scoring system
hhubey
Is there a write-up of why the scoring system on
FIBS (Gamesgrid?) works the way it does?
Back4U2 BBL
> FIBS (Gamesgrid?) works the way it does?
URL : http://fibs.com/#ratings
__________
Nardy
Belgian Backgammon League
Adam Stocks
sensitive information, and although it uses a totally different formula, it
yields results that are fairly close to the Fibs system. Ken Arnold allowed
me to have the formula to construct a graphical representation of it, which
can be seen on my website http://www.stocks49.freeserve.co.uk/.
There are two graphs to show the MWC versus rating difference for various
matchlengths, for both Fibs and Gamesgrid.
Adam
"hhubey" <hhu...@nj.rr.com> wrote in message
news:3C25361D...@nj.rr.com...
hhubey
determined from some formula?
Gregg Cattanach
Hopefully this answers some of your questions.
From GamesGrid webpage:
"How are ratings calculated? What do they mean?
The ratings system is designed to yield stable ratings that accurately
reflect the relative abilities of the players.
Each player begins with a rating of 1500. Ratings are increased or decreased
when a player wins or loses rated matches. The ratings change for each match
is based on the length of the match and the ratings difference between the
players. The longer the match, the more ratings points are risked.
Since the higher-rated player is more likely to win a match, especially as
the match length increases, a lower-rated player gets favorable odds when
playing a higher-rated player. For example, suppose a player rated 1550
challenges a player rated 1850 to a 7-point match. The player rated 1550
stands to gain 7.7 rating points if he wins, and lose only 2.9 rating points
if the higher-rated player wins. Conversely, The player rated 1850 stands to
gain only 2.9 rating points if she wins, but lose 7.7 rating points if the
lower-rated player wins.
These numbers reflect the match winning chances (MWC) for each player, based
on the difference in rating and length of the match. (In the example, the
lower-rated player has approximately 27% MWC, and the higher-rated player
73% MWC). The win/lose numbers are shown in the invitation dialogue.
There is an initial adjustment period of 500 games (called "the ramp").
During this period, your rating will change more rapidly, in order to more
quickly adjust to an appropriate level. Using the same example as above, if
the lower-rated player had only 5 experience points, he would stand to gain
30.8 rating points if he wins, and lose 11.5 rating points if he loses. (The
risks for the higher-rated player remain the same, assuming more than 500
experience points.) As your experience approaches 500 games, the rate of
change decreases to the standard rate.
Do I get more ratings change if I win by more points?
The amount of ratings change is based on the length of the match and the
difference in ratings between the players. There is no bonus for winning
more points than are necessary to win the match. For example, in a 7 point
match, winning 1024-0 has the same ratings impact as winning 7-6. In
particular, there is no benefit to winning a gammon if a simple win gives
you the match. You win the match either way, and allow your opponent to save
time and face.
For this reason, experienced players will offer a simple resignation in a
hopeless position, when the current cube value is sufficient to give their
opponent the match. Their opponent should accept it, and record the win,
even if a gammon is possible. While they are within their rights to play the
game out, there is no ratings benefit to wasting the time. They should
accept their match victory graciously, and move on to the next match.
Why do some players start higher than 1500?
GamesGrid offers players who have proven their backgammon skill in other
venues the opportunity to start above 1500. This helps those players get
matches at a competitive level. Also, it is fairer to their opponents, who
would give up substantial equity by playing a high-rated player at 1500 just
because they are new to GamesGrid.
In order to start at a higher rating, an account must be a newly created on
GamesGrid (no experience points) with a minimum membership commitment of 6
months. Transfer of membership from another GamesGrid account for this
purpose is not permitted.
We currently recognize the following equivalencies for players from other
backgammon sites:
FIBS: rating-50
Netgammon: rating-450
VOG: rating-400
Winners of major international tournaments may start at 1800. The maximum
rating transferred is 1800, with a minimum membership investment of six
months from date of transfer.
The equivalencies may change from time to time, and we reserve the right to
modify the formulae at any time in order to maintain the fairness of the
rating system."
End of quote.
--
Gregg Cattanach
Zox at GamesGrid, Zone
http://gateway.to/backgammon
gcattana...@prodigy.net
"Back4U2 BBL" <nardy.p...@skynet.be> wrote in message
news:3c256ea4$0$10326$ba62...@news.skynet.be...
nravier
If I understand well : rF=r-50, rN=r-450 then rating rF-rN=400 and
ratingFIBS=ratingNETGAMMON + 400 ???
jacques dit bezout sur FIBS.
"Gregg Cattanach" <gcattana...@prodigy.net> a écrit dans le message de
news: RnxV7.1874$Ah3.49...@newssvr16.news.prodigy.com...
Back4U2 BBL
> If I understand well : rF=r-50, rN=r-450 then rating rF-rN=400 and
> ratingFIBS=ratingNETGAMMON + 400 ???
You'll find a rating calculator here:
http://home.nordnet.fr/~fhochede/
It will give you an idea of how ratings can fluctuate
--
__________
Nardy Pillards
Belgian Backgammon League
hhubey
Gregg Cattanach wrote:
> Here is a cut and paste from GamesGrid's description of their rating system.
> Hopefully this answers some of your questions.
>
> From GamesGrid webpage:
>
> "How are ratings calculated? What do they mean?
> The ratings system is designed to yield stable ratings that accurately
> reflect the relative abilities of the players.
>
> Each player begins with a rating of 1500. Ratings are increased or decreased
> when a player wins or loses rated matches. The ratings change for each match
> is based on the length of the match and the ratings difference between the
> players. The longer the match, the more ratings points are risked.
>
> Since the higher-rated player is more likely to win a match, especially as
> the match length increases, a lower-rated player gets favorable odds when
> playing a higher-rated player. For example, suppose a player rated 1550
> challenges a player rated 1850 to a 7-point match. The player rated 1550
> stands to gain 7.7 rating points if he wins, and lose only 2.9 rating points
> if the higher-rated player wins. Conversely, The player rated 1850 stands to
> gain only 2.9 rating points if she wins, but lose 7.7 rating points if the
> lower-rated player wins.
>
> These numbers reflect the match winning chances (MWC) for each player, based
> on the difference in rating and length of the match. (In the example, the
> lower-rated player has approximately 27% MWC, and the higher-rated player
> 73% MWC). The win/lose numbers are shown in the invitation dialogue.
How do we know that someone rated 1850 has MWC of 73% when
playing against someone rated 1550? Is this an empirical determination
or is it derived from some other equations? Obviously the point win/loss
e.g. 7.7 for the 1550 and 2.9 for 1850 closely matches the 73% and 27%
MWC.
Again are these derived empirically or based on some calculation?
Douglas Zare
hhubey wrote:
> Gregg Cattanach wrote:
>
> [...]
> > These numbers reflect the match winning chances (MWC) for each player, based
> > on the difference in rating and length of the match. (In the example, the
> > lower-rated player has approximately 27% MWC, and the higher-rated player
> > 73% MWC). The win/lose numbers are shown in the invitation dialogue.
>
> How do we know that someone rated 1850 has MWC of 73% when
> playing against someone rated 1550? Is this an empirical determination
> or is it derived from some other equations? Obviously the point win/loss
> e.g. 7.7 for the 1550 and 2.9 for 1850 closely matches the 73% and 27%
> MWC.
>
> Again are these derived empirically or based on some calculation?
A few people were born with 1850-shaped birthmarks on their foreheads, and others
belonged to the notorious "1550" prison gang. Several billion backgammon games
between these groups were recorded from the late 19th century through World War
II, and analyzed in a surprisingly technical filibuster of Strom Thurmond in 1975.
Really, what sort of empirical observation can you imagine?
The elo system or backgammon is modelled on the elo system of chess--do you want
someone to recount and justify the ideas and inaccuracies in the chess rating
system? As in chess, people greatly disagree about what rating they ought to have,
and three players might be intransitive favorites (a over B over C over A). So,
what fine details can you expect? You can analyze the system. You can try to test
whether it is an accurate prediction of the probability that one player will win.
Much is known and already discussed about the rating system. A good starting point
is the section on ratings in http://www.bkgm.com/rgb/rgb.cgi?menu . For more
mathematical analysis of the system, check
http://www.gammonvillage.com/news/article_display.cfm?resourceid=764 and the
references contained in that article.
Douglas Zare
hhubey
Douglas Zare wrote:
> hhubey wrote:
>
> > Gregg Cattanach wrote:
> >
> > [...]
> > > These numbers reflect the match winning chances (MWC) for each player, based
> > > on the difference in rating and length of the match. (In the example, the
> > > lower-rated player has approximately 27% MWC, and the higher-rated player
> > > 73% MWC). The win/lose numbers are shown in the invitation dialogue.
> >
> > How do we know that someone rated 1850 has MWC of 73% when
> > playing against someone rated 1550? Is this an empirical determination
> > or is it derived from some other equations? Obviously the point win/loss
> > e.g. 7.7 for the 1550 and 2.9 for 1850 closely matches the 73% and 27%
> > MWC.
> >
> > Again are these derived empirically or based on some calculation?
>
> A few people were born with 1850-shaped birthmarks on their foreheads, and others
> belonged to the notorious "1550" prison gang. Several billion backgammon games
> between these groups were recorded from the late 19th century through World War
> II, and analyzed in a surprisingly technical filibuster of Strom Thurmond in 1975.
Were you among the prison gang or do you wish you were?
What the hell is your problem? Either provide the answer or keep quiet or
say you don't know.
Everyone has an interest. Mine at this point happens to be (among many
others) to provide a decent rating system based on clearly stated
assumptions (or axioms) or empiral observations and a provably
honest random number system for the Century of Internet Gaming.
Back4U2 BBL
>
> What the hell is your problem? Either provide the answer or keep quiet or
> say you don't know.
>
> Everyone has an interest. Mine at this point happens to be (among many
> others) to provide a decent rating system based on clearly stated
> assumptions (or axioms) or empiral observations and a provably
> honest random number system for the Century of Internet Gaming.
>
And I'm sure if you will do the effort to search either the newsgroup or
the internet, you'll find an answer for your many unasked questions.
--
__________
Nardy Pillards
Douglas Zare
hhubey wrote:
> Douglas Zare wrote:
>
> > hhubey wrote:
> >
> > > Gregg Cattanach wrote:
> > >
> > > [...]
> > > > These numbers reflect the match winning chances (MWC) for each player, based
> > > > on the difference in rating and length of the match. (In the example, the
> > > > lower-rated player has approximately 27% MWC, and the higher-rated player
> > > > 73% MWC). The win/lose numbers are shown in the invitation dialogue.
> > >
> > > How do we know that someone rated 1850 has MWC of 73% when
> > > playing against someone rated 1550? Is this an empirical determination
> > > or is it derived from some other equations? Obviously the point win/loss
> > > e.g. 7.7 for the 1550 and 2.9 for 1850 closely matches the 73% and 27%
> > > MWC.
> > >
> > > Again are these derived empirically or based on some calculation?
> >
> > A few people were born with 1850-shaped birthmarks on their foreheads, and others
> > belonged to the notorious "1550" prison gang. Several billion backgammon games
> > between these groups were recorded from the late 19th century through World War
> > II, and analyzed in a surprisingly technical filibuster of Strom Thurmond in 1975.
>
> Were you among the prison gang or do you wish you were?
Any guesses whether more people found your comment funny than mine?
> What the hell is your problem? Either provide the answer or keep quiet or
> say you don't know.
I have a problem with you because you seem to be a troll. You ask questions that you
should be able to answer with a little thought, such as whether the rating system was
derived empirically. Come on, think about it.
After asking simple questions, you ignore informative responses that you receive, and
repeat your questions. For example, did you read the references I gave after the
facetious paragraph? It would be more appropriate to thank the people who have taken
their time to help you. At least this time you didn't insert inflammatory accusations in
unrelated threads.
I'm happy to help people who want help, but not those who want to be spoon-fed
information rather than to be given references which contain the information.
> Everyone has an interest. Mine at this point happens to be (among many
> others) to provide a decent rating system based on clearly stated
> assumptions (or axioms) or empiral observations and a provably
> honest random number system for the Century of Internet Gaming.
I hope you do a good literature search. I recommend http://www.google.com and
http://www.ams.org/mathscinet (the latter might only work with some ISP's, such as
academic institutions).
hhubey
I see the error of my ways. Pursue your interests. ONe day
you too will become an adult and will have pursuits bigger
than being funny on rec.gammon.
qwerrk
> Now that I see your site, and see that you are a pre-adult
> I see the error of my ways. Pursue your interests. ONe day
> you too will become an adult and will have pursuits bigger
> than being funny on rec.gammon.
>
> Douglas Zare wrote:
>
> > hhubey wrote:
> >
> > > Douglas Zare wrote:
> > >
> > > > hhubey wrote:
> > > >
> > > > > Gregg Cattanach wrote:
> > > > >
> > > > > [...]
I'd trade "pre-adult" for post-asshole anytime. Perhaps it's just
your rhoids acting up? After you stop by the proctologist's, go visit
the ophthalmologist and SEE whether something might be done with your
errant vision.
TRHanlon