Re: calculate play ranking based on games played
- From: ime@xxxxxxxxx (Randy Hudson)
- Date: Fri, 9 Nov 2007 02:35:26 +0000 (UTC)
In article <2007110814473516807-kirakun@earthlinknet>,
Kira Yamato <kirakun@xxxxxxxxxxxxx> wrote:
On 2007-11-08 12:17:21 -0500, Robert Israel
<israel@xxxxxxxxxxxxxxxxxxxxxxxxxxxxx> said:
= (x+1)/(n+2)
So your 6 out of 6 should be worth a ranking of 7/8 or 87.5%, while
the other player's 99 out of 100 should be worth 100/102 or about 98%.
Interesting... So, in a loose sense, you are penalizing every player's
winning percentage by at least
1 - (n+1)/(n+2) = 1/(n+2)
so that players who have played more games will be less penalized.
It's not really a penalty, more of a regression toward the mean. The less
information available (fewer games completed), the greater the regression.
If the performances are below 50% actual wins, the regression will raise the
score of the player who has played fewer games more than the player who has
played more games.
--
Randy Hudson
.
- References:
- calculate play ranking based on games played
- From: formulateaffinity
- Re: calculate play ranking based on games played
- From: Robert Israel
- Re: calculate play ranking based on games played
- From: Kira Yamato
- calculate play ranking based on games played
- Prev by Date: Textbook Solution Manuals for sale...
- Next by Date: Re: Confirmation of Shannon's Mistake about Perfect Secrecy of One-time-pad
- Previous by thread: Re: calculate play ranking based on games played
- Next by thread: Re: calculate play ranking based on games played
- Index(es):
Relevant Pages
|
