar loss of memory as being experienced by Alexander Downer.

The conversation took place in my shop in the CBD.

Present in the discussion in my shop was ACF Rating Officer Graham Saint, ACF Rating Officer Bill Gletsos and myself. I thought with both ACF ratings officers in my shop at the same time it was a good time to discuss the 337 rating points chopped off of Lewski’s rating from only 7 games (2 points scored).The conversation went nothing like you have suggested.

If you have lost your memory on this and numerous other matters this may explain your overall behaviour in recent months “Peter last played in 1988” – Has your computer also lost its memory??I already pointed out that was a typo and should have been 1998.

four four two
13-04-2006, 01:03 AM
If he is as strong as his old rating he will quickly get them back...you cant be serious Kevin?:lol:

The amount of games you would have to win after losing 300+ points in one tournament is going to take you a fair while.:whistle:

Alan Shore
13-04-2006, 01:25 AM
True.. Glicko is good because it gives much more accurate representations.

For the record, I beat Lewski in the Doeberl lightning last year after having a few drinks... either he is on the decline or drinking improves my play!

Kevin Bonham
13-04-2006, 01:30 AM
If he is as strong as his old rating he will quickly get them back...you cant be serious Kevin?:lol:

The amount of games you would have to win after losing 300+ points in one tournament is going to take you a fair while.:whistle:

Not if you are consistently performing at 300+ points above your new rating and your rating is tagged as very unreliable on account of inactivity. Glicko is very very fast in such circumstances.

Bill might be able to provide a more accurate figure but I am suspecting it would take not more than 15 games performing at the old rating to recover to within 50 points of the old rating. Indeed it may be faster than that.

Rincewind
13-04-2006, 07:54 AM
For the record, I beat Lewski in the Doeberl lightning last year after having a few drinks... either he is on the decline or drinking improves my play!

...or one game of blitz is not a statistically significant measure of chess ability.

You be the judge. ;)

Garvinator
13-04-2006, 09:20 AM
...or one game of blitz is not a statistically significant measure of chess ability.

You be the judge. ;)
or that one blitz game plus the alcohol impaires chess ability judgement:P :cool:

WhiteElephant
13-04-2006, 09:44 AM
For the record, I beat Lewski in the Doeberl lightning last year after having a few drinks... either he is on the decline or drinking improves my play!

In my first year at uni, I got totally pissed during O-Week, and turned up to one of these club open days, where clubs try to recruit new members. I staggered past the chess club where there were 3 chess nerd types sitting in front of chessboards looking very arrogant about their chess abilities. I flopped down into one of the chairs and started making moves on all 3 boards. They thought I was so smashed I didn't know which board I was playing on! Anyway I somehow ended up swindling all 3 of them, the looks on their faces were hilarious.

I have also had some marathon all night chess/ drinking sessions and I swear my chess was far better in the morning after a full night on the grog :)

four four two
13-04-2006, 09:53 AM
So are you an expert in the Stolichnoya attack or the Bacardi defense?:hmm:

Maybe the Jim Beam gambit?;)

WhiteElephant
13-04-2006, 09:58 AM
So are you an expert in the Stolichnoya attack or the Bacardi defense?:hmm:

Maybe the Jim Beam gambit?;)

During uni days, when I was a seasoned campaigner, half a slab of VBs was no problem for me. These days I stick to Southern Comfort, but after one or two drinks I am already thinking about the hangover the next day.

four four two
13-04-2006, 10:01 AM
So long as you dont go down the path of the West coast coolers...for that is the path of the DARK SIDE.:D

WhiteElephant
13-04-2006, 10:06 AM
So long as you dont go down the path of the West coast coolers...for that is the path of the DARK SIDE.:D

Nothing beats those old Tropicana casks for tackiness. They were something like 5 bucks per cask and could keep you going all night :)

Another way we used to economise was get a bottle of Jim Beam between 4 guys, go into the nearest Hungry Jacks and order 4 large cokes. We'd drink about a quarter of the cokes then surreptitiously fill them up with the Jim Beam. Voila. Instant party.

peter_parr
13-04-2006, 12:04 PM
Drug claims I am about 2000 Playchess.
Now he says I fluctuate from 1800-2200 after I proved my bullet rating is 2218 last played 10 April 2006 (4239 games).

2218 is higher then either of these (close to my ACF rating of 2227)
I do note however that your highest rating is about 2900 on ICC.

I will play very carefully against you if we meet in a tournament game in the post Glicko/Gletsos era.

antichrist
13-04-2006, 12:09 PM
That was a typo. It should have been 1998.
I have corrected the original post.

I thought I could remember Peter playing in a North Sydney Comp or something about 2-3 years ago??

Are the wheels falling off my gray matter that much?

peter_parr
13-04-2006, 12:11 PM
I see you are still making up supposed conversations.


The conversation went nothing like you have suggested

It appears that there are some signs your memory is beginning to return on this matter.

Let us hope your memory starts returning on other more important matters that have not yet been addressed (re-the 2005 NSWCA, AGM) which remains crystal clear in my memory.

Bill Gletsos
13-04-2006, 12:18 PM
It appears that there are some signs your memory is beginning to return on this matter.I remember the conversation. It is your account of it that is fanciful.

Let us hope your memory starts returning on other more important matters that have not yet been addressed (re-the 2005 NSWCA, AGM) which remains crystal clear in my memory.My memory of that is quite clear and differs from yours.

peter_parr
13-04-2006, 12:31 PM
Toby Lewski (rating 1951?) scored 2/7 against an average rating of 1725 in the Doeberl Cup (Easter) 2005. His expected score was 5.53.

He scored 3.5 points less then he should have.
On any normal rating system (FIDE) etc he would expect to lose no more than 50 rating points.

Under the ACF Glicko rating system he lost 337 rating points – down from 1951? To 1614?


Sounds completely statistically correct to me.


Will you try and answer a simple question for me which Gletsos and Drug do not answer.

If you alone were the ratings officer and this result occurred would you in your opinion adjust Lewski’s rating by 50 rating points as FIDE would or by 337 rating points as Glicko would on the 7 games played?

Simple question – (50 or 337) Your choice either FIDE system or Glicko.

Example: A customer walks into my shop and is considering playing in the next weekender. He has not played for about 8 years and checks his rating. I of course encourage him to play. If a normal (FIDE rating system) was used and he had a bad result he may lose 50 points – fair enough for a 7 game event.

It is my duty to inform him however that if he has a bad result he may lose about 350 rating points under the Glicko rating system (unique to Australia) in only 7 games.

Is our message to encourage such a player to compete again or do we inflict a very heavy penalty for inactivity.

The problem is he may not play in the weekender and if he does and loses 350 rating points we may lose him forever.

Kevin Bonham
13-04-2006, 12:54 PM
Simple question – (50 or 337) Your choice either FIDE system or Glicko.

Glicko, for the reasons given above - a player performing that far below rating is more likely simply not playing at anything like the strength of the former rating. Glicko has been shown to be a better predictor of future results than FIDE.


Example: A customer walks into my shop and is considering playing in the next weekender. He has not played for about 8 years and checks his rating. I of course encourage him to play. If a normal (FIDE rating system) was used and he had a bad result he may lose 50 points – fair enough for a 7 game event.

It is my duty to inform him however that if he has a bad result he may lose about 350 rating points under the Glicko rating system (unique to Australia) in only 7 games.

Is our message to encourage such a player to compete again or do we inflict a very heavy penalty for inactivity.

The problem is he may not play in the weekender and if he does and loses 350 rating points we may lose him forever.

There is indeed an issue here with marketing the Glicko system to inactive players. The best way to deal with it may be to tell a player that under the Glicko system an old rating is basically restarted from scratch and the player gets a new rating that is mainly determined by the new games they played but with some small influence from the old.

arosar
13-04-2006, 01:17 PM
My memory of that is quite clear and differs from yours.

How do you remember the conversation? And when did it take place?

AR

PHAT
13-04-2006, 03:37 PM
The conversation went nothing like you [Parr] have suggested.
Nobody believes YOU.

Bill Gletsos
13-04-2006, 03:52 PM
Nobody believes YOU.Four of my supposed answers to Parr's supposed 5 questions are one word answers.
Now although I may have given one word answers on a bulletin board, anyone who knows me knows I never give one word answers to those types of questions. :hand:

PHAT
13-04-2006, 04:15 PM
Now although I may have given one word answers on a bulletin board, anyone who knows me knows I never give one word answers to those types of questions. :hand:

You take two paragraphs to say "yes", two sentences to say "no" but only two words to say "what income?"

four four two
13-04-2006, 05:04 PM
Glicko has been shown to be a better predictor of future results than FIDE.


Where has this been "shown"?:hmm:
And if glicko is more accurate than the elo system why arent other countries or FIDE using it?:hmm: Are they just obstinate dinosaurs?:hmm:

Vlad
14-04-2006, 09:47 AM
Where has this been "shown"?:hmm:
And if glicko is more accurate than the elo system why arent other countries or FIDE using it?:hmm: Are they just obstinate dinosaurs?:hmm:

A "fresh" first-year student is shopping in one of the Boston grocery shops. He fills in a huge trolley of products and slowly moves to a check out counter with a sign "Less than 8 items".

The person on the check out tells him – You must be either from Harvard or MIT.

How do you know? – proudly replies the student.

Well, people in Harvard can’t count, while people in MIT can’t read. :doh:

Kevin Bonham
14-04-2006, 11:32 AM
Where has this been "shown"?:hmm:

ACF ratings officers have carried out extensive tests of predictive efficiency of the Glicko system. In fact, the system is probably better at predicting future results if made even more dynamic but this would result in occasional cases of very large errors.


And if glicko is more accurate than the elo system why arent other countries or FIDE using it?:hmm: Are they just obstinate dinosaurs?:hmm:

In FIDE's case, yes.

FIDE have only got away with having such an unresponsive rating system for so long because their ratings floors and scrapping of sub-floor results before a player has a rating (which were, between them, causing significant inflation) cancelled out the unresponsiveness of their system. Juniors gaining points more slowly than they deserved did not matter because they were often slightly overrated to begin with.

Now that FIDE have fixed the rating floor problem, the unresponsiveness one is bound to bite them and they will have to make more reforms.

Can't vouch for other nations. There are various ways to tweak an ELO system to produce effects similar to Glicko. Glicko is by no means a straightforward system to run and some federations may have avoided it on practical grounds. Others may just be slavishly following FIDE.

Vlad
14-04-2006, 11:55 AM
I think what people in Australia do not appreciate is that people who run ratings here do understand the ratings much better than rating officers in most other countries. My experience in Russia with rating officers was very bad. The same story applies to FIDE. At the best these guys can count and are able to substitute numbers into excel spreadsheet. They do not know what is mean or standard deviation. They can't compare two different rating systems. They just follow what FIDE does.

antichrist
14-04-2006, 06:44 PM
I know some players who may prefer Glicko only because it gives them a chance to win prizemoney - which is a consideration.

Is it fair to expect someone to play for ages to recover their former strength before having a chance to win prizes.

Of course there is the other side of one's pride in their high rating.

My table tennis game went right down due to inactivity and age. But on return I was put back into my old rating and have buckely's chance of ever winning anything. Is this fair?

Cat
23-04-2006, 05:25 PM
ACF ratings officers have carried out extensive tests of predictive efficiency of the Glicko system.

Yes, but where has it been shown?


In FIDE's case, yes.

Unlike the ACF???!!!:liar: :liar: :liar: :liar:


FIDE have only got away with having such an unresponsive rating system for so long because their ratings floors and scrapping of sub-floor results before a player has a rating (which were, between them, causing significant inflation) cancelled out the unresponsiveness of their system. Juniors gaining points more slowly than they deserved did not matter because they were often slightly overrated to begin with.

Whereas the ACF rating officers have got away with it by calling people names and being very,very rude.


Others may just be slavishly following FIDE.
While others worship the Glicko.

Cat
23-04-2006, 09:42 PM
I know some players who may prefer Glicko only because it gives them a chance to win prizemoney - which is a consideration.

Is it fair to expect someone to play for ages to recover their former strength before having a chance to win prizes.

Of course there is the other side of one's pride in their high rating.

My table tennis game went right down due to inactivity and age. But on return I was put back into my old rating and have buckely's chance of ever winning anything. Is this fair?

These are very good points A/C. It's really our expectations that's at fault, we should accept the rating system the way it is and learn to appreciate the special opportunities it affords. This is a very attractive benefit I'd never really thought about but it makes perfect sense.

Ride the Glicko deflation down into rating ranges that make winning prizemoney easy, in full knowledge that provided you've played enough games, a really reliable Glicko rating will hold you there forever.

For example, if Ian Rogers played enough loosing games to get his Glicko rating down to (say) 1200!!!, he'd then be able to go on to win every prize in every comp he would enter in sure knowledge the Glicko would freeze his rating there permanently. If he accepted this earnings shortfall intially, he'd triple or quadrouple his earnings for every year thereafter, or at least until the ELO came back.

I know you're a lawyer A/C, but have you ever thought of trying your hand at financial management, you're so creative?

Kevin Bonham
24-04-2006, 11:54 AM
Cat's post above is absolute fiction. :hand:

Kevin Bonham
24-04-2006, 12:08 PM
Yes, but where has it been shown?

OK, it hasn't been publicly released, so perhaps I should have said "found".


Unlike the ACF???!!!

Indeed unlike the ACF which continues to adopt and fine-tune an advanced and modern rating system.


Whereas the ACF rating officers have got away with it by calling people names and being very,very rude.

So what has Graham Saint ever done to you? :hmm:

A lot of the attacks on the Ratings Officers were very rude to begin with, not to mention uninformed. Do you really think that without the responses Bill has sometimes made, that ACF Council would have taken any different line over the Glicko?

Incidentally I note there has not been that much complaint about the Glicko system on this board lately. While there is still the odd flare-up (eg Peter Parr about inactive players) this section of the forum is much quieter than it used to be.


While others worship the Glicko.

No-one worships it; it is just a matter of fact that it is a major improvement on the ELO system. It remains imperfect and open to sensible ideas for ways to improve its predictive efficiency.

Cat
25-04-2006, 08:36 AM
OK, it hasn't been publicly released, so perhaps I should have said "found".

Publish it then , so we can all have a good laugh!


Indeed unlike the ACF which continues to adopt and fine-tune an advanced and modern rating system.

Really, so when are you going to let us know about it? What's it called? I know, we could call it the Billko!


Do you really think that without the responses Bill has sometimes made, that ACF Council would have taken any different line over the Glicko?

Unequivocally, yes!


Incidentally I note there has not been that much complaint about the Glicko system on this board lately. While there is still the odd flare-up (eg Peter Parr about inactive players) this section of the forum is much quieter than it used to be.

Muted might be a better word.



No-one worships it; it is just a matter of fact that it is a major improvement on the ELO system. It remains imperfect and open to sensible ideas for ways to improve its predictive efficiency.

Spare us pleeeease! Glicko has been repaired more times than my old Land Rover and I'd still trust my old Land Rover any day!

Cat
25-04-2006, 08:38 AM
Cat's post above is absolute fiction. :hand:

No it's not its completely real! Just ask A/C

Frank Walker
25-04-2006, 10:28 AM
He has been taking lessons from BG on how to defend the Glicko.


no, i just received a large sum of money.

Kevin Bonham
25-04-2006, 11:46 AM
Publish it then , so we can all have a good laugh!

It is up to the Ratings Officers to decide what they wish to publish. Why don't you try asking them? :lol:


Really, so when are you going to let us know about it? What's it called? I know, we could call it the Billko!

Just more mindless trolling from someone who has a grudge against the Glicko presumably because his "report" on it (like other babblings) was given the short shrift it deserved in terms of its numerous defects.


Unequivocally, yes!

Your delusions grow both more and less interesting by the day.


Spare us pleeeease! Glicko has been repaired more times than my old Land Rover and I'd still trust my old Land Rover any day!

If changes aren't made then you pretend the ACF is not responsive to criticism of the system. If changes are made you pretend it has been tweaked too often. Either way you're just a troll. :hand:

qpawn
25-04-2006, 12:34 PM
As someone with tertiary maths qualifications I am not satisfied with Kevin Bonham's claim that "extensive tests" have been done with the Glicko system: where is a link that shows such a statistical comparison of ELO and Glicko? I am not having a go at Kevin; I just want to know from a quantitative standpoint how the glicko stands up.

From what I know I can only conclude that whatever current rating system being used is a total sham and failure. When I was at the MCC weekender there were players VERY scared of the number of points you can lose in one tournamnet; it is DEFINITELY enough to put people off playing. I can share that feeling; I am happy to lose games if I learn something and hence in reality get better. However, numerically, it looks as if I have got worse!! :eek: Also, what is the aim of a rating system? Surely it's to give a player some idea of where they stand in ability. Yet my rating is an absolute joke; in that event I took a player approx 600 ELO points above me to 3 and a half hours and lost in a time scramble when I was equal if not better. I am left with NO IDEA where I stand. Then there was the last game where I played someone on a rating as measly as mine. There was no way that either of us played to that rating; we played a lot better.

Wouldn't a better method of setting a provisional rating be to have an initial randomised tactics/strategy test with the result correlated to ELO? At least then one's initial rating is done objectively instead of blind GUESSWORK.

This is how initial ratings may as well be set in the current system:

Ratings officer: Joe, what do you think your rating is?
Joe : I dunno.
Officer: OK we will do it like this. I will take the number of pretty girls you have seen in one night stands, add in your bloodcount, and then multiply your starsign number by 5. That will give you your rating.
JOE: Sounds groovy .

:lol:

bergil
25-04-2006, 12:42 PM
Put the ego away and just play the game of chess. How bloody ridiculous is it to not play in fear of losing ratings points? They are pathetic and no loss to the game!! :hand:

Kevin Bonham
25-04-2006, 01:37 PM
Well look who's back. :D Someone who a mere twelve days ago wrote "It is quite clear that I have no interest in posting here anymore" and that he had "gone to other chess sites where I am having a far more meaningful exchange of chess ideas than here." What a pathetic lack of resolve! Let me advise you, don't ever take up smoking. :lol:


As someone with tertiary maths qualifications I am not satisfied with Kevin Bonham's claim that "extensive tests" have been done with the Glicko system: where is a link that shows such a statistical comparison of ELO and Glicko? I am not having a go at Kevin; I just want to know from a quantitative standpoint how the glicko stands up.

I got the information about testing from extensive discussions with the ACF ratings officers. One of them, a Mr Gletsos I believe, has been rumoured to post here now and then.

You may find many items on Professor Mark Glickman's site (http://math.bu.edu/people/mg/ratings.html) of interest although they don't specifically relate to your question.

Of course you will be all at sea here since this is an applied maths issue and your degree is in pure maths ("calculus, optimisation, matrices" etc). A ratings system is a real-world exercise in problem solving, and by your own admission, "I have one serious limitation: I can only do maths out of a textbook. I cannot problem solve at all."


From what I know I can only conclude that whatever current rating system being used is a total sham and failure. When I was at the MCC weekender there were players VERY scared of the number of points you can lose in one tournamnet; it is DEFINITELY enough to put people off playing.

You seem to be forgetting that the point of a ratings system is to assess player skill and therefore if players are playing far more strongly or weakly than their previous rating indicates (a common issue with ratings based on a small number of games or where a player has not played for a while) then large ratings changes are to be expected.


I can share that feeling; I am happy to lose games if I learn something and hence in reality get better.

If losing a game has caused you to improve then this will be reflected in your further results and your rating will soon head upwards. This is what is great about Glicko; if you have a shocker and lose a pile of points you can recover them very quickly if you are as good as your old rating said.


Also, what is the aim of a rating system? Surely it's to give a player some idea of where they stand in ability. Yet my rating is an absolute joke; in that event I took a player approx 600 ELO points above me to 3 and a half hours and lost in a time scramble when I was equal if not better.

All the rating system cares about is that you lost. It draws no distinction between losing in 10 moves and losing in 100.

At our club there is a player rated around 1300 who in terms of natural chess talent is probably a 1650 or 1700 player. Yet his time management is poor, causing him to frequently lose on time or settle for draws against seemingly weaker opposition in good positions. Hence his rating stays down. The purpose of the rating system is to assess your performance as a tournament player. Ability to win time scrambles is a part of that.


I am left with NO IDEA where I stand. Then there was the last game where I played someone on a rating as measly as mine. There was no way that either of us played to that rating; we played a lot better.

Post the game!


Wouldn't a better method of setting a provisional rating be to have an initial randomised tactics/strategy test with the result correlated to ELO? At least then one's initial rating is done objectively instead of blind GUESSWORK.

But it then fails to take into account factors relevant to tournament performance such as opening preparation, concentration, performance under pressure, match toughness, time allocation and everything else.

The system uses tournament results to predict tournament results. It does not use tournament results to assess whether people are grandmasters when nobody is watching.

Initial ratings are very inaccurate because they are based on limited data. However because of the dynamism of the Glicko system and its incorporation of volatility and reliability factors, if your start rating is inaccurate it rapidly gets scrubbed. In ELO you start with an inaccurate rating and are stuck with its legacy for years.

Cat
25-04-2006, 03:18 PM
Well look who's back. :D Someone who a mere twelve days ago wrote "It is quite clear that I have no interest in posting here anymore" and that he had "gone to other chess sites where I am having a far more meaningful exchange of chess ideas than here." What a pathetic lack of resolve! Let me advise you, don't ever take up smoking. :lol:

Up to your old bullying tricks, eh? qpawn asks a perfectly reasonable question and you can't help being snide. Why aren't the moderators stepping in?



You may find many items on Professor Mark Glickman's site (http://math.bu.edu/people/mg/ratings.html) of interest although they don't specifically relate to your question.

The Australian Glicko is an unrecognisable mutant deviant of the original, so your link is not much help.


Of course you will be all at sea here since this is an applied maths issue and your degree is in pure maths ("calculus, optimisation, matrices" etc). A ratings system is a real-world exercise in problem solving, and by your own admission, "I have one serious limitation: I can only do maths out of a textbook. I cannot problem solve at all."

What arrogance! And you wonder why people don't take you seriously? The problem is the real world doesn't fit into the Glicko capsule.


You seem to be forgetting that the point of a ratings system is to assess player skill and therefore if players are playing far more strongly or weakly than their previous rating indicates (a common issue with ratings based on a small number of games or where a player has not played for a while) then large ratings changes are to be expected.



If losing a game has caused you to improve then this will be reflected in your further results and your rating will soon head upwards. This is what is great about Glicko; if you have a shocker and lose a pile of points you can recover them very quickly if you are as good as your old rating said.



All the rating system cares about is that you lost. It draws no distinction between losing in 10 moves and losing in 100.

At our club there is a player rated around 1300 who in terms of natural chess talent is probably a 1650 or 1700 player. Yet his time management is poor, causing him to frequently lose on time or settle for draws against seemingly weaker opposition in good positions. Hence his rating stays down. The purpose of the rating system is to assess your performance as a tournament player. Ability to win time scrambles is a part of that.


But it then fails to take into account factors relevant to tournament performance such as opening preparation, concentration, performance under pressure, match toughness, time allocation and everything else.

The system uses tournament results to predict tournament results. It does not use tournament results to assess whether people are grandmasters when nobody is watching.

Initial ratings are very inaccurate because they are based on limited data. However because of the dynamism of the Glicko system and its incorporation of volatility and reliability factors, if your start rating is inaccurate it rapidly gets scrubbed. In ELO you start with an inaccurate rating and are stuck with its legacy for years.

We know what it's suppose to do, but it's like buying a car from a Skoda dealer, never quite lives up to expectations.

Vlad
25-04-2006, 03:35 PM
As someone with tertiary maths qualifications ...
:lol:

Just to have an idea about how tertiary your qualifications are tell me what is the derivative of function "x to the power of x".

P.S. This is a problem I give to "smart" %$$&$ in my graduate math class when they get bored with the routine material.

qpawn
25-04-2006, 04:22 PM
I am not going to reciprocate Kevin's abuse. I have no doubt that the PM message he has sent to me is abusive so I have not read it.

Clearly we have very different ideas about what chess ability is and how it should be measured. At least my skepticism towards the rating system has 1 benefit for me; I take every opponent seriously whatever his or her rating. I also go into every game with every chance of winning; even a titled player can make mistakes or have gaps in their skillset. For an example of a GM who has huge gaps in his skillset I would take Nakamura. That guy's positional understanding is pathetic. How he got to 2600 ELO I do not know. If I played him in a positional game I would fancy my chances of getting a full point.

I understand chess extremely well in spite of my rating. I will seek to improve my other understanding on other sites. Bye.

Lucena
25-04-2006, 05:39 PM
I am not going to reciprocate Kevin's abuse. I have no doubt that the PM message he has sent to me is abusive so I have not read it.

Clearly we have very different ideas about what chess ability is and how it should be measured. At least my skepticism towards the rating system has 1 benefit for me; I take every opponent seriously whatever his or her rating. I also go into every game with every chance of winning; even a titled player can make mistakes or have gaps in their skillset. For an example of a GM who has huge gaps in his skillset I would take Nakamura. That guy's positional understanding is pathetic. How he got to 2600 ELO I do not know. If I played him in a positional game I would fancy my chances of getting a full point.

I understand chess extremely well in spite of my rating. I will seek to improve my other understanding on other sites. Bye.

But you didn't answer Drug's question! I will have to answer on your behalf.

d/dx (x^n) = n.x^(n-1)

so d/dx (x^x) = x.x^(x-1)

=x^x :owned: :uhoh:

Watto
25-04-2006, 05:44 PM
From what I know I can only conclude that whatever current rating system being used is a total sham and failure. When I was at the MCC weekender there were players VERY scared of the number of points you can lose in one tournamnet; it is DEFINITELY enough to put people off playing. I can share that feeling; I am happy to lose games if I learn something and hence in reality get better. However, numerically, it looks as if I have got worse!! :eek: Also, what is the aim of a rating system? Surely it's to give a player some idea of where they stand in ability. Yet my rating is an absolute joke; in that event I took a player approx 600 ELO points above me to 3 and a half hours and lost in a time scramble when I was equal if not better. I am left with NO IDEA where I stand. Then there was the last game where I played someone on a rating as measly as mine. There was no way that either of us played to that rating; we played a lot better.
gpawn... I match opponents rated 600 points above me quite often... quite often I even end up with a theoretically won position but the fact is at the moment I'm losing those positions far far more often than I win them- through loss of concentration, blunders, losing the thread, not finding the right plan whatever- simply not being good enough for the length of the game. Until I start winning those positions there's not much point in my whinging about my rating. Your rating is a reflection of your results, not your strength at any given moment. It's a low rating but, well, it will go up if your results improve. If they don't, well tough.
The relative merits of different rating systems are a separate issue and to be honest I don't think your low rating is really relevant. Any rating system would reflect on the result of the game played, not the apparent quality. If you always play like a grandmaster but invariably at the crucial moment play like a demented grandmaster, thus losing the game, your rating will reflect that sad and tragic result and only that. ;)


Clearly we have very different ideas about what chess ability is and how it should be measured. At least my skepticism towards the rating system has 1 benefit for me; I take every opponent seriously whatever his or her rating. I also go into every game with every chance of winning; even a titled player can make mistakes or have gaps in their skillset. For an example of a GM who has huge gaps in his skillset I would take Nakamura. That guy's positional understanding is pathetic. How he got to 2600 ELO I do not know. If I played him in a positional game I would fancy my chances of getting a full point.

I understand chess extremely well in spite of my rating. I will seek to improve my other understanding on other sites. Bye.
I agree with playing the position not the opponent’s rating but your point about Nakamura: I'm assuming you're joking ... or else you've never analysed a game with a strong player. It's like expecting to match a Davis Cup player at tennis when you're a social doubles player. There would be no contest.
p.s. Go to other sites by all means, but come back sometimes for a visit. :)

PHAT
25-04-2006, 05:58 PM
How bloody ridiculous is it to not play in fear of losing ratings points? They are pathetic and no loss to the game!! :hand:

Speaking for myself, I have no fear at all about my rating rising/falling. However, many (most?) players have a lot of self worth invested in that mear four digit number.

You may think them "pathetic" - but never ever say that they would be "no loss to the game." That is not what a potential state president should ever say. Wake up dude, before you alienate half of your potentialy voters.

Vlad
25-04-2006, 06:03 PM
But you didn't answer Drug's question! I will have to answer on your behalf.

d/dx (x^n) = n.x^(n-1)

so d/dx (x^x) = x.x^(x-1)

=x^x :owned: :uhoh:

Sorry mate, but that is not correct answer.:)

Vlad
25-04-2006, 06:21 PM
I'm assuming you're joking ... or else you've never analysed a game with a strong player.

A very good point, but unfortunately for some people it is not persuasive enough.:((

Here is an example. From time to time I play socially in the Hyde Park or the Spanish Club and quite often the rating difference for me and my opponents is at least 500 points. So, the way games usually go - at first I do not think too much; if I get in trouble then I start thinking hard. Quite often I can save games when I am a rook or even a queen down. So some players get impression that they are easily beating me and just extremely unlucky to blunder in the last moment. Especially persuasive was in that regard one of the players who kept saying that I can't play chess and he is just extremely unlucky. So at some moment I started counting how many times straight he will loose. During a few weeks I managed to count to 17. I lost my 18-th game at the end of the evening in the Spanish Club when I was completely drunk. :owned:

PHAT
25-04-2006, 06:27 PM
Well look who's back. :D Someone who a mere twelve days ago wrote "It is quite clear that I have no interest in posting here anymore" and that he had "gone to other chess sites where I am having a far more meaningful exchange of chess ideas than here." What a pathetic lack of resolve! Let me advise you, don't ever take up smoking. :lol:



I will seek to improve my other understanding on other sites. Bye.

Well look who's bullying people off the BB. Someone a mere year ago (you) was telling us all how I frighten off new posters. :hand:

antichrist
25-04-2006, 06:51 PM
KB: ......The purpose of the rating system is to assess your performance as a tournament player. Ability to win time scrambles is a part of that...........

A/C
And whether that time scramble happens to be guillotine finish so what?

A time scramble with a digital clock is still a time scramble.

bergil
26-04-2006, 04:36 AM
Speaking for myself, I have no fear at all about my rating rising/falling. However, many (most?) players have a lot of self worth invested in that mear four digit number.

You may think them "pathetic" - but never ever say that they would be "no loss to the game." That is not what a potential state president should ever say. Wake up dude, before you alienate half of your potentialy voters.They are not potential voters because they are not prepared to front up. While its true the game would be better off with greater participation, what does the game really get from those that are so shallow? :hmm:

PHAT
26-04-2006, 08:48 AM
While its true the game would be better off with greater participation, what does the game really get from those that are so shallow? :hmm:

You have answered your own question - "greater participation."

There are thresholds in the numbers of people required to:

1. Initiate self sustaining growth of community interest in X.
2. Initiate irreversable decine of community interest in X.

I don't know about you, but I am in favor of chasing (1) because (2) is the ace of spades.

four four two
26-04-2006, 09:52 AM
They are not potential voters because they are not prepared to front up. While its true the game would be better off with greater participation, what does the game really get from those that are so shallow? :hmm:

You think people who think about their rating are shallow?...do you tell most juniors they are shallow?:hmm:

bergil
26-04-2006, 11:30 AM
You think people who think about their rating are shallow?...do you tell most juniors they are shallow?:hmm:
Golems who refuse to play rated games because they fear to lose their precious rating and are only concern about their rating for whatever perceived status it gives them are indeed shallow. :whistle:

I made no mention of juniors, which is typical of your usual lack of attentiveness when replying to a post. :doh: But I doubt very much they are only playing to have a rating and not because they enjoy playing. Juniors are not scared about losing their rating because they are not scared their best chess is behind them. :hand:


You have answered your own question - "greater participation."
There are thresholds in the numbers of people required to:

1. Initiate self sustaining growth of community interest in X.
2. Initiate irreversable decine of community interest in X.

I don't know about you, but I am in favor of chasing (1) because (2) is the ace of spades.
Agreed but I don't see ELO changing that.

Kevin Bonham
26-04-2006, 11:59 AM
Up to your old bullying tricks, eh? qpawn asks a perfectly reasonable question and you can't help being snide. Why aren't the moderators stepping in?

Possibly because qpawn has some ?! form on this board including unreasonable attacks on the board as a whole and engaging in pointless insults just because he doesn't like the way the debate is going (http://chesschat.org/showpost.php?p=95229&postcount=435).

Perhaps if he reaches the other end of the board the side controlling him will get the choice whether to demote him to a Cat, a firegoat or a Sweeney, although given that he often moves backwards I feel that this is unlikely to ever occur.


The Australian Glicko is an unrecognisable mutant deviant of the original, so your link is not much help.

I suggest if you asked Prof Glickman for his view on this he would disagree with you. It is however worth noting that the ACF employs Glicko-2 which is discussed on said site.


What arrogance! And you wonder why people don't take you seriously? The problem is the real world doesn't fit into the Glicko capsule.

Explain. (Actually, don't. We have been through enough of your failed attempts before and you are clearly hopelessly biased against it.)


We know what it's suppose to do, but it's like buying a car from a Skoda dealer, never quite lives up to expectations.

More likely your expectations of what any rating system can do on the data available to it are unrealistic.

four four two
26-04-2006, 12:28 PM
Juniors are not scared about losing their rating because they are not scared their best chess is behind them. :hand:


Most juniors are obsessed with their current rating...and their future rating. They are concerned with performing well amongst their peers. :whistle:

Kevin Bonham
26-04-2006, 12:29 PM
I am not going to reciprocate Kevin's abuse. I have no doubt that the PM message he has sent to me is abusive so I have not read it.

Well, firstly I don't believe you when you claim not to have read it any more than I believe you when you say you will not post here any more.

Secondly the only PM I ever sent you was sent on 3 April in response to a hilarious and very abusive PM you sent me the same day. I wonder if you would grant me permission to post your PM on the board. :hmm:

Thirdly, as for abuse in my comments above, that you can't stay away despite saying you will is demonstrably true thus far, and the rest was all in your own words. :lol:


Clearly we have very different ideas about what chess ability is and how it should be measured. At least my skepticism towards the rating system has 1 benefit for me; I take every opponent seriously whatever his or her rating.

You are actually hardly alone in this. I do the same to a degree, not because I am sceptical about the rating system but because I know that (i) any rating system can only do so much and (ii) even an opponent you are predicted to score 99% against should not be taken for granted.


That guy's positional understanding is pathetic. How he got to 2600 ELO I do not know. If I played him in a positional game I would fancy my chances of getting a full point.

Nobody else would. :D


I will seek to improve my other understanding on other sites. Bye.

We've heard that line before. See ya later. :lol:

Which brings me to:


Well look who's bullying people off the BB. Someone a mere year ago (you) was telling us all how I frighten off new posters.

Which you do, brand new posters who arouse your suspicions in some (often spurious) way can certainly bring out your worst, eg your disgraceful treatment of fletch. A poster who has been here posting heavily for months and has Coffee Lounge access is hardly a new poster, and as the link above shows qpawn has been dishing abuse out hence ought to expect it back.

Furthermore far from me "bullying" (ah, that lame PC cliche de jour again) him off, it seems he cannot stay away! :owned:

PHAT
26-04-2006, 05:31 PM
Agreed but I don't see ELO changing that.
I have NEVER EVER said Elo ws better than Glicko. However, they are at different degrees of shithouse.

WhiteElephant
28-04-2006, 01:04 PM
Golems who refuse to play rated games because they fear to lose their precious rating and are only concern about their rating for whatever perceived status it gives them are indeed shallow. :whistle:


I think many people play because of a desire to increase their ratings and I don't see a problem with that.

Ratings are the primary measure of an individual's improvement. It is fun to watch your rating increase and compare your own rating to that of others.

It is a great thrill to beat someone with a higher rating than you.

When I play chess on the internet, I often tell myself that I will keep on playing until my rating reaches a particular level (which has resulted in a sleepless night or two hehe).

You would like to think that some people play for the beauty of the game etc blah blah, which is fantastic, but the reality is that many people loive to see their rating increase. Why the animosity towards those people, bergil?

bergil
28-04-2006, 02:11 PM
I think many people play because of a desire to increase their ratings and I don't see a problem with that. True but that's not the only reason anyone plays.


Ratings are the primary measure of an individual's improvement. It is fun to watch your rating increase and compare your own rating to that of others.Not quite, rating are an indication of your playing strength. If you play above you'll increase if below decrease. It is still fun to watch it slide?


It is a great thrill to beat someone with a higher rating than you.
That's because you have boosted your ego by besting someone considered according to their rating as better than you and you are proud of it. Fair enough, most do and there doesn't seem anything wrong with that.


You would like to think that some people play for the beauty of the game etc blah blah, which is fantastic, but the reality is that many people love to see their rating increase. Why the animosity towards those people, bergil?People play for any number of reasons, purely for enjoyment, to be social, the competitiveness, challenge of the game and so on.

Yes they do but not many enjoy the decrease but don't stop playing solely because of it. In your reality I must be playing with a host of egotists and when I consider those I know in the game, I find that's just not true or at least not my reality.

There is nothing wrong in being proud of your rating and the results you have gained to achieve it. But to turn around and whinge that you don't want to play agian because I might lose my current pecious rating is childish and not to be encouraged.

four four two
28-04-2006, 02:19 PM
I dont think its the loss of ratings points as such but the way Glicko savages people for being inactive that is the issue.
People have many reasons for being inactive for a period of over a year...family,work,study,health. Should they be punished for being inactive?:hmm:

If you didnt play for a year Bergil and then had an ordinary "comeback" tournament that caused you to lose 150+ points,which would consequently affect how you will be paired against players for the next couple of tournaments,would you be a little annoyed or indifferent?:hmm:

bergil
28-04-2006, 02:37 PM
I dont think its the loss of ratings points as such but the way Glicko savages people for being inactive that is the issue.
People have many reasons for being inactive for a period of over a year...family,work,study,health. Should they be punished for being inactive?:hmm: Its not punishment but an indication of performance.


If you didnt play for a year Bergil and then had an ordinary "comeback" tournament that caused you to lose 150+ points,which would consequently affect how you will be paired against players for the next couple of tournaments,would you be a little annoyed or indifferent?:hmm:
Mate I have both a large increase of 76 points and and a larger decrease of 106 points happen to me and yes I am indifferent.

four four two
28-04-2006, 02:50 PM
Its not punishment but an indication of performance.


So why doesnt a player who hasnt played in 3mths and a player who hasnt played in over a year have their games rated the same way?:hmm:

Isnt the reason they are rated differently because of inactivity,rather than performance?:hmm:

If it was performance a player who plays semi regularly should have their games rated the same way as someone who has had to take a break from tournament chess. They lost to the same level of player and yet lose more rating points. If that isnt punishment,what is?:hmm:

Kevin Bonham
28-04-2006, 03:02 PM
So why doesnt a player who hasnt played in 3mths and a player who hasnt played in over a year have their games rated the same way?:hmm:

Because the less fresh a result the less reliable it is as an indicator of current playing strength. If someone doesn't play for years they might return at at least the same strength (as Igor Goldenberg certainly has, and I can think of some other cases) but more often they play much more weakly on return. If a player returns a poor result on return from a long break the most likely cause is the long break and not just it being an unlucky bad result.

Rincewind
28-04-2006, 03:11 PM
I don't think one can or should argue that that a player is more likely to play at a lower strength on their return. Afterall, the lack of rated games in the ACF list could be due to them being overseas playing all over the place and becoming strong.

IMHO Glickman argues persuasively in para 2 of http://math.bu.edu/people/mg/glicko/glicko.doc/glicko.html where he says...

The problem with the Elo system that the Glicko system addresses has to do with the reliability of a player's rating. Suppose two players, both rated 1700, played a tournament game with the first player defeating the second. Under the US Chess Federation's version of the Elo system, the first player would gain 16 rating points and the second player would lose 16 points. But suppose that the first player had just returned to tournament play after many years, while the second player plays every weekend. In this situation, the first player's rating of 1700 is not a very reliable measure of his strength, while the second player's rating of 1700 is much more trustworthy. My intuition tells me that (1) the first player's rating should increase by a large amount (more than 16 points) because his rating of 1700 is not believable in the first place, and that defeating a player with a fairly precise rating of 1700 is reasonable evidence that his strength is probably much higher than 1700, and (2) the second player's rating should decrease by a small amount (less than 16 points) because his rating is already precisely measured to be near 1700, and that he loses to a player whose rating cannot be trusted, so that very little information about his own playing strength has been learned.

- Prof Mark Glickman


To my way of thinking, you are modelling a stocastic process and the longer the period since the last measurement the lower the reliability of that measure. This is essentially what the RD factor and its aging is all about.

four four two
28-04-2006, 03:13 PM
Strength is one thing...performance is another. You only have to consider Gary Lane's performance in Doeberl.

While I would agree with you Kevin that players who have had a prolonged absence from tournament chess[3 yrs or more] generally dont play at the strength of their last rating ,I dont believe they are 200 points+ weaker in strength either ...which is what will happen to their rating if they have a very ordinary tournament when they first come back.

Kevin Bonham
28-04-2006, 06:08 PM
As for the question asked by four four two in the title of his post, the dynamism of Glicko is already reduced for top players (Bill could tell you the cutoff). It was incorporated because it is harder to maintain a rating where it is if you are at the very top of a pool, so for the top players the Glicko system is implemented in a less dynamic fashion. I am hazy on the fine details so ask Bill if you want to know.


I don't think one can or should argue that that a player is more likely to play at a lower strength on their return. Afterall, the lack of rated games in the ACF list could be due to them being overseas playing all over the place and becoming strong.

The latter bit is true, and you would also have cases where a player who first played as a very weak junior later returned to play as a stronger adult. However I'm predicting that empirically, players who have not played a rated game in, say, >4 years, will perform on average below their old rating. It would be interesting to see some data on this.

Also if a player's rating gets greatly reduced by a very poor performance on return, it will go up rapidly with more games with performance at their actual strength.

WhiteElephant
28-04-2006, 06:17 PM
Ok, I offer myself as a guinea pig. I am about to play a tournament at Box Hill. It is my second tournament in about 10 years and I currently have a ?? next to my rating.

This means if I have a bad tournament, I'm screwed. Since I am the second seed, chances are my rating will drop. But I am curious to see how I perform in my comeback tournament and how my rating will be affected.

Rincewind
28-04-2006, 06:36 PM
Ok, I offer myself as a guinea pig. I am about to play a tournament at Box Hill. It is my second tournament in about 10 years and I currently have a ?? next to my rating.

This means if I have a bad tournament, I'm screwed. Since I am the second seed, chances are my rating will drop. But I am curious to see how I perform in my comeback tournament and how my rating will be affected.

Not sure what value you have in being a guinea pig. Hypotheticals on the effect of performance vs rating change are easily performed and just as if not more enlightening. Regarding the strength of players returning to tournament chess, every case is different and the results will depend on the specifics of your absence.

Good luck in the tourny.

WhiteElephant
28-04-2006, 06:44 PM
Not sure what value you have in being a guinea pig. Hypotheticals on the effect of performance vs rating change are easily performed and just as if not more enlightening. Regarding the strength of players returning to tournament chess, every case is different and the results will depend on the specifics of your absence.

Good luck in the tourny.

Thanks :)

I am hoping that since I play a fair bit of internet chess and work in the industry my play won't be too adversely affected.

four four two
28-04-2006, 06:55 PM
Any predictions on how many points you will lose WE?:hmm:
Under 100 or over 100?:hmm: ;)

WhiteElephant
28-04-2006, 07:04 PM
Any predictions on how many points you will lose WE?:hmm:
Under 100 or over 100?:hmm: ;)

I'm pretty worried because it's a round robin and there are talented juniors galore - Chris Wallis, Eugene Schon, Derek & Sally Yu. Everyone is going on about how much Sally has improved so I will need to be in top form. A 100 point drop is being kind :)

Bill Gletsos
28-04-2006, 07:18 PM
Ok, I offer myself as a guinea pig. I am about to play a tournament at Box Hill. It is my second tournament in about 10 years and I currently have a ?? next to my rating.That is incorrect.
You dont currently have a ?? next to your rating.
In fact since symbols were added to ratings in December 2000 your rating only had a ?? next to it in the September & December 2004 rating periods.

You had been inactive between the Decmber 1997 rating period and the March 2005 rating period.

Your rating in December 1997 was 1740.
In April 2000 you received the 150 point uplift so your rating was listed as 1890.

In December 2000 it was published as 1890 with no symbol.
In August 2002 your published rating became 1890?
In September 2004 it became 1890??

In the March 2005 rating period you played 5 rated games with a performance of 2003 and you rating increased to 1942 with no symbol.

In the September 2005 list it changed to 1942?

Your current March 2006 rating is listed as 1942?

WhiteElephant
28-04-2006, 07:32 PM
That is incorrect.
You dont currently have a ?? next to your rating.
In fact since symbols were added to ratings in December 2000 your rating only had a ?? next to it in the September & December 2004 rating periods.

You had been inactive between the Decmber 1997 rating period and the March 2005 rating period.

Your rating in December 1997 was 1740.
In April 2000 you received the 150 point uplift so your rating was listed as 1890.

In December 2000 it was published as 1890 with no symbol.
In August 2002 your published rating became 1890?
In September 2004 it became 1890??

In the March 2005 rating period you played 5 rated games with a performance of 2003 and you rating increased to 1942 with no symbol.

In the September 2005 list it changed to 1942?

Your current March 2006 rating is listed as 1942?

Aaah ok. I thought it was ?? last time I checked but you are right. Good news for me, my rating won't drop as much as I thought :owned:

Vlad
28-04-2006, 07:43 PM
As for the question asked by four four two in the title of his post, the dynamism of Glicko is already reduced for top players (Bill could tell you the cutoff). It was incorporated because it is harder to maintain a rating where it is if you are at the very top of a pool, so for the top players the Glicko system is implemented in a less dynamic fashion. I am hazy on the fine details so ask Bill if you want to know.


Ohh, well, now I understand why these guys are so hugely overrated; say Lane or Wohl, for example. Everybody else has his ACF rating below his FIDE rating. These guys have them vice versa. That clearly means Bill should adjust the coefficient the other way 442 is suggesting.:hmm:

Kevin Bonham
28-04-2006, 07:44 PM
WE: And you have the 52 points you gained on your previous return to act as a buffer! In a system other than Glicko you would have gained less points for that performance.

WhiteElephant
28-04-2006, 07:49 PM
WE: And you have the 52 points you gained on your previous return to act as a buffer! In a system other than Glicko you would have gained less points for that performance.

Yeah I know, I am not complaining about the huge rating swings for inactive players as last time it worked in my favour. I am just saying that I am worried about what CAN happen this time :) Of course, if I play well, it might bump me over 2000 which is my goal.

Kevin Bonham
28-04-2006, 07:50 PM
Ohh, well, now I understand why these guys are so hugely overrated; say Lane or Wohl, for example. Everybody else has his ACF rating below his FIDE rating. These guys have them vice versa. That clearly means Bill should adjust the coefficient the other way 442 is suggesting.:hmm:

Wohl is a special case - he was inactive during a period in which the top end of the list was generally shedding points (I believe this was because the Glicko system was compacting the rating pool after it had become overstretched by the 336 point rule and other ELO issues, but there are a range of theories).

Bill Gletsos
28-04-2006, 08:11 PM
Wohl is a special case - he was inactive during a period in which the top end of the list was generally shedding points (I believe this was because the Glicko system was compacting the rating pool after it had become overstretched by the 336 point rule and other ELO issues, but there are a range of theories).Actually Wohl was 2461 on the July 2000 FIDE list. He was 2485 on the ACF August 2000 list which was an Elo based list.

Wohl's FIDE rating then went on a bit of a roller coaster ride dropping to 2371 in July 2001, rising again to 2422 in October 2002 before again dropping to 2374 in July 2003, before again rising to 2399 in April 2004 again dropping to 2359 in October 2004 before increasing steadily to 2452 in January 2006.

Currently Wohl is rated 2458 on the ACF list and 2432 on the FIDE list.

four four two
28-04-2006, 08:58 PM
Ohh, well, now I understand why these guys are so hugely overrated; say Lane or Wohl, for example. Everybody else has his ACF rating below his FIDE rating. These guys have them vice versa. That clearly means Bill should adjust the coefficient the other way 442 is suggesting.:hmm:

I dont think I was suggesting Gary Lane was overrated,I was however trying to illustrate to Bergil that Lane was a 2400+strength player despite his performance in Doeberl.

If Lane hadnt haved played for 1-2 years and had his games rated the same way as the average club player, his rating would have dropped by a fair margin. The question you have to ask yourself is,if the Glicko fluctuations are acceptable for a 2000/1800/1600 player then why are they not appropiate for players above 2200?

Could it be that the "inactivity" factor in our version of Glicko is flawed?:hmm:

Garvinator
28-04-2006, 09:05 PM
The question you have to ask yourself is,if the Glicko fluctuations are acceptable for a 2000/1800/1600 player then why are they not appropiate for players above 2200?

Could it be that the "inactivity" factor in our version of Glicko is flawed?:hmm:
I believe that part of the reason why there is a difference for the top players in Australia is that they would have so few opportunities to get those points back.

It would take them about 100 wins or even more against the general chess masses to get the points back. They have very few opportunities to play against players who have higher acf ratings than themselves.

four four two
28-04-2006, 09:23 PM
Well if a 2200+ player was to drop a ton of points from a tournament for being inactive for a couple of years then they should be able to recover them quickly if they play to their previous rating ,that is if you follow Kevin's rating theory.:whistle:

As for enough tournaments...well there's Doeberl,Begonia,NSW and Vic championships,Vic open,etc.;)

Bill Gletsos
28-04-2006, 09:33 PM
I believe that part of the reason why there is a difference for the top players in Australia is that they would have so few opportunities to get those points back.

It would take them about 100 wins or even more against the general chess masses to get the points back. They have very few opportunities to play against players who have higher acf ratings than themselves.Given you dont know what you are talking about you shouldnt just mindlessly speculate.

Garvinator
28-04-2006, 09:34 PM
As for enough tournaments...well there's Doeberl,Begonia,NSW and Vic championships,Vic open,etc.;)
I dont agree. Out of the tournaments you mention plus a few more like Gold Coast Open, Gary Lane would play very few, if any, players around his rating. Certainly not enough to get back the amount of points lost. He certainly wouldnt get to play anyone above 2645 acf;)

Garvinator
28-04-2006, 09:35 PM
Given you dont know what you are talking about you shouldnt just mindlessly speculate.
feel free to correct so we can all learn:D

Kevin Bonham
28-04-2006, 09:47 PM
Actually Wohl was 2461 on the July 2000 FIDE list. He was 2485 on the ACF August 2000 list which was an Elo based list.

Wohl's FIDE rating then went on a bit of a roller coaster ride dropping to 2371 in July 2001, rising again to 2422 in October 2002 before again dropping to 2374 in July 2003, before again rising to 2399 in April 2004 again dropping to 2359 in October 2004 before increasing steadily to 2452 in January 2006.

Currently Wohl is rated 2458 on the ACF list and 2432 on the FIDE list.

What I had in mind was the period where Wohl was well over 2500 ACF. Perhaps this was just a result of him having gone inactive with a high ACF rating and then had a few good results on return. For instance, he was 2538 ACF before losing 80 points in the last list, and, GM-strength as many of his recent overseas results have been, I'm not sure he's quite that good.

Did Wohl get the 70 points or was he just resuming on 2485 and then going to 2509 in March 05 on the basis of a good result?

Kevin Bonham
28-04-2006, 09:52 PM
Well if a 2200+ player was to drop a ton of points from a tournament for being inactive for a couple of years then they should be able to recover them quickly if they play to their previous rating ,that is if you follow Kevin's rating theory.:whistle:

Bill can clarify (if he wants) whether the different treatment at the top of the list applies to those with unreliable ratings or not. I'm not at all sure whether it does.


I dont think I was suggesting Gary Lane was overrated,I was however trying to illustrate to Bergil that Lane was a 2400+strength player despite his performance in Doeberl.

Yes, but what's the relevance? We know a regularly active player has good and bad days and they are no big deal; it doesn't mean a good or bad performance by an inactive player should be similarly dismissed as just having a good or bad event.

Bill Gletsos
28-04-2006, 09:52 PM
feel free to correct so we can all learn:DPut as simply as possible it is this.

The opportunity/possibility for a player who loses rating points to regain those points is less for players at the top end as opposed to players at other positions in the rating list.

We noted that if a player rated say 1700 lost to a player rated 1400 (a 15% chance) he could get all his points back in one game by beating a player rated 2000 (also a 15% chance). However if you are rated 2400 and lose to a 2100 then to get the points back you would need to beat a 2700. There are no 2700 rated players in Australia. Testing showed that there was a slight disadvantage to the top players compared to the vast majority of other players. To try and alleviate this we implemented a sliding scale where the points lost or gained were scaled depending on the players rating at the start of the rating period. If your rating was equal to or less than 2200 then no scaling was applied. If your rating was 2300 and greater then a scale of 0.75 was applied. For ratings between 2200 and 2300 this was a straight line from 1.0 to 0.75.

Bill Gletsos
28-04-2006, 09:55 PM
What I had in mind was the period where Wohl was well over 2500 ACF. Perhaps this was just a result of him having gone inactive with a high ACF rating and then had a few good results on return. For instance, he was 2538 ACF before losing 80 points in the last list, and, GM-strength as many of his recent overseas results have been, I'm not sure he's quite that good.

Did Wohl get the 70 points or was he just resuming on 2485 and then going to 2509 in March 05 on the basis of a good result?He got the 70 point uplift which was applied to all players who had been active between March 2000 and December 2003.

Vlad
28-04-2006, 09:58 PM
0.75 sounds to me like almost no scaling. Can not see why would you bother at all.

Bill Gletsos
28-04-2006, 11:07 PM
0.75 sounds to me like almost no scaling. Can not see why would you bother at all.Testing at the time showed that dropping below 0.75 resulted in a loss of predictive accuracy that we deemed unacceptable.

Garvinator
28-04-2006, 11:53 PM
Put as simply as possible it is this.

The opportunity/possibility for a player who loses rating points to regain those points is less for players at the top end as opposed to players at other positions in the rating list.

We noted that if a player rated say 1700 lost to a player rated 1400 (a 15% chance) he could get all his points back in one game by beating a player rated 2000 (also a 15% chance). However if you are rated 2400 and lose to a 2100 then to get the points back you would need to beat a 2700. There are no 2700 rated players in Australia. Testing showed that there was a slight disadvantage to the top players compared to the vast majority of other players. To try and alleviate this we implemented a sliding scale where the points lost or gained were scaled depending on the players rating at the start of the rating period. If your rating was equal to or less than 2200 then no scaling was applied. If your rating was 2300 and greater then a scale of 0.75 was applied. For ratings between 2200 and 2300 this was a straight line from 1.0 to 0.75.

:hmm: actually this was part of what I was trying to say:
However if you are rated 2400 and lose to a 2100 then to get the points back you would need to beat a 2700. There are no 2700 rated players in Australia. so I was on the right track;)

PHAT
28-04-2006, 11:54 PM
If Lane hadnt haved played for 1-2 years and had his games rated the same way as the average club player, his rating would have dropped by a fair margin. The question you have to ask yourself is,if the Glicko fluctuations are acceptable for a 2000/1800/1600 player then why are they not appropiate for players above 2200?

Beautiful! A perfect illistration of how a system (Glicko) can be wrong footed. :clap:

PHAT
28-04-2006, 11:58 PM
Given you dont know what you are talking about you shouldnt just mindlessly speculate.
You have not learned a thing. You still abuse those who do not hold your view and you do not have the brains to teach. unless you can produce a defence of Glicko's downward volitily STFU and if possible, for ever.

PHAT
29-04-2006, 12:04 AM
Bill can clarify (if he wants)

OMG. What do you mean "if he wants."

FFS this equivelent to a DOP being required to to explain has a pairing was arrived at. The ratings officer MUST be able to explain how a drop is justified. If he cannot do that, he should go away, forever.

Bill Gletsos
29-04-2006, 12:08 AM
You have not learned a thing. You still abuse those who do not hold your view and you do not have the brains to teach.This particlar aspect was explained in depth at the time on the old ACF board board. As Garvin was in the top 15 posters on that board then I expected he should have remembered it. If he didnt, then he should not have speculated and added confusion.

unless you can produce a defence of Glicko's downward volitily STFU and if possible, for ever.There is no need to defend Glicko's downward volatility as there is none.

Garvinator
29-04-2006, 12:09 AM
You have not learned a thing. You still abuse those who do not hold your view and you do not have the brains to teach. unless you can produce a defence of Glicko's downward volitily STFU and if possible, for ever.
Matt, can you please shut up. This could actually be a thread that posters can learn something from. Whether or not you agree with glicko2, at least it is helpful to understand what is going on and at least what the rating officers were/are thinking when they make changes.

Secondly, I didnt take Bill's response to me as abuse. As I have replied, I do believe that I was kinda on the right track, just didnt express it well, which has been clarified by Bill.

Bill Gletsos
29-04-2006, 12:09 AM
Beautiful! A perfect illistration of how a system (Glicko) can be wrong footed. :clap:The only one wrong footed around here is you.
As usual you are just clueless.

PHAT
29-04-2006, 12:10 AM
Testing at the time showed that dropping below 0.75 resulted in a loss of predictive accuracy that we deemed unacceptable.

Show us. Surely that is not to big an ask.

The reason you MUST show us is that we, the great unwashed players, want to know the evidence that lead the great bot to deem it unacceptable.

PHAT
29-04-2006, 12:16 AM
As usual you are just clueless.

As usual you are a technician with no insight into the fact that modelling of real phenominae is more difficult than the simple mathematics.

Bill Gletsos
29-04-2006, 12:16 AM
OMG. What do you mean "if he wants."As usual I see you are selectively quoting and taking peoples statements out of context.

Kevin essentially asked did the scaling of rating changes for those above 2200 apply to all players above 2200 or just those with reliable ratings. Given all explantions have never mentioned any restriction based on a players reliability then it could be conclued that the answer is that it applies to all players rated above 2200.

FFS this equivelent to a DOP being required to to explain has a pairing was arrived at. The ratings officer MUST be able to explain how a drop is justified. If he cannot do that, he should go away, forever.I have explained the scaling for players rated above 2200 a number of times in the past.

Bill Gletsos
29-04-2006, 12:21 AM
As usual you are a technician with no insight into the fact that modelling of real phenominae is more difficult than the simple mathematics.You are just a joke.
This was evidenced by your abysmal attempts at using "trending" to predict future ratings movements.
As for your own "Trirank" rating system you started rabbitting about over 3 years ago that has never seen the light of day.

As usual you are a complete waste of time.

Kevin Bonham
29-04-2006, 12:26 AM
OMG. What do you mean "if he wants."

I mean that contrary to myth, the duties of his position do not include being required to answer every question here. Obviously what he's doing needs to be justifiable, but that doesn't require a duty of constant public comment - that is an optional extra.

bergil
29-04-2006, 12:26 AM
You are just a joke.
This was evidenced by your abysmal attempts at using "trending" to predict future ratings movements.
As for your own "Trirank" rating system you started rabbitting about over 3 years ago that has never seen the light of day.

As usual you are a complete waste of time.
Trirank? What was that and how did it work?

Bill Gletsos
29-04-2006, 12:31 AM
Trirank? What was that and how did it work?Kerry Stead summed it up best on the old ACF board.

Trirank, rhymes with wank

A coincidence? I think not!

PHAT
29-04-2006, 12:49 AM
I have explained the scaling for players rated above 2200 a number of times in the past.

Good. Then it will be no problem for a man of your hard drive to show him where.

Bill Gletsos
29-04-2006, 12:57 AM
Good. Then it will be no problem for a man of your hard drive to show him where.As I have explained previously I am not anyones research assistant.

PHAT
29-04-2006, 12:57 AM
You are just a joke.
This was evidenced by your abysmal attempts at using "trending" to predict future ratings movements.


There was some VERY interseting work done by me about 1 1/2 years ago on this and Glicko is in trouble. So is the ACF revenue stream.


As usual you are a complete waste of time.

:lol: You wish. V..

bergil
29-04-2006, 01:11 AM
As usual you are a complete waste of time.
Bill that last statement is not correct. My typing skills have increased as has my patience for self flagellating mentally infirmed ignoramuses! :P

Bill Gletsos
29-04-2006, 01:21 AM
Bill that last statement is not correct. My typing skills have increased as has my patience for self flagellating mentally infirmed ignoramuses! :P:P

Kevin Bonham
29-04-2006, 02:30 AM
There was some VERY interseting work done by me about 1 1/2 years ago on this and Glicko is in trouble.

Rubbish. Your sole attempt at predictions using trending was a crushing failure (http://www.chesskit.com/auschess/cgi-bin/yabb/YaBB.pl?board=auschess;action=display;num=10706229 81) and all your other "work" has been untested theoretical blather apart from possibly some fatally methodically flawed player surveys.

Rincewind
29-04-2006, 09:37 AM
As usual you are a technician with no insight into the fact that modelling of real phenominae is more difficult than the simple mathematics.

I would argue that modelling is, by definition, a mathematical exercise. Maths is the language of science.

PHAT
29-04-2006, 04:12 PM
As I have explained previously I am not anyones research assistant.

IDIOT. You claim that you research Glicko behaviour using different constants. Are you do that for you or us. If it is for yourself, we don't want you. If it is for us, publish.

PHAT
29-04-2006, 04:13 PM
...ignoramuses!

ignorami

PHAT
29-04-2006, 04:16 PM
Rubbish. Your sole attempt ...

You have only seen one of two. Unlike the first attewmpt that dudded, #2 is a goer. The ACF will not get it as long as Gletsos has anything to do with the ratings.

PHAT
29-04-2006, 04:21 PM
I would argue that modelling is, by definition, a mathematical exercise. Maths is the language of science.

:hmm: I would say that maths is the most imortant TOOL of modelling. Modelling takes a broard understanding of the system being modelled. Gletsos can use a calculator to get the answers, but he does not know what the questions are.

Bill Gletsos
29-04-2006, 04:28 PM
:hmm: I would say that maths is the most imortant TOOL of modelling. Modelling takes a broard understanding of the system being modelled. Gletsos can use a calculator to get the answers, but he does not know what the questions are.If you think Graham Saint and I use a calculator you are a fool. As for understanding rating systems you have continually demonstrated you have absolutely no clue.

PHAT
29-04-2006, 04:47 PM
If you think Graham Saint and I use a calculator you are a fool.

You are truly autistic.

Bill Gletsos
29-04-2006, 04:48 PM
You are truly autistic.Coming from someone who is truly clueless.

Rincewind
29-04-2006, 04:48 PM
:hmm: I would say that maths is the most imortant TOOL of modelling. Modelling takes a broard understanding of the system being modelled. Gletsos can use a calculator to get the answers, but he does not know what the questions are.

In the same way that language is the most important tool in literature. Yes, you need to know how to tell a story. Perfectly formed but uninspired prose has less value than a mistake ridden but uplifting piece. However, the delineation between tool and product (in both cases) is difficult to make.

Kevin Bonham
05-05-2006, 01:52 PM
You have only seen one of two. Unlike the first attewmpt that dudded, #2 is a goer. The ACF will not get it as long as Gletsos has anything to do with the ratings.

You don't have to disclose the secret herbs and spices. Produce some predictions using it that we can compare with Glicko like you did with your first attempt.