Ratings inflation

For discussion pertaining to Chess, Net-Chess, or general interests.
Post Reply
wharris
Posts: 1
Joined: Tue Mar 07, 2000 11:53 am

Ratings inflation

Post by wharris » Wed Jun 02, 2004 6:02 am

Hi all:

Sorry to raise this - it's been talked about enough before - but I can't help noticing that the inflation of ratings is getting worse. Since I've started playing here I've gone from about 1600 (which was inflated to begin with) to about 2050, without getting much better at chess.
I know that ratings are purely relative, but the inflation seems to be so serious that they can no longer be relied upon as guides even to relative playing strength. It's not really a problem - this is by far my favourite correspondence chess site - but it would be nice to have some external comparability.
Not sure how anyone else feels, or what can be done to fix it. Maybe the first thing is to ask if anyone out there knows why it's happening. Answers on a postcard please...

Billy.

ruymoultor
Posts: 23
Joined: Mon Feb 09, 2004 8:03 pm
Contact:

Post by ruymoultor » Thu Jun 03, 2004 4:32 pm

one thing that could help would be getting rid of the rating floor.

jonwat
Posts: 14
Joined: Sun Mar 07, 2004 4:27 am

re : ratings

Post by jonwat » Sun Jun 06, 2004 5:31 am

Another idea that may help would be to not give new members any rating until they have earned one.

wulebgr
Posts: 65
Joined: Wed Jan 28, 2004 1:12 pm
Location: PNW USA
Contact:

Post by wulebgr » Tue Jun 08, 2004 1:53 pm

When I create a database of all my completed games, which includes those that timed out, ChessBase calculates my performance rating 200 points lower than the SlowChess calculator. No doubt different formulas are at play, but I suspect that the average rating of those who time out is the key difference.

Current SlowChess rating: 2275, based on 79 games against average rating of 1920, 54-14-11 (w-l-d)
ChessBase Performance rating: 2078, based on 125 games against average rating of 1805, 100-14-11

Of course, if these calculations were applied across the board, the ratings would fall even more, as my opponents' ratings would be lower.

Perhaps there is a solution to the rating inflation on this site buried in this information.

katchum
Posts: 176
Joined: Thu Aug 10, 2000 3:16 am

Post by katchum » Tue Jun 08, 2004 5:26 pm

Let's ask G. Miller to invent a new algorithm that makes it very difficult to stay at 2000.

By the way: where is the live chess you promised me? :roll:
I don't mind, you don't have to change anything, it's good enough for me for now, but still...

gmiller
Site Admin
Posts: 1388
Joined: Sun Mar 14, 1999 11:13 am
Location: Jeffersonville, IN
Contact:

Post by gmiller » Tue Jun 08, 2004 11:03 pm

The Elo rating system has two parameters, the first representing the maximum a provisional player's rating can vary from their average rating (e.g. if they lost every game, or won every game). The recommended ranges for this variable are between 400 and 800, I picked 800, if ChessBase uses a lower value, then that would explain the difference. I bumped the value down to 400, so let's see how that goes, it'll at least remove slow the addition of points into the rating pool anyway.
By the way: where is the live chess you promised me?
After looking at other free blitz chess servers, theres just no one on them. So it doesn't really make sense to try to dillute the playing population with one more server.

katchum
Posts: 176
Joined: Thu Aug 10, 2000 3:16 am

Post by katchum » Wed Jun 09, 2004 8:16 am

Hmm, interesting maneuver. I don't 'feel' it yet.

mic
Posts: 134
Joined: Sun Jan 18, 2004 4:31 pm

Post by mic » Fri Jun 11, 2004 6:24 am

:roll: :idea: :!: It really doesnt matter about chess ratings,,,just enjoy the site. I have noticed 1 individual who's rating is 2600 something who has won every game in 25 moves or, less and has lost only 1 game...either he/she is too good for the site or, they are cheating? I am more concerning about that than about why my rating is 400 pts higher than my uscf rating of 1747. Course that rating is from the late 70's early 80's when ELO ratings were more accurate if you want to get technical. :evil: :twisted: :roll: :wink: :wink:

Post Reply