by **ktown** » Thu Jun 01, 2006 2:26 pm

You can look at Vernon and Teix as case studies of their algorithm. Look at their OPS #s for the past 4 years (neither steals a lot, so it's fair).

Year Wells Teix

2003 .909 .811

2004 .809 .929

2005 .783 .954

2006 .994 .827

Up until this week, the O-rank was a total of 2005-2003. When they updated the O-Rank this week, they basically switched their calculations to use 2006-2004.

It looks like Teix should go up when you replace .811 with .827, but the weighting makes a big difference.Say for simplicity we weight current year as 1/2, last year as 1/3 and 2 years ago as 1/6 of the total.

Teix's high O-rank had leaned heavily on his past 2 years, but when you use this year's stats as half, he suddenly drops and Vern goes way up hitting .994 so far despite dropping a .909 year.