Glicko RD increase

Sort:

Hookflash

Nov 23, 2013

Just a quick question about the glicko rating system (as implemented by chess.com): At what rate does RD increase during inactivity?

Rsava

Nov 26, 2013

Here is some light reading on it:

http://www.chess.com/article/view/chess-ratings---how-they-work

baddogno

Nov 26, 2013

It's really simple formula. I don't know why the answer doesn't just jump right out at you...

$r = r_0 + \frac{q}{\frac{1}{RD^2} + \frac{1}{d^2}}\sum_{i=1}^{m}{g(RD_i)(s_i-E(s|r,r_i,RD_i))}$

Where the following is defined:

let $g(RD_i) = \frac{1}{\sqrt{1 + \frac{3 q^2 (RD_i^2)}{\pi^2} }}$

let $E(s|r,r_i,RD_i) = \frac{1}{1+10^{(\frac{g(RD_i)(r-r_i)}{-400})}}$

let $q = \frac{ln(10)}{400} = 0.00575646273$

let $d^2 = \frac{1}{q^2 \sum_{i=1}^{m}{(g(RD_i))^2 E(s|r,r_i,RD_i) (1-E(s|r,r_i,RD_i))}}$

$r_i$ represents the ratings of the individual opponents.

$s_i$ represents the outcome of the individual games. A win is 1, a draw is $\frac{1}{2}$ , and a loss is 0.

Step 3: Determine New Ratings Deviation[edit]

The function of the prior RD calculation was to increase the RD appropriately to account for the increasing uncertainty in a player's skill level during a period of non-observation by the model. Now, the RD is updated (decreased) after the series of games:

$RD'=\sqrt{\left(\frac{1}{RD^2}+\frac{1}{d^2}\right)^{-1}}$

Step 3: Determine New Ratings Deviation[edit]

See also[edit]