# ranking formula

1 view

### fiore

Dec 17, 2003, 7:16:48 PM12/17/03
to
i am looking for a ranking or elo formula to use in a backgammon club, i
would like to use it with excel.
i searched in the web, i admit I didnt understand almost anything.
i was told that the official ranking system is the one used on FIBS server
and that an experience player doesnt have any weight after 400 games, is it
true?
last but not least: how do u calculate a rating change in double
consultation tournaments?
thanks for helping
fiore

Dec 17, 2003, 9:22:00 PM12/17/03
to
fiore <fiore_...@galactica.it> wrote:

>i searched in the web, i admit I didnt understand almost anything.
>i was told that the official ranking system is the one used on FIBS server
>and that an experience player doesnt have any weight after 400 games, is it
>true?

Basically, the FIBS formula calculates your rating change based on the
length of the match and the difference in ratings between the two
players. Even though the math looks scary, it's very straightforward.

As a special case, players who have less than 400 points of experience
get bigger ratings swings... the general idea is to help you find your
"true" rating more quickly. But <400 is really a special case of the
fairly simple general formula.

So it's not that the player has no influence over 400 points, it's
just that the extra newbie skew goes away at 400.

>last but not least: how do u calculate a rating change in double
>consultation tournaments?

I assume you mean doubles tournaments, where two players consult over
each move? Personally, I wouldn't count these at all.

-Patti
--
pat...@gammon.com | Thinking about you as innocent
http://www.gammon.com/ | is sort of like thinking about
Check out www.tribe.net! | a viking playing badminton.

### fiore

Dec 18, 2003, 5:23:06 AM12/18/03
to
i guess u missed a line of my post :)
what i am basically looking for is the rating formula.

"Even though the math looks scary, it's very straightforward"
How is this mysterious simple one?
thanks
fiore

### Elizabeth Barker

Dec 17, 2003, 9:44:41 PM12/17/03
to
You want to talk to Liz Barker at www.gammonitis.com ,

She has already developed ranking systems for both knockout and swiss
formats using excel.

Hope she can help...

Andrew
"fiore" <fiore_...@galactica.it> wrote in message

### Andrew Sarjeant

Dec 17, 2003, 9:48:20 PM12/17/03
to
You want to talk to Liz Barker at www.gammonitis.com ,

She has already developed ranking systems for both knockout and swiss
formats using excel.

Hope she can help...

Andrew

"fiore" <fiore_...@galactica.it> wrote in message

### Michael Crane

Dec 18, 2003, 8:58:09 AM12/18/03
to
Up to your old tricks again, Liz, posting as a third party? :-)

Michael - who always does his own posts! :-)))

Due to the overwhelming amount of spam received at this email address, only
direct or cc'd emails will be accepted, all other types, including forwarded
will be deleted from my server.

"Elizabeth Barker" <eb006...@blueyonder.co.uk> wrote in message
news:_u8Eb.36310\$5F2....@news-binary.blueyonder.co.uk...

### fiore

Dec 18, 2003, 11:31:36 AM12/18/03
to
well, i appreciate this all, but, how can i talk to Liz Barker? ;)
ciao
fiore
"Andrew Sarjeant" <andrew....@blueyonder.co.uk> ha scritto nel
messaggio news:py8Eb.36362\$5F2....@news-binary.blueyonder.co.uk...

Dec 18, 2003, 12:47:11 PM12/18/03
to
fiore <fiore_...@galactica.it> wrote:

>How is this mysterious simple one?

Here is what you get when you ask FIBS for it's rating formula:

> help formula
NAME
formula - The formulas used to calculate rating changes

DESCRIPTION
These are the formulas used to determine the ratings of a player:
Let's say that two players P1 and P2 were playing a n-point match.
The ratings of the players are r1 for P1 and r2 for P2 .

Let D = abs(r1-r2) (rating difference)
Let P_upset = 1/(10^(D*sqrt(n)/2000)+1) (probability that underdog wins)
Let P=1-P_upset if the underdog wins and P=P_upset if the favorite wins.

For the winner:
Let K = max ( 1 , -experience/100+5 )
The rating change is: 4*K*sqrt(n)*P
For the loser:
Let K = max ( 1 , -experience/100+5 )
The rating change is: -4*K*sqrt(n)*P

The 'experience' of a player is the sum of the lengths of all matches
a player has finished. Every player starts with a rating of 1500 and
an experience of 0.

ratings

>

Let's walk through this in English.

D is the difference in the two ratings. If you're rated 1850 and I'm
rated 1750, then D=100. Easy enough.

n is the match length. The real workhorse of the formula is in the
P_upset line; that calculates the probability of the lower-ranked
player winning the match.

P then becomes P_upset if the lower-rated player wins, or 1 - P_upset
if the higher-rated player wins.

K is the factor that scales for people with lower experience. Once
a player hits 400 experience, K is always 1.

So essentially, the winner get 4 * the square root of the match
length * the probability of his winning, and the loser loses that
much.

-Patti
--