New Ratings System

 Put simply, the higher it is the more your rating will change per game. Over time it will slowly increase, but as a game finishes it wil decrease. :)

RD is pretty much Standard Deviation. Because our ratings are just statistical, the actual number will not represent your true rating.

Your RD is 59, therefore:

There is a 67% chance your rating is between 1,124 and 1242.

There is a 95% chance your rating is between 1,065 and 1,301.

Hope that helps! 

 ?..

RD doesnt mean that..

My RD is 63 and rating is 1800+

Haddads RD is 36 and his rating is 1900+,

the correct response of what RD is is the post above yours.

Yes, it does mean that, although perhaps I wasn't clear enough.

Your RD of 63, coupled with your rating of 1838, means that:

There's a 67% chance of your rating being between 1,775 and 1,901.

There's a 95% chance of your rating being between 1,712 and 1,964.

One, then two standard deviations away.

The reason I find the concept of RD suspect is that it purports to measure the unpredictable or the unknown. But it seems to me that what is unknown cannot be measured.

It is very difficult, perhaps impossible, to know the real playing strength of a person. Therefore it seems to me that the rating should just reflect their known playing performance evenly, and not purport to do more than that.

 But theres 100% chance my rating is between 1800-1850.. becuase you alredy know my rating?

Im confused :P

You have to remember, your rating isn't a precise number like a measurement on a ruler, it's a statistical estimate! So it's all based on percentages.

 Your rating is never a constant. It is a variable, and where it is now is simply put as a number this is the best guess. Your rating can never be certain. (have you noticed that when you finish a game it changes?)

If I understand Nytik correctly (and I very well might not) your rating is:

67% likely to be within R (rating) - or + your RD:
1838 - 63 = 1775;
1838 + 63 = 1901;
which means there is 67% chance that your REAL playing skill (not a rating variable) is between 1775 and 1901.

and 95% likely to be within R - or + 2* your RD:
1838 - (63*2, which is 126) = 1712;
1838 + 126 = 1964;
which means that there is a 95% chance the your REAL playing skill is between 1712 and 1964.

Hope this sorta clears things up... your rating is an estimate, not a constant, and is subject to change. It is not an exact measure of your playing skill. This is basically the range.

oh, thanks alot! you cleared it up! :)

Well, I COULD'VE said that, but it would've taken all day. 

 lmfao

i love chess.com rating system fair and square

the trouble is that it was changed when my rating was at a low point. now ive had some really good days since then but its still amost 100 points less than it was, as my rating tends to fluctuate.

I think that ratings should be viewed as a representation of your known playing performance, not of unknown 'playing strength'. As such they should be fairly easy to compute, and probabilities and subjective guesses should have no part in it.

Ratings simply don't work like that.

Look at it this way:

I have two masses, 5kg and 10kg. The 10kg mass will ALWAYS be more massive than the 5kg one. There's no questioning that.

Now we have two chess players, rated 1500 and 1800. The player rated 1800 is higher, but he will not always win! This is why probabilities are, and must be, incorporated into any accurate rating formula.

Now we have two chess players, rated 1500 and 1800. The player rated 1800 is higher, but he will not always win! This is why probabilities are, and must be, incorporated into any accurate rating formula.

*I agree, the player rated 1800 will not always win. Then if they lose their rating will go down somewhat and the 1500 players' rating will go up somewhat. That is, their ratings will reflect this after it happens, not before.

You seem to be missing my point. In the universe, we cannot have an absolute value unless we will have an absolute result. As the result is not absolute (we do not know which player will win) then their ratings cannot be either.

Well, it may be that we are both missing each other's point. My point was that ratings should be taken to reflect known peformance in the past, not unknown performance in the future. I am not sure what is meant that "we cannot have an absolute value unless we will have an absolute result." We do have the results of the previous games and they are enough to base a rating on.

In baseball  a .300 hitter has had 3 hits in every 10 at bats. A .200 hitter has had 2 hits in every 10 at bats. These are absolute numbers, but we can only say that the .300 hitter is more likely to get a hit his next time at bat than the .200 hitter.

And this is the basis of RD!