- Blind mode tutorial
lichess.org
Donate

online Rating Calculation website (based on Glicko2) to calculate my rating change?

#1

For example based on "https: // database. lichess.org " - Sample given.
How can I calculate the rating change based on the information given below? - Black (2000 elo) has won and gained one rating point and white lost 4 rating points.
is there any way I can calculate this based on below information ( I assume you will need rating deviation?)
[Result "0-1"]
[WhiteElo "2100"]
[BlackElo "2000"]
[WhiteRatingDiff "-4"]
[BlackRatingDiff "+1"]
[TimeControl "300+0"]
Below are based on ELO ( Not Glicko2 and just mentioning them as reference)
https://ratings.fide.com/calculator_rtd.phtml ( I found this ) and have entered the above values . K=20
https://imgur.com/a/8lNsKwu
( looking for something similar for calculating rating change for games that I play ( RD - How to get this from PGN?)

For example based on "https: // database. lichess.org " - Sample given. How can I calculate the rating change based on the information given below? - Black (2000 elo) has won and gained one rating point and white lost 4 rating points. is there any way I can calculate this based on below information ( I assume you will need rating deviation?) [Result "0-1"] [WhiteElo "2100"] [BlackElo "2000"] [WhiteRatingDiff "-4"] [BlackRatingDiff "+1"] [TimeControl "300+0"] Below are based on ELO ( Not Glicko2 and just mentioning them as reference) https://ratings.fide.com/calculator_rtd.phtml ( I found this ) and have entered the above values . K=20 https://imgur.com/a/8lNsKwu ( looking for something similar for calculating rating change for games that I play ( RD - How to get this from PGN?)
#2

Based on the above example : are the rating changes symmetrical ?
[WhiteRatingDiff "-4"]
[BlackRatingDiff "+1"]
Here white lost and therefore lost -4 points and black won 1 point.
Now, had White won - then would be have got +4 and black loose 1 point?
That is in the event of white winning - would the score update be below?
[WhiteRatingDiff "+4"]
[BlackRatingDiff "-1"]

Based on the above example : are the rating changes symmetrical ? [WhiteRatingDiff "-4"] [BlackRatingDiff "+1"] Here white lost and therefore lost -4 points and black won 1 point. Now, had White won - then would be have got +4 and black loose 1 point? That is in the event of white winning - would the score update be below? [WhiteRatingDiff "+4"] [BlackRatingDiff "-1"]
#4

As you've seen, the rating deviation isn't in the PGN. The formula for rating change in Glicko(2) is complex and the point of the RD is to create a normal distribution range for what the player's actual rating "could be". The idea being the more games you play the more confidence can be put in the given rating and therefore can make the range smaller. If you stop playing for a while, the RD is supposed to drift higher because there's less confidence.

I have no idea how Lichess is modifying the RD over time but that is part of the Glicko concept.

For my Chessvis app, I decided to go with the old ELO calculation. If it's good enough for FIDE... And I didn't want to get involved in some time based RD change. Lichess has a volume of games being played that makes that effort worthwhile.

Not sure if that answers your question. But basically, without the RD value you can't figure out the new Glicko rating for a game.

As you've seen, the rating deviation isn't in the PGN. The formula for rating change in Glicko(2) is complex and the point of the RD is to create a normal distribution range for what the player's actual rating "could be". The idea being the more games you play the more confidence can be put in the given rating and therefore can make the range smaller. If you stop playing for a while, the RD is supposed to drift higher because there's less confidence. I have no idea how Lichess is modifying the RD over time but that is part of the Glicko concept. For my Chessvis app, I decided to go with the old ELO calculation. If it's good enough for FIDE... And I didn't want to get involved in some time based RD change. Lichess has a volume of games being played that makes that effort worthwhile. Not sure if that answers your question. But basically, without the RD value you can't figure out the new Glicko rating for a game.
#6

Thanks @chessvis_com and @tpr . Really appreciate taking time and explaining in detail.
@chessvis_com
is it not possible to reverse engineer the RD based on the data found in PGN.
For example -PGN Already states - the initial rating and rating-change - input and output .
Rating change = function of ( Player rating , opponent rating , RD of player and RD of opponent).
we know the rating change and also the Player and opponent ratings , only variable is RD - cant we reverse engineer this ?

Thanks @chessvis_com and @tpr . Really appreciate taking time and explaining in detail. @chessvis_com is it not possible to reverse engineer the RD based on the data found in PGN. For example -PGN Already states - the initial rating and rating-change - input and output . Rating change = function of ( Player rating , opponent rating , RD of player and RD of opponent). we know the rating change and also the Player and opponent ratings , only variable is RD - cant we reverse engineer this ?
#7

#6
You can reverse engineer the RD
rating change = function (game result, RD, rating difference)
As you know rating change, game result and rating difference, you can calculate RD

#6 You can reverse engineer the RD rating change = function (game result, RD, rating difference) As you know rating change, game result and rating difference, you can calculate RD
#8

Also the PGN only shows rounded ratings. If you win against an opponent 700 pts below your rating, then after the game rating change is displayed as "+0" but if you compare ratings on your profile before and afterwards, you will see a tiny change, something like from 1761.50 to 1761.52.

Also the PGN only shows rounded ratings. If you win against an opponent 700 pts below your rating, then after the game rating change is displayed as "+0" but if you compare ratings on your profile before and afterwards, you will see a tiny change, something like from 1761.50 to 1761.52.
#10

Check this link
https://lichess.org/forum/team-superorganization/draft-about-elo#7

It's about the difficulty of understanding elo for amateurs,
It is almost as hard as playing the game of chess.
It is incomplete.

Check this link https://lichess.org/forum/team-superorganization/draft-about-elo#7 It's about the difficulty of understanding elo for amateurs, It is almost as hard as playing the game of chess. It is incomplete.

This topic has been archived and can no longer be replied to.