I think no one should have a rating higher than their "best win"

My rating is currently 1757 and my best win is 1710. I cannot credibly claim to be better than a player rated 1711.

I think the mods should cap peoples rating to their "best win".

Well, whilst your point is valid, there is a slight flaw in it. I just won a game against someone who is now a 1904 (but was a 2000 before he lost the game), I'm only a 1492. Why did I win; because he made an unfortunate blunder losing a piece in the endgame from which he could not come back. I would not be happy to now take on a rating of 1904 because it is way above my actual level of play..

I disagree. For example if you only play people who have the same rating as you and you win, then your rating becomes higher than your best win. Also after you defeat an opponent their rating drops, so if you play people with the same rating as you, and you win mor ethan you lose, your rating will be inevitably higher than your best win.

I did'nt say that you automatically get the rating of your "best win", just that you shouldn't have a rating higher than your best win.

That's not the way ratings work.  It's "ok" to have a rating higher than your best win, as your rating is based also on your winning percentage, not soley on who you beat.

The only thing you can do if you don't like it is play tougher players  .

Various contributing factors and possibilities. Already outlined by the staff, or discussed in previous forums as far as I know.

Try typing 'ratings' (and maybe follow up with 'glicko') in the top right box.

Should give you all the info you need.

That is ridiculous. How would anyone ever gain in ratings? IE this site just started so everyone joining is a 1200..... by your logic no one can be higher than their best win, so everyone would be stuck at 1200.....

Yea you didn't really think this through. The rating system is fine as it is.

I say your rating shouldn't be 100 points higher than your best win.

Sad to  repeat this ..."best win" could be a "time out" !! Frequently , like "biggest upset" , it is !!

By the way I meant to remark also that what I consider my best result  was a draw with a 2000 rated player....

This seems pretty simple to understand.

He is saying your rating will be CAPPED by your best win.

Example...ChessNetwork mainly plays people under 2000 in quick.

He is rated over 3000.  He could still be rated very high, but only as high as the highest rated player he's beaten. 

So if he'd beaten say a 2300 rated player, he would not be able to progress any higher rating wise unless he played and beat a 2301+ rated player.

Makes perfect sense to me.  I just think a lot of you didn't get it.

@ImmortalGamer and OP

There are 3 different possibilities for player matchups.

1. Player 1 is rated higher than player 2.

2. Player 2 is rated higher than player 1.

3. Player 1 and player 2 have the same rating.

In Situation 1 and 2, one of the players is essentially playing a pointless game. They can lose rating for losing, but for winning they gain nothing.

Situation 3, both players lose rating for losing but win nothing for winning.

Your proposed change would bring an end to rated chess and all because you simply don't understand something.

I think you should be careful where you point that brain of yours.

If all you have a problem with is ChessNetworks very high rating then all you really need to do is tweak chess.com's rating system to be more realistic. You shouldn't gain 1 rating point for beating someone over 1000 points below yours. But that's how it works now. If thats what you have a problem with, then I think we can agree on that. But it seems like your trying to apply that logic across the board.

So if kasparov beats 100 people in a row all rated at 2000 elo you think he is only 2000 himself? Surely he would be drawing and losing a bunch of them if he was only 2000 so some weight must be added to how often he wins.

Also the system spectaculaly fails - lets imagine that kasparov joins chess.com and is given a starting rating of 2800, how will he ever improve this? Lets assume he beats the next best player rate 2750elo, Kasparovs rating must go down for his win! Now no-one can ever be higher then 2750, so Kasparov beats the third best player at 2700elo....down he goes again!

If your rating couldn't progress beyond your best win then no-one could move beyond the rating they started (assuming we all start the same) and if we didn't all start the same then the high raters gain no points for beating those below them and have it all to lose!

what you're neglecting to consider is that when the sample size increases...your rating adjusts and comes closer and closer to an accurate depiction of your skill level...

if you've played 100 matches and won them all against 1200....your best win is against a 1200 rated player...however you're obviously league above that...see my point?

My friend,

              Here in chess.com you should always look for not only rating but best win, time per move and the average opponent. These all are very important factors when evaluating a player. I think that is why they are providing these.

The ELO takes into account if all you play is low players. you dont get many rating points at all. If you lose you lose a ton.

In every sport even if you beat a total cupcake you go up, a win is a win and the ELO takes into account how good of a win it is.

IF what you guys are saying had been added when ELO started then nobody would ever change from the 1200 rating. What point of a rating system would there be if everyone was 1200?

Here is an example:

A player plays 100 games against players rated exactly 1600 and wins them all.  His best win is therefore against a 1600. What should this player be rated?

According to the rating system, a 1600 rated player would have a 50-50 record in these hundred games, so clearly this player should be rated higher than 1600. Even a 1700 rated player doesn't win 100% of the time against 1600 rated players, so our example player should be rated higher than 1700.

It's simply too simplistic to say A player cannot be rated higher than their best win.

immortalgamer as has been pointed out by peter etc it doesn't make sense.  What does make sense is the recent change that was made to stop your rating rising if you beat someone much lower rated than you.  It used to be a rating point per win even if you were 2000 rating points higher.  perhaps this doesn't apply to live chess yet, otherwise chessnetwork couldn't keep increasing his rating.

User A joins, plays 1000 games against 1200 players, wins 950 of them.
His rating is through the roof.

User B joins, plays 1000 games against 1400 players, wins 50 of them.
His rating is now absolute crap.

User A and B are the same person.

This is why I agree with the OP.

Mark Glickman's website explains how the rating system works:

http://math.bu.edu/people/mg/glicko/

I know that the OP feels that it's unfair, but the system works correctly.  Your rating is high because of your winning percentage, not because of who you beat.