algorithm - Help on Elo System specifics for large group competitions -



algorithm - Help on Elo System specifics for large group competitions -

i'm trying implement elo-based ranking scheme sport website deals with.

there's few one thousand competitors, , each competition sees 50 500 of them go against clock. fastest man wins.

my initial thought race 50 people can treated 50*49/2 = 1225 one-on-one matches.

i of these comparisons in 1 go, , adjust each competitors rating @ end. i.e. if someone's rating 1600 remains 49 comparisons make, , adjusted sum of changes @ end. doesn't seem right... should doing?

the problem have if 1 (normally strong) competitor has terrible day (e.g. injury) can beaten 40+ people beat. have lower ratings him, , such rating gets pummelled. recommended k-factor of 32 see swings of thousands of points in single event... if drastically cut down k factor (say, 1) things better, sense flawed.

instead of summing of adjustments, should averaging them in way? or taking extreme value? got head in bit of twist here!

any help appreciated, give thanks you!

rather treating race of 50 1225 one-on-one matches think improve results treating 50 one-on-one matches it's player vs. max(everybody else). unless there's reason racing against 10 opponents more hard racing against best of 10 opponents in winner-takes-all scenario?

algorithm ranking

Comments

Popular posts from this blog

iphone - Dismissing a UIAlertView -

intellij idea - Update external libraries with intelij and java -

javascript - send data from a new window to previous window in php -