# Glicko RD

• 8 years ago · Quote · #1

What is GlickoRD???

My score is 73 apparently!

See the full stats page on a profile for reference

• 8 years ago · Quote · #2

RD partly differentiates Glicko fro Elo.

Glicko is a rating sysem. RD= rating deviation. As a person plays more games the RD lowers under the assumption that the more games played, the more accurate the rating. if a person doesn't play much over a period of time, the RD will increase slowly. So you can look at a person's RD to determine how accurate his rating is. But also, you can look at his RD to determine how many points you'll win/lose by playing him. If he has a high RD (somewhat inaccurate) lose to him will affect you less to compensate for the inaccuracy. If you have a high RD you'll have a lot of rating change after each game, if you have a low RD you'll have slighter changes. Your RD doesn't just lower the more you play, but rather lowers in relationship to the RD of your current opponent. So, everything in sort of interconnected and relatively confusing.

• 8 years ago · Quote · #3
• 8 years ago · Quote · #4

C

Thanks,

Although what if you have more one opponent? I assume the formulaic explanation from the link accounts for that?

Also, what is a high/or low RD? ie on a scale of 1-100

• 8 years ago · Quote · #5
All I know is that the initial RD is assumed to be 350 (at FICS). That's with 0 games played.
• 8 years ago · Quote · #6

Thanks for the heads up Batgirl :-)

• 8 years ago · Quote · #7

From the freechess.org page linked above:

"Each player can be characterized as having a true (but unknown) rating that may be thought of as the player's average ability."

The only problem with this assumption is that it assumes that each time a player plays a game, that put as much effort and thought into that game as they do for every other game.  I know for myself this is entirely untrue.  Sometimes I play for fun, sometimes I try new things, and mostly I could care less what my rating is.  Sometimes I will beat people a couple hundred points higher than myself, and sometimes the reverse.  So, for myself, the whole rd itself is probably highly innacurate, because I really do not play each game in the same way.

• 8 years ago · Quote · #8
Is this "Glicko RD" value related to standard deviation?
• 8 years ago · Quote · #9
yes
• 8 years ago · Quote · #10
we start with 350 here at chess.com as well.
• 8 years ago · Quote · #11

wow, I didn't know that

• 7 years ago · Quote · #12
batgirl wrote:

oh, I forgot:

http://www.freechess.org/Help/HelpFiles/glicko.html

Batgirl scores again!!  What a freaking awesome article!!

(Did I mention I love stats?)

• 7 years ago · Quote · #13
Another helpful article to read would be http://www.chess.com/article/view/chess-ratings---how-they-work
• 5 years ago · Quote · #14

It's simple Bayseian analysis.

Find the unknown value by using known derived values to estimate it.

• 5 years ago · Quote · #15

Adesso ho finalmente capito tutto anche io :-)

Grazie

• 5 years ago · Quote · #16
onosson wrote:

From the freechess.org page linked above:

"Each player can be characterized as having a true (but unknown) rating that may be thought of as the player's average ability."

The only problem with this assumption is that it assumes that each time a player plays a game, that put as much effort and thought into that game as they do for every other game.  I know for myself this is entirely untrue.  Sometimes I play for fun, sometimes I try new things, and mostly I could care less what my rating is.  Sometimes I will beat people a couple hundred points higher than myself, and sometimes the reverse.  So, for myself, the whole rd itself is probably highly innacurate, because I really do not play each game in the same way.

There's a problem with your reasoning as well. In a rated game, both players are assumed to be playing to the maximum of their abilities. If you're trying new things, or simply not putting in enough effort in a game, then make that unrated. That's why you have the option of playing an unrated game.

About the Glicko system, the rating displayed on your profile is actually your mean, or average, rating, and the RD is the standard deviation from this mean. Technically, the way Glicko works assumes that your playing power varies according to the logistic distribution having the mean and standard deviation just mentioned.

• 2 years ago · Quote · #17

Is low rating is good or bad and what is the lowest and highest points in this ratings.

• 21 months ago · Quote · #18

Wow!

xactxx states; "About the Glicko system, the rating displayed on your profile is actually your mean, or average, rating, and the RD is the standard deviation from this mean."

"mean" is not the same as "average". This is what happens when non-mathematicians start using mathematical terms.

"mean" has a certain meaning and "average" has a pretty average one.

Do not confuse them.

I would like to thank radonato for opening up a whole new can of worms by mentioning "Bayseian analysis" - even though she/he got the spelling wrong. Have a close look at it, though. You use it every day - especially if you have a spam filter.

I agree with onosson. I never play an unrated game. I want my whole chess life to be rated. Sure, sometimes I am Lord of the Universe and sometimes I have had a few too many beers and sometimes I blunder whilst sober. Rate them all. That is me. That is my rating. Otherwise, what is the point of rating? If you can pick and choose which of your  games are rated then my rating will shoot up sooner than look at you.

Take your rating like a Man. It is you. It is a very good indication of how you play.

• 10 months ago · Quote · #19
[COMMENT DELETED]
• 2 months ago · Quote · #20

Starchip,

Actually, mathematicians and statisticians use the terms as synonyms. They both refer to the sum of the numbers divided by the number of data points, in the most basic form.

Are you thinking of midrange, median, or mode by chance?

Back to Top