Wednesday, January 14, 2009

The DL Adjustment

Thanks to this link, I was able to find enough numbers to figure out the Elias Rankings injury adjustment. As a result, the rankings can now almost be recalculated with exact accuracy (However, projecting becomes a whole other issue). First, I want to show how the injury adjustment is calculated.

A: Adjustment
D: Days Missed

D=<60: A=364/(364-D)

D>60: A=(424-D)/(364-D)

Previously, I was giving all players over 60 days missed an adjustment of 364/304. You will notice, however, that there is a significant difference between the actual adjustment and that assumed adjustment. Fixing this error has led me to be able to reproduce the rankings (of the positions I've gotten to so far) with 100% accuracy.

If you want to try to reproduce the American League Catcher rankings, you can find all the data you need, right here. Just multiply each counting stat by the DL adjustment described above, rank them by statistical category, give each player a score from 0 to 31, sharing points in the case of ties. Then add up each player's score and divide by 217 (31 possible points in 7 different categories), multiplying by 100 because the score is a percentage. Voila, you've calculated the 2007-2008 Elias Rankings for American League catchers. The top 20 percent (6) are Type A free agents while the next 20 are Type B.

So why will this be difficult to project? While it's easy to work backward and notice when there is an error, there is basically no way to publicly get reliable disabled list data, accurate to the day. You may think that one or two days makes no difference. However, to those of you who went ahead and calculated the rankings, go ahead and give Ramon Hernandez 40 days -- the total of days I found in the transactions log on -- on the disabled list instead of 41. You will notice his score will drop from 73.733 to 72.350. If Dioner Navarro was closer to him, that could have been the difference between a Type A and B ranking.

I'm not ready to pronounce my system bug free just yet, but this is a significant step that should improve the accuracy of the projections for the 2009 season.

No comments: