Are Chess.com ratings accurate?

Personally I have found better rated players that I am stronger then and worse rated players that are stronger. Any thoughts if this is a coincidence or are they just off?

I feel that here on chess.com, in the long run, ratings have an relatively normal distribution.

It is very likely that the wins and losses have been due to the difference in the playing styles of you and your opponents.

Heck, right now I have a very low rating, because my chess is a little rusty, but in the past, I have even been able to beat Little Chess Partner.

The ratings are "accurate" in the sense they are calculated correctly.  Some things that will influence ratings here that may cause them to be less correlated to real playing strength (for example I feel my rating is high and it is finding its way to a more reasonable level even though my play is improving):

1.  Having a string of opponents time out to you

2.  Timing out in a string of your own games

2.  Colors are not always random

3.  Thematic tournaments may cater to a player's strengths

4.  In CC it is a bit of an open book exam - we will perform better with access to opening books and databases during the opening stages and will be less likely to blunder with time and an analysis board.  Some of us use these tools a lot.  Others are playing from their iphone on the train with no opening explorer etc.

5.  Other factors

So they are as accurate as they can be given the factors at play and the degree of player choice involved.  You can't compare CC and Live ratings for these reasons, and CC ratings may be less relevant comparing one player to another as well as a result.  Over time though I think they are a good indicator of your own personal improvement.

The above relates to turn based ratings.  The live ratings should be as accurate as any other live ratings, relative to the pool of players here.

I agree.  Much of the time when I make my moves, I do it on campus using my kindle.  I don't really rely on the opening books, because doing so would take forever.

Timing out is also a big factor.  My rating would be significantly different without them.

 I feel ya. :/ I'm in college too and my laptop lags out and such too lol.

Would you believe it - my chess.com rating is exactly correct at 1712

 lol

Be proud of it.  I wish I had a 1712.

Yes, of course I am happy. Be proud to be part of the chess community - it is a wonderful experience at all levels of play.

Don't feel so bad, bro. In live chess, 600 of those points come from clocking your opponent.

I think I'm better than I actually am most of the time.

To the OP: If you want to dispute the accuracy of chess.com's rating system, you'll have to do some work on finding mathematical flaws in glicko (http://www.glicko.net/glicko/glicko2.pdf). Good luck. ^_^

heinzie wrote:

The Glicko system is very good indeed but there MAY be a bias (difference of means) between the live chess.com rating and USCF ratings. This bias would affect all players the same way. For example, it is known that on Free internet chess Server (FICS) which also uses the Glicko system, the standard rating is overevaluated up to 200 points for most players. But as the above example of 1712 seems to indicate, there would not be no such a bias here. It is very strange to notice that the bias is different for players of Master level, which should not be the case since the fundamental equations are the same for USCF, FIDE, FICS and chess.com ratings (the Glicko system is an improvement of the Elo system), so a difference of strengh of 500 points between two players in one ranking list should stay 500 in another. That could mean that masters play under their real strengh on chess servers compared to OTB tournament games, maybe due to the lack of stake. I noticed the same trend here on chess.com, with several USCF players rated >2000 having a standard live rating under 2000, while there seems to be no bias for weaker players.

No rating system is truly accurate, play is effected by too many variables to be accurately determined by mathematical formulas. I find that many amateur players will/can play within an interval of 200 points, plus or minus, of their rating. Professionals are another, different story.

Lower rated opponents are sometimes those chess players who are underrated by constantly playing opponents stronger than them. 

USCF or FIDE raitings is the most accurate as you can get in OTB play since OTB play we take a lot more seriously. Chess.com raiting is nothing. I dont count other chess servers raitings just the national federations or FIDE.

For my part I play OTB (FIDE and national) and on at least two chess servers at several time controls + a tactical problem server. That provides about 8+ ratings and I can make a mean of all ratings and have a more stable "overall" rating for my own evaluation.