As a follow up to my last post on objective Probabilistic Model of Range (PMR), I down loaded the latest retrosheet data, complete through 2010. I thought I might actually extend the model with vectors, but few events have hit location attached.
On the other hand, retrosheet data is very consistent from year to year with classifying balls in play as grounders, flies, liners and pops. So I added that to the model. The model now contains three objective parameters, batter-handedness, pitcher handedness, and stadium, and well as the slightly subjective batted ball type.
I also decide to look at these models longer term, so I’ll do a series of five-year studies, covering fielding from 2006 to 2010. For each model for a season, I used five years of data, not including the year in question. This way, none of the data for the season in question was used to train the model. I used the 2005 through 2010 data. For example, the model for 2006 is built from 2005, 2007, 2008, 2009, 2010. The 2009 data is built from 2005-2008 and 2010. The only exception are the models for Target Field, which only existed in 2010. I also only built the model with data on the visiting fielders, so a great or terrible defender for a team would not influence the model that much.
Just to review, PMR determines the probability of a batted ball being turned into an out based on a set of parameters. Adding those probabilities up for each ball in play gives us the expected number of outs. Calculating an index by the formula (100*actual outs/predicted outs) allows a ranking, with number over 100 good and numbers under 100 poor. When I used BIS data, I included a direction for the ball, and a measure of distance. Those elements are subjective, however, so leaving them out removes the biases of the scorers. Since the number of balls in play an individual fielder can handle is small in any year, by looking at a longer term model we should get a better picture of who are the best glove men.
Compiling the data for five years, the Red Sox were the best defensive team in the majors.
| Team | In Play | Actual Outs | Predicted Outs | Actual DER | Predicted DER | Index |
| BOS | 17014 | 11800 | 11524.4 | 0.694 | 0.677 | 102.4 |
| NYA | 17125 | 11878 | 11636.5 | 0.694 | 0.680 | 102.1 |
| COL | 17595 | 12096 | 11894.7 | 0.687 | 0.676 | 101.7 |
| NYN | 17490 | 12168 | 11973.4 | 0.696 | 0.685 | 101.6 |
| TOR | 17110 | 11874 | 11691.9 | 0.694 | 0.683 | 101.6 |
| ANA | 17487 | 12023 | 11880.8 | 0.688 | 0.679 | 101.2 |
| TEX | 17830 | 12260 | 12117.6 | 0.688 | 0.680 | 101.2 |
| DET | 17710 | 12214 | 12084.1 | 0.690 | 0.682 | 101.1 |
| SEA | 17881 | 12370 | 12253.6 | 0.692 | 0.685 | 101.0 |
| SLN | 17996 | 12452 | 12327.8 | 0.692 | 0.685 | 101.0 |
| ATL | 17229 | 11918 | 11816.6 | 0.692 | 0.686 | 100.9 |
| SFN | 16821 | 11752 | 11653.6 | 0.699 | 0.693 | 100.8 |
| PHI | 17548 | 12164 | 12077.2 | 0.693 | 0.688 | 100.7 |
| TBA | 17142 | 11856 | 11768.6 | 0.692 | 0.687 | 100.7 |
| ARI | 17339 | 11945 | 11874.9 | 0.689 | 0.685 | 100.6 |
| CHN | 16667 | 11615 | 11544.6 | 0.697 | 0.693 | 100.6 |
| LAN | 16889 | 11745 | 11690.8 | 0.695 | 0.692 | 100.5 |
| CLE | 18081 | 12369 | 12310.3 | 0.684 | 0.681 | 100.5 |
| OAK | 17441 | 12122 | 12075.0 | 0.695 | 0.692 | 100.4 |
| MIN | 17929 | 12336 | 12283.0 | 0.688 | 0.685 | 100.4 |
| SDN | 17237 | 12032 | 12019.3 | 0.698 | 0.697 | 100.1 |
| KCA | 17765 | 12096 | 12096.4 | 0.681 | 0.681 | 100.0 |
| BAL | 18048 | 12392 | 12430.5 | 0.687 | 0.689 | 99.7 |
| WAS | 18063 | 12432 | 12469.4 | 0.688 | 0.690 | 99.7 |
| FLO | 17406 | 11830 | 11860.0 | 0.680 | 0.681 | 99.7 |
| CIN | 17540 | 12098 | 12143.1 | 0.690 | 0.692 | 99.6 |
| MIL | 17408 | 11921 | 11968.0 | 0.685 | 0.687 | 99.6 |
| PIT | 18280 | 12382 | 12481.3 | 0.677 | 0.683 | 99.2 |
| HOU | 17443 | 11939 | 12083.1 | 0.684 | 0.693 | 98.8 |
| CHA | 17630 | 12044 | 12197.7 | 0.683 | 0.692 | 98.7 |
I was somewhat surprised to see the Yankees second. The team has brought in a few good defenders in recent years. The Rockies were the best team in the NL, and their speedy outfielders make a difference in the big park. The White Sox inhabit the bottom of the rankings, with the Astros bringing up the rear in the NL.
Over the next week studies will include each position, as well as defenses behind pitchers.