http://research.microsoft.com/en-us/pro ... tails.aspx
heres the formual for true skill that you where going to research,
and a virtual calculor as well..
http://atom.research.microsoft.com/true ... lator.aspx
(and I believe the same system is used in golf and tennis).
But, it also gives points based on place finished and there is no "defense" for playing against someone who is better.
and how should the number of people involved in the game affect the points, and how should the difficulty affect the number of points?
New point = (old points + ( k * ( score – win %) ) )
Win %
1
Win Probability = ------------------------------------------------------------------------
10^(( Opponent’s average rating – Player’s Rating)/400) + 1)
Score:
N + 1 – Rank Finished
----------------------------
N
where:
DNF = 0
K:
K = ( N * puzzle rating)
Puzzle rating:
Easy: = 1
Medium 2
Hard = 4
Very hard = 8
sudoking wrote:What if instead of (loser's score / winner's score) we just used (loser's score / winner's score) ^ 2 i.e. squared to amplify the difference?
that would still have the same effect, winniers points increases constatly
strmckr wrote:that would still have the same effect, winniers points increases constatly
see 2 post.
P.S.: The other weakness of the ranking system has been adressed by others in this thread. I just want to confirm:
It is obvious that a 16000 player has an estimated win percentage of at least 99.9% against a 1000 player. Thus if he wins a race he should win only fractions of a point and if he looses a race he should loose tons of points. In the current ranking system he can still win 1,25 points and lose only 100 points (because the formula (loser's score / winner's score) * 20 is capped at 100 points). Therefore the points will go up.
Return to Suggestions & Bug Reports
Users browsing this forum: No registered users and 0 guests