bug-gnubg
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Bug-gnubg] match winning chances


From: Thomas Koch
Subject: Re: [Bug-gnubg] match winning chances
Date: Tue, 13 May 2008 17:06:04 +0200

That is exactly what I am searching for. Thank you very much. I suppose that even with a biased cube error rate I will get pretty good results. The rating difference D must be D = R_favorite - R_underdog, right?
Now that you are thinking about recalculating the values submitted by Kees van den Doel, how about integrating the formula into GnuBG? I think that would be quite handy and appreciated by most users!
Thanks for your efforts and good luck with the calculations. I am looking forward to the reviewed formula!

Thomas Koch



2008/5/12 Michael Petch <address@hidden>:
Considering that all you need is EPM and EPC (The error rates for the Unforced moves, and cubes) - the player records will suffice. The value N in the formula is the match length. These 3 values can simply be plugged into this reduce formula:

R = (8798 + 25526/N)*EPM+(863 - 519/N)*EPC,

R = ratings differnce based on checker, cube play, and match length values.

On 5/12/08 10:33 AM, "Michael Petch" <address@hidden> wrote:


The calculation that Gnubg uses is based upon Kees van den Doel:

http://www.cs.ubc.ca/~kvdoel/tmp/ratings/

The calculation and the variables you plug in are summarized at the top. How the calculation is arrived at it discussed in the 90% of the article after that.

Once you get a Ratings difference from that equation you then use the Fibs formula to convert the ratings difference into MWC. The part of the fibs formula you want is here:
Take the ratings difference computed in the part above and use it as the value for variable "D" in the formula below. The match length is variable "n"

First we compute U (Chances of Underdog to win) as U = 1/(10^(D*SQRT(n)/2000)+1)
To compute the chances of the favorite winning its simple computed as   1-U

So if you were playing a 15 point match (n) and the rating difference computed by Kee's forumla at the link provided was 47.13 (for example), you would get soemthing like:

D = 47.13 (Rating difference)
n = 15 (Match length)

U = 1 / (10^(47.13*sqrt(15)/2000)+1)
U = .447655

Chance of underdog Winning - U * 100 = 44.77%
Chance of Favorite winning - (1-U) * 100 = 55.23%

I haven't done any investigation, but I'm not sure if the information needed to do the ratings difference calculation (Part 1 above) is available through standard player records, but I am pretty sure if you use the relational database to store the records that all the info you need is there.

Let me know if this helps at all.

Mike

On 5/12/08 10:11 AM, "Thomas Koch" <address@hidden> wrote:

Is there a possibility to derive (e.g. calculate) my match winning chances against a certain person from the gnubg error rate shown in the player records? Let's say my error rate in the player record is 0.0200 and my opponents error rate is 0.0300. How can I calculate my match winning chances for a 5pt, 11pt, 25pt- match etc.? I would like to get something like the "luck adjusted result" from the match statistics which tells me that I am e.g. 60% favorite to win a 5pter against this opponent. What is the formula?

Thanks in advance,
Thomas Koch



_______________________________________________
Bug-gnubg mailing list
address@hidden
http://lists.gnu.org/mailman/listinfo/bug-gnubg



_______________________________________________
Bug-gnubg mailing list
address@hidden
http://lists.gnu.org/mailman/listinfo/bug-gnubg



Michael Petch
Quortech Solutions Ltd.
Cell: (403)804-5700
Home: (403)209-1628
Off:  (403)516-2612


This e-mail is the property of QuorTech Solutions, Inc. and iTechnology Customer Service and Support, Inc. and may contain confidential and privileged material for the sole use of the intended recipient(s). Any review, use, distribution or disclosure by others is strictly prohibited. If you are not the intended recipient (or authorized to receive for the recipient), please contact the sender and delete all copies of the message.



--
Der Backgammon Blog: http://graswurzelbackgammon.wordpress.com
reply via email to

[Prev in Thread] Current Thread [Next in Thread]