Change to the rating system

Change to the rating system

Help

Cookies help us deliver our Services. By using our Services or clicking I agree, you agree to our use of cookies. Learn More.

L

Joined
23 Aug 03
Moves
929
28 Sep 03

There is something I don't understand. I just bet a guy on 'Move timeout' and got my rating reduced (??!) by 3 points, while his grew for a point 🙄 He had a p 1224, and I a p 1341.
The point is that I was so much ahead of him, but he was slow in moving, and I wanted to play some more, and not the same game I would winn in a few moves anyways , so I decided to claim a victory. It seems that this really doesn't pay off, but what's the point then?🙄

ea
Santa.

The Mall.

Joined
11 Jul 02
Moves
66753
28 Sep 03

Originally posted by Longbowman
There is something I don't understand. I just bet a guy on 'Move timeout' and got my rating reduced (??!) by 3 points, while his grew for a point 🙄 He had a p 1224, and I a p 1341.
The point is that I was so much ahead of him, but he was slow in moving, and I wanted to play some more, and not the same game I would winn in a few moves anyways , so I ...[text shortened]... ed to claim a victory. It seems that this really doesn't pay off, but what's the point then?🙄
i think its because you are both on provisional ratings that wont happen after your 20 games.😏

L

Joined
23 Aug 03
Moves
929
29 Sep 03

Ah so, thanx. I think this should be noted somewhere so people can see, I wouldn't have lost some points had I known that...
I've just realized something else that is odd. I had a Timeout victory once before, and at that time both myself and the opponent had a provisional rating of 1200 and it was our first game. He stopped playing, I claimed a victory and received 200 points, which I saw after playing the first 5 games. 🙄

So does one's rating get recalculated after 20 games, or it just stays and is not provisional any more?

f
Quack Quack Quack !

Chesstralia

Joined
18 Aug 03
Moves
54533
02 Oct 03

How about: A player who has played more than twenty games uses that number 400.
Then for players with less games:

Mathematically: X games already played means you affect your opponents rating by 20X.

for instance: A player with say 8 games under their belt would only have an effect of 160.

This way a 1200 player would not affect your rating at all.

f
Quack Quack Quack !

Chesstralia

Joined
18 Aug 03
Moves
54533
02 Oct 03

SORRY ABOUT THE ABOVE

I have carefully reread the rating calculations I feel so silly,

Why not PA the Provisional adjusment

PA = 0 if they have finished no games
= 0.05 if finished 1 game
etc that is 0.05 times the number of games your opponent has completed
= 1 if more than 20 games

stitching you up

Joined
08 Apr 02
Moves
7146
02 Oct 03

Originally posted by Longbowman
Ah so, thanx. I think this should be noted somewhere so people can see, I wouldn't have lost some points had I known that...
I've just realized something else that is odd. I had a Timeout victory once before, and at that time both myself and the opponent had a provisional rating of 1200 and it was our first game. He stopped playing, I claimed a victor ...[text shortened]... one's rating get recalculated after 20 games, or it just stays and is not provisional any more?
you would have lost points against him win, lose or draw. Even if you'd won by check mating him instead of claiming check mate it would not have made a difference.

The idea of the provisional rating system is that after the 20 games you have a roughly accurate rating. The reason you lost points was because he was lower rated than you. I don't understand the system exactly, but I guess it works on a rolling average of your wins - you gain a massive number of points for your first victory, but if you then beat someone lower than you the system can't continue to increase your rating because you haven't shown yourself to be better than someone of your rating, only better than someone of a lower rating, so you lose a few points. Does that make any sense (I'm not sure it does)

f
Quack Quack Quack !

Chesstralia

Joined
18 Aug 03
Moves
54533
03 Oct 03

Perhaps the provisional adjustment PA could also be less than 1 if the player has showed dramatic fluctuation in their rating over the last 20 games. This would then reduce the affect of grandmasters climbing their way to dizzying heights from 1200 or the affect of those players who embark on an unacheivable numbers of games etc. The simplest way to measure dramatic fluctuation could simply be to subtract the lowest rating for the last 20 games from their highest. I suspect most players stabilize within a 100 point range if they take every game on equal merit and their opponents are not unfairly rated.

I make many comments on the rating system here, but don't get me wrong, I think it is very good. I had a look on some other sites knot to mention names and their ratings are @#%!ed.

L

Joined
23 Aug 03
Moves
929
04 Oct 03

thanx people, I think I understand the rating system better now. It does make sense belgian freak

t
Xebite

in front of you

Joined
06 Jan 03
Moves
15730
15 Feb 04

I just fould this:
http://chess.go.ro/rules/fide/elo.html#Mathematical_Details
it says that the ELO rating is exactly (or almost) the rating we have here at RHP 🙂
th

s
Red Republican

Auckland

Joined
08 Jun 03
Moves
6680
16 Feb 04

When is your opponents rating determined? At the start of the game or the end?

An example - I start a game with a player who is 50 points better than me. I have slipped a little behind, but have not given up.

I note my opponent has not moved for a while - in fact, I could time him out but I usually don't. Many others do win by timeout and when the game resumes, he is 300 points down. I then lose - am I disadvantaged by being lenient and losing to a poorly ranking opponent?

Of course, with an end based ranking, I get an advantage if my opponent improves. But timeouts can dramatically change a player rating - and if it works on an end based rating, first to time-out gains more points.



t

Joined
19 Aug 02
Moves
101523
16 Feb 04

Originally posted by steerpike
When is your opponents rating determined? At the start of the game or the end?

An example - I start a game with a player who is 50 points better than me. I have slipped a little behind, but have not given up.

I note my opponent has not moved for a while - in fact, I could time him out but I usually don't. Many others do win by timeout and when the ...[text shortened]... er rating - and if it works on an end based rating, first to time-out gains more points.



All rating calculations for wins and losses are calculated at the END of the game.

-trekkie

s
Red Republican

Auckland

Joined
08 Jun 03
Moves
6680
16 Feb 04

Originally posted by trekkie
All rating calculations for wins and losses are calculated at the END of the game.

-trekkie
Thought so. Have had a few games where a similar situation has happened. So - if you time out someone out as soon as you can, you get maximum points because if you let someone else in first, the ratings will have dropped by the time you claim a win.

I think a few have worked this out for themselves.

Joined
07 Oct 01
Moves
51059
30 Oct 04

I hope this is not ridiculously out of date, but I noticed you have
S=1 for a win, .5 for a draw, and -1 for a loss. However elsewhere in the forums in the area of general knowledge s=0 for a loss.
I suspect the -1 is an error because it makes the set of values asymmetric.

s=1, s=.5, s=0 is a symmetric set.
s=1, s=.5, s=-1 is not a symmetric set.

What was your real intention?

Raymond Zachary
razacharypc@comcast.net

R

London

Joined
05 Mar 03
Moves
6047
30 Oct 04

Originally posted by Ray Zachary
I hope this is not ridiculously out of date, but I noticed you have
S=1 for a win, .5 for a draw, and -1 for a loss. However elsewhere in the forums in the area of general knowledge s=0 for a loss.
I suspect the -1 is an error because it makes the set of values asymmetric.

s=1, s=.5, s=0 is a symmetric set.
s=1, s=.5, s=-1 is not a symmetric set.

What was your real intention?

Raymond Zachary
razacharypc@comcast.net
Indeed, s=0, not -1 for a loss; this is stated in the FAQ
http://www.redhotpawn.com/help/index.php?help=faq

Chrismo's main purpose in the original post was to descripe the provisional rating system that he was about to introduce. His description of the 'normal' rating system conntained a typo which you seem to have been the first person to notice in the year and a bit since he posted it.

Time to close this thread?

Joined
07 Oct 01
Moves
51059
20 Feb 05

Dear Sir:
I have done an Excel Spreadsheet on the basic scoring algorithm (disregarding new players) to gain insight on the effects of playing opponents at different levels. It is noteworthy that whenever one plays another opponent at the same level, one will lose three times as many rating points by losing as he will gain by winning. This lack of symmetry is inherent in the algorithm which is shown in this article. It seems to me that several years ago when I first studied your algorithm it was symmetric. Is that an error or is it intentional. If it is intentional, I am curious about the reasoning.
Thanks for your comments.
Raymond Zachary
razacharypc@comcast.net