PDA

View Full Version : Ratings Questions Thread (was Question for Barry Cox)



bobby1972
05-04-2006, 10:34 AM
Barry,i drew with some one 1313 (and i was very lucky i posted the position under the dandy summer swiss )and lost 13.5 poins ,now if i had won i would have gained .4 points.does this mean that i have to beat him 33.75 times to recuperate my lost points.i personaly like this rating system ,but this seems odd,does this mean that no matter how many high rated players you beat if you draw or loose against lower rated players you may as well not play.

Vlad
05-04-2006, 10:43 AM
Yes, it does. The easiest way to get points is to stop playing. In a few years Bill will upgrade everyone's rating including players who did not play.

Then the relevant question will be "why do you care about rating if you do not play?":))

P.S. This is why I think upgrades should be proportional to the number of games played by each player.

pax
05-04-2006, 01:50 PM
Barry,i drew with some one 1313 (and i was very lucky i posted the position under the dandy summer swiss )and lost 13.5 poins ,now if i had won i would have gained .4 points.does this mean that i have to beat him 33.75 times to recuperate my lost points.i personaly like this rating system ,but this seems odd,does this mean that no matter how many high rated players you beat if you draw or loose against lower rated players you may as well not play.

It's a misconception. If you lose to a player rated X points below you, and beat a player rated X points above you, your net rating change should be zero.

pax
05-04-2006, 02:02 PM
Yes, it does. The easiest way to get points is to stop playing. In a few years Bill will upgrade everyone's rating including players who did not play.


Actually this isn't correct. Bill retrospectively removed 'upgrade' points from players who were inactive over a long period.

Rincewind
05-04-2006, 02:05 PM
It's a misconception. If you lose to a player rated X points below you, and beat a player rated X points above you, your net rating change should be zero.

...all other things being equal. It would depend a great deal on the RD values as well as the ratings of those involved.

The other loss of symmetry is that while 2000 players occasionally draw or lose to players 700 points below them they don't often get the chance to play players rated 700 point higher.

Vlad
05-04-2006, 02:18 PM
...all other things being equal. It would depend a great deal on the RD values as well as the ratings of those involved.

The other loss of symmetry is that while 2000 players occasionally draw or lose to players 700 points below them they don't often get the chance to play players rated 700 point higher.

Another loss of symmetry arises when you play in a swiss and say you are close to the top. Whenever you play with somebody who is 400+ below you (after the first rounds), you are playing with a player who is performing much better than his rating, which means he is underrated. The same is true for players who close to the bottom. They only play 2000 player when he/she has a bad tournamnet or when he is overrated. This is a well known phenomenon in the professional chess. Players playing in swiss tournaments are pretty much bounded by 2650 because of that. If they had a chance to play in Linares they would be 2700.

Vlad
05-04-2006, 02:29 PM
Actually this isn't correct. Bill retrospectively removed 'upgrade' points from players who were inactive over a long period.

This amendment makes the system funny. If the rule is that the points are taken away only when you haven't played for a long period of time, you strategy is to play one game a year with a player 1000 below your rating and collect the points.

This rule reminds me the australian tax system, which is probably the most complicated tax system in the world. Instead of completely rewriting the rules the government makes amendments here and there (sometimes retrospectively) that make the rules more nad more complicated.

pax
05-04-2006, 02:33 PM
Players playing in swiss tournaments are pretty much bounded by 2650 because of that. If they had a chance to play in Linares they would be 2700.

Most of the top successful 'Open Swiss' players would be mercilessly crushed if they played Linares.

bobby1972
05-04-2006, 02:42 PM
thanks very much for all the info everyone

Vlad
05-04-2006, 02:44 PM
Most of the top successful 'Open Swiss' players would be mercilessly crushed if they played Linares.

Sorry mate, but this is not what observed. Just take example of Levon Aronyan or Ruslan Ponomarev or Khalifman, etc. All these players got points from their first Linares.

Bill Gletsos
05-04-2006, 04:23 PM
Barry,i drew with some one 1313 (and i was very lucky i posted the position under the dandy summer swiss )and lost 13.5 poins ,now if i had won i would have gained .4 points.does this mean that i have to beat him 33.75 times to recuperate my lost points.i personaly like this rating system ,but this seems odd,does this mean that no matter how many high rated players you beat if you draw or loose against lower rated players you may as well not play.Actually you would need to beat the 1313 rated player 32 times.
However under Elo its worse. Based on a K factor of 15 and using the normal distribution the loss costs you 7.435 but each win only gains you 0.646. Therefore you would need to win 115 times. :wall:
If the logistic distribution is used each loss costs you 7.295 but each win gains you 0.205 and you would therefore need to win 36 times.

So Glicko is better for you. ;)

Tomas Kessler
05-04-2006, 04:32 PM
However the rating of a player could be markedly different from elo given the same results. Who was the bight spark who foisted this system on the chess playing community in Australia? Just curious, I have my suspicions but I just want to see if I am right.

Bill Gletsos
05-04-2006, 04:44 PM
However the rating of a player could be markedly different from elo given the same results.Yes but its the difference in ratings that is important not the absolute value. The difference between Bobby1972 and his 1300 odd opponent is likely to be pretty much the same.
Also in the majority of cases under Elo juniors would be much lower rated compared to adults than they are under Glicko.

Altecman
22-04-2006, 05:37 PM
[QUOTE=bobby1972]does this mean that i have to beat him 33.75 times to recuperate my lost points.iQUOTE]

You wont beat he 33.75 more time bobby. Sorry about disapointing you

bobby1972
24-04-2006, 10:53 AM
Its Ok I Just Beat A Few 2200 + And The Damage Is Fixed,

Altecman
28-04-2006, 09:27 PM
Its Ok I Just Beat A Few 2200 + And The Damage Is Fixed,

lol, yeah ps. I think Voon is my new bunny, beat him at Easter cup as well lol 4:3

MichaelBaron
02-05-2008, 07:37 PM
I am very curious is there anyone without a specialised degree/skills who can understand how the Glicko system works. I had no trouble learning how to do elo rating calculations in half an hour (that was before fide published its rating calculator online) but i still have no idea how Glicko works:hmm:

schess
14-03-2011, 05:47 PM
I played on the weekend at a chess club and my rating is 961 but I losed to an unrated player does this change my rating?

Rincewind
14-03-2011, 06:14 PM
I played on the weekend at a chess club and my rating is 961 but I losed to an unrated player does this change my rating?

Most likely but it would depend if that player gets a rating in the next rating period (they need a total of 9 games up to the end of May). If they do not obtain a rating then I think it will not affect your rating. But if they do get a rating then that game will contribute towards a rating adjustment for you.

That is my understanding anyway, Bill Gletsos can confirm or deny for sure.

schess
19-03-2011, 04:07 PM
Hi,
I have a question about your calculator? My rating is 1209 in the march acf rating list and i beat somebody with a 1268 rating in the march acf rating list so does acf calculale that i beat him as a rating of 1268 or the rating he gets in may when the new acf list comes out

Kevin Bonham
19-03-2011, 04:29 PM
My rating is 1209 in the march acf rating list and i beat somebody with a 1268 rating in the march acf rating list so does acf calculale that i beat him as a rating of 1268 or the rating he gets in may when the new acf list comes out

The ACF uses "intermediate ratings" so it will calculate as if you beat him at a rating somewhere between (about half way between most likely) 1268 and his new rating.

Next ACF list comes out at start of June.

schess
30-05-2011, 09:35 PM
When does the next acf rating list come out?

Adamski
30-05-2011, 10:12 PM
Next ACF list comes out at start of June. I.e., very soon!

schess
15-06-2011, 12:39 PM
Hi Barry,
How many Fide rated games do unrated players need to play to get a rating?
Currently I played 5 and still I don't have a rating.

Thanks

Rincewind
15-06-2011, 12:48 PM
Hi Barry,
How many Fide rated games do unrated players need to play to get a rating?
Currently I played 5 and still I don't have a rating.

Thanks

If you mean ACF rated games for an ACF rating then 9. My site only have ACF ratings and I don't store, calculate or make available FIDE ratings.

If you really do mean a FIDE rating then the regulation is available here

http://www.fide.com/fide/handbook.html?id=72&view=article

antichrist
09-11-2011, 12:05 AM
Firegoat in the other place reakons that Baz reakons that George Xie now has his extra rating points according to Glicko, will that get him over the line for GM rating or it means nothing?

Kevin Bonham
09-11-2011, 12:41 AM
Firegoat in the other place reakons that Baz reakons that George Xie now has his extra rating points according to Glicko, will that get him over the line for GM rating or it means nothing?

Whatever firegoat reckons about ratings can generally be safely ignored as he is almost a complete ignoramus on the subject and has a lot of chips on his shoulder about it. But he has one thing right: GM titles are determined by FIDE ratings not ACF. FIDE does not use Glicko ... yet ... :lol:

antichrist
09-11-2011, 01:08 AM
I was a knight up against George once, was scared of getting into time trouble and blew my lead, was only a rapid