November 20, 2006
Probabilistic Model of Range, Centerfielders, 2006
Let's take a look at how PMR rates the centerfielders this year:
Probabilistic Model of Range, Centerfielders. Model is Based on 2006 Data Only. Minimum 1000 Balls in Play.
Player | In Play | Actual Outs | Predicted Outs | DER | Predicted DER | Difference |
Ichiro Suzuki | 1017 | 114 | 106.04 | 0.112 | 0.104 | 0.00782 |
Ryan Freel | 1211 | 127 | 119.69 | 0.105 | 0.099 | 0.00603 |
Shane Victorino | 1691 | 161 | 151.18 | 0.095 | 0.089 | 0.00581 |
Carlos Beltran | 3517 | 357 | 338.76 | 0.102 | 0.096 | 0.00519 |
Alfredo Amezaga | 1580 | 155 | 146.95 | 0.098 | 0.093 | 0.00509 |
Coco Crisp | 2814 | 246 | 232.39 | 0.087 | 0.083 | 0.00484 |
Corey Patterson | 3360 | 345 | 329.33 | 0.103 | 0.098 | 0.00466 |
Joey R Gathright | 3272 | 341 | 325.89 | 0.104 | 0.100 | 0.00462 |
Aaron Rowand | 2742 | 251 | 238.64 | 0.092 | 0.087 | 0.00451 |
Johnny Damon | 3378 | 306 | 294.10 | 0.091 | 0.087 | 0.00352 |
Rocco Baldelli | 2368 | 228 | 219.80 | 0.096 | 0.093 | 0.00346 |
Randy Winn | 1366 | 137 | 132.45 | 0.100 | 0.097 | 0.00333 |
Jim Edmonds | 2471 | 223 | 215.35 | 0.090 | 0.087 | 0.00309 |
Brady Clark | 2748 | 250 | 241.95 | 0.091 | 0.088 | 0.00293 |
Willy Taveras | 3304 | 335 | 325.37 | 0.101 | 0.098 | 0.00292 |
Reggie D Abercrombie | 1833 | 172 | 168.04 | 0.094 | 0.092 | 0.00216 |
Mike Cameron | 3723 | 367 | 360.50 | 0.099 | 0.097 | 0.00174 |
Brian N Anderson | 2996 | 305 | 300.58 | 0.102 | 0.100 | 0.00148 |
Steve Finley | 3013 | 287 | 283.68 | 0.095 | 0.094 | 0.00110 |
Juan Pierre | 4103 | 380 | 375.88 | 0.093 | 0.092 | 0.00101 |
Curtis Granderson | 4014 | 385 | 381.35 | 0.096 | 0.095 | 0.00091 |
Vernon Wells | 3918 | 332 | 330.00 | 0.085 | 0.084 | 0.00051 |
Eric Byrnes | 3208 | 270 | 268.41 | 0.084 | 0.084 | 0.00050 |
Andruw Jones | 4109 | 377 | 375.19 | 0.092 | 0.091 | 0.00044 |
Choo Freeman | 1021 | 101 | 100.81 | 0.099 | 0.099 | 0.00018 |
Chris Duffy | 2053 | 166 | 165.87 | 0.081 | 0.081 | 0.00006 |
So Taguchi | 1095 | 90 | 89.97 | 0.082 | 0.082 | 0.00003 |
Marlon Byrd | 1272 | 125 | 125.07 | 0.098 | 0.098 | -0.00006 |
Gary Matthews Jr. | 3909 | 333 | 334.90 | 0.085 | 0.086 | -0.00049 |
Chone Figgins | 2455 | 242 | 243.74 | 0.099 | 0.099 | -0.00071 |
Torii Hunter | 3715 | 343 | 347.24 | 0.092 | 0.093 | -0.00114 |
Nate McLouth | 1072 | 84 | 86.24 | 0.078 | 0.080 | -0.00209 |
David DeJesus | 1561 | 149 | 153.04 | 0.095 | 0.098 | -0.00258 |
Mark Kotsay | 3261 | 281 | 294.51 | 0.086 | 0.090 | -0.00414 |
Cory Sullivan | 2666 | 225 | 236.48 | 0.084 | 0.089 | -0.00430 |
Grady Sizemore | 4455 | 409 | 431.13 | 0.092 | 0.097 | -0.00497 |
Ryan M Church | 1172 | 122 | 128.48 | 0.104 | 0.110 | -0.00553 |
Rob Mackowiak | 1415 | 119 | 127.40 | 0.084 | 0.090 | -0.00594 |
Kenny Lofton | 2999 | 241 | 259.05 | 0.080 | 0.086 | -0.00602 |
Jose A Bautista | 1323 | 114 | 122.09 | 0.086 | 0.092 | -0.00612 |
Jay Payton | 1196 | 104 | 111.90 | 0.087 | 0.094 | -0.00661 |
Ken Griffey Jr. | 2753 | 229 | 256.68 | 0.083 | 0.093 | -0.01006 |
Jeremy T Reed | 1535 | 129 | 146.35 | 0.084 | 0.095 | -0.01130 |
The first thing I notice is that Ken Griffey shouldn't be in centerfield any more. Kenny Lofton outlived his usefulness as well at the position. On the other hand, Crisp did provide the Red Sox with good defense in center, better than Johnny Damon. Damon, however, did improve the Yankees as Bernie Williams was a big negative there in 2005.
Somewhat surprising is the neutrallity of Gary Matthews Jr. He actually made two fewer outs than expected. It's the problem of the spectacular catch. That's what we remember, that's what we see on the highlight reels, so we assume he's a great fielder. Sometimes, however, to make those plays you need to play deep, and singles fall in front of you. Can any Texas fans comment on how deep Matthews plays?
The Phillies looked very good at the position with both Victorino and Rowand. And Ichiro did a great job subbing in center this season. And among the regulars, Beltran clearly deserved his gold glove.
So Coco WAS a good CF...Never really understood why people continued to criticize his D all season long...
Coco gets his rep from having terrible instincts and takes crap routes/jumps on a lot of plays. I'm not buying these numbers.
Coco did very well in left field last year.
hmmm, i was about to comment on coco's ranking, but others beat me to the punch. this really surprised me because they've been saying he's -14 runs according UZR on SoSH. any idea how your system could come up with something so different from mgl's?
How is the predicted DER calculated?
You know, I'm surprised that Grady Sizemore doesn't rank better, considering that he's always billed as an Andruw Jones-quality center fielder. I still think Grady is potentially an MVP-type player (he bats out of postion at first in the lineup), but his negative impact on defense in the PMR is surprising, considering the hype. Maybe, David, your "spectacular catches" theory comes into play here. Although Grady is certainly no Jim Edmonds when it comes to manufacturing dives, the amount of credit he gets from Indians' broadcasters and radio guys for making probably routine plays in fashionable style is probably not surprising. (See Grady's Ladies.)
I suspect there's a connection between Coco Crisp's high rating in center field and Manny Ramirez's low rating in left field. Ramirez is very deferential on balls to the gap that the center fielder also could catch, and the geometry of Fenway (no foul territory to speak of in the outfield) also encourages corner outfielders to shade toward the gaps. Can you determine how much of Coco's plus rating by PMR and Manny's minus rating is specifically due to balls in the gap? Home/road splits would also show some sign of this, since they would have a more normal separation in road parks.
BMC, every ball in play has a probability of being turned into an out by a particular fielder. I just add up those probabilities for each ball in play, divide by balls in play, and that's predicted DER.
Amos, there are a number of differences between my system and MGL's, the biggest of which is how we factor in the park. But I have no idea how much data he uses to build the model, either, or where that data comes from.
re: victorino & roward - an embarassment of riches
the phillies hae an embarassment of riches in centerfield with victorino, who ranks third from the top in this list, and roward, who in in the top ten in this list, in centerfield. Victorino can actually hit--he was international league player of the year, and between his defensive and offensive abilities, actually earns win shares on a pace equal to Pat Burrell's. Victorino is a switch-hitter, and while only about 5'9", has vicious line-drive power, hitting doubles and triples and spraying the ball to all parts of the park. He has ability to walk and his OBA should improve as he gains more experience.
Rowand is also, I believe, a very good player, even though his offensive contributions are fairly set. His OBA is what it is, and his slugging average is what it is, but he is willing to swing and hit with men on base to drive them home, and his willingness to go through walls and his attitude in the clubhouse made a big difference in the phils last year. One seems reluctant to part with him.
having said that, rowand would benefit by playing less games and more games against lefties and fewer against righties.
One can easily envision using Roward and Victorino as late inning replacements for Burrell, who can't field at all anymore, and also using Victorino in RF, with Rowand and Burrell as right handed power bats behind Howard.
If the Phils actually sign a Carlos Lee, which would be a good pickup, he'd have to play in left field, which means they's still have to play Rowand in center, Victorino in right and Lee in Left.
If Lee played in RF and Burrell in LF, then doubly so would the Phils need Victorino in CF and Rowand to come in and caddy Burrell in lf as a defensive replacement.
--art kyriazis
David,
Thanks for answering my question. I guess I meant to ask how the probability for a particular player is calculated. I'm just wondering if players are exceeding low expectiations or falling short of unreasonable ones.