Sie sind nicht angemeldet.
06.12.2005, 12:18
06.12.2005, 13:34
Zitat
BITTE: Nicht aus dem derzeitigen Tabellen-Output parsen, danke.
Dieser Beitrag wurde bereits 1 mal editiert, zuletzt von »fast_tam« (06.12.2005, 13:35)
07.12.2005, 16:30
Here's the full rating algorithm:
K=32 //zone compatible k
S=250 //zone compatible scaling
C=0.5 //weighting of k-adjustment
p = 1 / (1+10^(Rate_loser-Rate_winner)/S) //expected winning-chance for winner
K_adjust = 1-abs(1-2p) //1.0 at equal rates, decreasing as rating-diff rises
K_final = K*K_adjust^C
RatingChange_winner = K_final*(1-p)
RatingChange_loser = -K_final*(1-p)
For info on ELO:
http://en.wikipedia.org/wiki/ELO_rating_system
About the K-Adjustment:
This is done to (hopefully) minimize the impact of noob-bashing. Note that it also decreases the points you can win if you play high-rated players. This also makes the high-rated players rate more stable, which is usually achieved by using lower K-Values for high-rated players. We have same K-Values for everyone, only decreasing it if the match is "unbalanced".
Dieser Beitrag wurde bereits 1 mal editiert, zuletzt von »plexiq« (07.12.2005, 16:34)
