No huge changes to the rating process this year, though I did make two fairly significant changes: better matchup data & a better representation of the average batter/pitcher. I rated a total of 219 batters and 196 pitchers.
Instead of relying on the MLB figures, I counted up all the PAs from last year’s TV games (both sides) and classified them according to match up. I included a separate category for switch hitters. Totals:
LHP RHP
--- ---
LHB 282 752
RHB 434 790
SHB 182 416
(Total gives an average of 39.7 PAs per game per team.)
How these translate to the ranking parameters for batters:
$freq{'R'}{vR} = 0.645; # freq of R arms seen by R batter
$freq{'R'}{vL} = 0.355; # freq of L arms seen by R batter
$freq{'L'}{vR} = 0.727; # freq of R arms seen by L batter
$freq{'L'}{vL} = 0.273; # freq of L arms seen by L batter
$freq{'B'}{vR} = 0.696; # freq of R arms seen by B batter
$freq{'B'}{vL} = 0.304; # freq of L arms seen by B batter
…and for pitchers:
$freq{'R'}{vR} = 0.403; # freq of R bats seen by R pitcher
$freq{'R'}{vL} = 0.597; # freq of L bats seen by R pitcher
$freq{'L'}{vR} = 0.686; # freq of R bats seen by L pitcher
$freq{'L'}{vL} = 0.314; # freq of L bats seen by L pitcher
Lots of lefty bats in this sample. In 2007 MLB, righty pitchers saw 47% lefties. In the 2009 TV ABL games it was 60%.
The second change was to use only the better players for the applied averages. Last year I used all the rated players to form the average. This year I took a guess at an ABL regular cut-off. I selected about the top 132 batters (12 per team). That put the cut off at Ryan Church. (He’s an approximation of the worst guy expected to get regular playing time.) I took the top 44 starters (4 per team), which cuts off at Kip Wells. Same for shorts, which cuts off at Chad Durbin. Allowed the top 11 closers, which is a cut-off at Chad Qualls. Didn’t have a good answer for weighting starters & relievers, so I threw them all together. As expected, the averages had a marked effect of lowering the ratings of everyone.
Thanks for the interesting look into the Perfectos ratings. I think the idea of adding if a player hits righties or lefties more into the ratings in intriguing.