February 03, 2005
Probabilistic Model of Range, Rightfielders
In the continuing series, here's the table for 2004 rightfielders. The results were certainly surprising to me.
Player | InPlay | Actual Outs | Predicted Outs | DER | Predicted DER | Difference |
Karim Garcia | 1361 | 100 | 88.88 | 0.073 | 0.065 | 0.00817 |
Kevin Mench | 1557 | 127 | 116.43 | 0.082 | 0.075 | 0.00679 |
Kevin Millar | 1336 | 96 | 87.01 | 0.072 | 0.065 | 0.00673 |
Brian Jordan | 1132 | 93 | 85.96 | 0.082 | 0.076 | 0.00622 |
Sammy Sosa | 3055 | 238 | 220.65 | 0.078 | 0.072 | 0.00568 |
Richard Hidalgo | 3519 | 261 | 242.20 | 0.074 | 0.069 | 0.00534 |
Jermaine Dye | 3609 | 257 | 238.50 | 0.071 | 0.066 | 0.00512 |
Craig Monroe | 1242 | 110 | 103.79 | 0.089 | 0.084 | 0.00500 |
Gabe Kapler | 1731 | 134 | 126.12 | 0.077 | 0.073 | 0.00455 |
Joe Borchard | 1397 | 99 | 93.35 | 0.071 | 0.067 | 0.00404 |
Brady Clark | 2371 | 218 | 210.91 | 0.092 | 0.089 | 0.00299 |
Jacque Jones | 3729 | 313 | 302.45 | 0.084 | 0.081 | 0.00283 |
Bobby Higginson | 3035 | 224 | 216.42 | 0.074 | 0.071 | 0.00250 |
Juan Encarnacion | 3137 | 247 | 239.42 | 0.079 | 0.076 | 0.00242 |
Abraham Nunez | 1644 | 144 | 140.83 | 0.088 | 0.086 | 0.00193 |
Jose Cruz | 4096 | 313 | 307.59 | 0.076 | 0.075 | 0.00132 |
Reed Johnson | 1190 | 88 | 86.65 | 0.074 | 0.073 | 0.00113 |
Gary Matthews Jr. | 1532 | 119 | 117.38 | 0.078 | 0.077 | 0.00106 |
Ichiro Suzuki | 4336 | 375 | 370.48 | 0.086 | 0.085 | 0.00104 |
Alexis I Rios | 2991 | 217 | 214.96 | 0.073 | 0.072 | 0.00068 |
Gary Sheffield | 3668 | 273 | 271.90 | 0.074 | 0.074 | 0.00030 |
Danny Bautista | 3525 | 265 | 265.00 | 0.075 | 0.075 | -0.00000 |
Vladimir Guerrero | 3665 | 308 | 308.45 | 0.084 | 0.084 | -0.00012 |
J.D. Drew | 3672 | 277 | 277.94 | 0.075 | 0.076 | -0.00026 |
Jay Gibbons | 1704 | 117 | 117.91 | 0.069 | 0.069 | -0.00054 |
Michael Tucker | 2674 | 209 | 211.41 | 0.078 | 0.079 | -0.00090 |
Lance Berkman | 2251 | 148 | 150.92 | 0.066 | 0.067 | -0.00130 |
Dustan Mohr | 1130 | 89 | 90.49 | 0.079 | 0.080 | -0.00132 |
Brian Giles | 4210 | 323 | 328.65 | 0.077 | 0.078 | -0.00134 |
Timo Perez | 1141 | 83 | 84.62 | 0.073 | 0.074 | -0.00142 |
Jody Gerut | 3111 | 242 | 246.65 | 0.078 | 0.079 | -0.00149 |
Austin Kearns | 1570 | 118 | 120.71 | 0.075 | 0.077 | -0.00173 |
Magglio Ordonez | 1134 | 95 | 98.02 | 0.084 | 0.086 | -0.00266 |
Ben Grieve | 1488 | 108 | 112.04 | 0.073 | 0.075 | -0.00272 |
Larry Walker | 1869 | 122 | 127.64 | 0.065 | 0.068 | -0.00302 |
Juan Rivera | 2219 | 152 | 160.11 | 0.068 | 0.072 | -0.00365 |
Wily Mo Pena | 1104 | 69 | 73.35 | 0.063 | 0.066 | -0.00394 |
Miguel Cabrera | 2558 | 171 | 181.36 | 0.067 | 0.071 | -0.00405 |
Bobby Abreu | 4240 | 311 | 329.66 | 0.073 | 0.078 | -0.00440 |
Rob Mackowiak | 1268 | 78 | 84.42 | 0.062 | 0.067 | -0.00506 |
Reggie Sanders | 1977 | 132 | 142.03 | 0.067 | 0.072 | -0.00507 |
Jeromy Burnitz | 1701 | 106 | 115.35 | 0.062 | 0.068 | -0.00550 |
Craig A Wilson | 1929 | 145 | 160.63 | 0.075 | 0.083 | -0.00810 |
Shawn Green | 1256 | 81 | 93.72 | 0.064 | 0.075 | -0.01013 |
Matt Stairs | 1483 | 108 | 127.51 | 0.073 | 0.086 | -0.01315 |
The conventional wisdom about Sammy Sosa is summed up nicely in a comment to this post:
Sosa is past his prime as a baserunner, is in serious decline as a fielder, and is losing it as a hitter.
The other thing I find very troubling about this list is the position of Ichiro Suzuki. If I could name one right fielder who I would think had extraordinary range, it would be Ichiro.
The model I'm presenting is very simple, and I'm aware of it's flaws and limitations. All of your comments have been very helpful in that regard. I'm going to push forward with the model as is for the rest of the postions before I start trying to improve the calculation, however. This way, we'll have a good baseline for comparison.
Of course, it could be that Sosa isn't as bad as we believe and Ichiro is positioning himself to make great plays at the expense of easy ones. I'll probably present a more detailed breakdown of the two shortly.
Posted by David Pinto at
09:24 AM
|
Defense
|
TrackBack (0)
Something else I noticed: Wily Mo Pena was excellent in CF (+0.00708), but he sucked in RF (-0.00394). Weird how that could be. You'd think a guy with great range in center would still have great range in right. Austin Kearns is also very lowly rated in RF (-.0.00173). Maybe it's just how the Reds position their outfielders. If that's so, then does this method hold very much weight?
Its not a problem for infielders, but for outfielders you really need to look beyond outs recorded and look at how often the plays they miss go for singles, doubles, or triples.
I don't think Ichiro has ever been great in ZR or UZR, either.
What really surprises me is the Karim and Kevins at the top of the list. If any players looked like the description "big fat oaf who should be a DH" its them.
Rallymonkey,
Interestingly, Tom Ruane at Diamond Mind has written in his "Gold Glove" review for 2001-03 that Ichiro is noticeably good at keeping doubles to singles. Tom also says that Ichiro is quite good (at times the best) RF, and that his putout totals were slightly suppressed by ball-hogging (if you want to call it that) by Cameron when they were still teammates. UZR and DRA also rate Ichiro before 2004 as a strong right fielder, though not spectacular.
Perhaps we need to look at Seattle CF ratings to see if Ichiro has again allowed the CF to take discretionary chances.
Is there evidence from SB attempts and success rate that Ichiro has lost any speed?
Centerfielders do not take discretionary chances. Everyball hit to the outfield that can be caught by the centerfielder should be caught by the centerfielder. That's his job.
I'd say with Ichiro setting the hit record with a large number of infield hits is an indicator of his not losing speed.
I think the point is that maybe Cameron and Andruw make .99 of the plays in a certain slice, while the RF make .01.
But, for other CF, maybe because they are slow, or because they shade the other way, or because they defer to a surly guy like Sosa, they only make .90 of those plays while the RF make .10 plays, in that same slice.
Essentially, if those CF had Frank Thomas in RF and Cecil Fielder in LF, they would make 1.00 of those plays, and even more in slices closer to those guys.
By the way, I'm starting a mini-project here:
http://www.battersbox.ca/archives/00002781.shtml
"Wily Mo Pena was excellent in CF (+0.00708), but he sucked in RF (-0.00394). Weird how that could be."
This is one where you have to say "small sample size" and leave it at that.
Its possible to be a better CF than corner. I was. I was very fast back in the day, but my eyesight is bad. I saw the ball much better from CF, was able to judge the angles much better than judging balls hit in my direction. If that's true about Pena, we'd need more data to say with certainty.
Also, players of my skill set don't make the majors.
Just to continue:
There are certain combination of parameters that are more noise than signal. Infield pop ups that don't travel more than 150 feet are noise, and don't tell you anything about the infielder.
***
This is why I would love if BIS, STATS, MLB.com or someone to track hang time. In my view, it is the most critical variable that is not tracked.... and, it's gosh-darn the easiest thing to track, especially for an OF.
I will point you to an excellent article by Robert Dudek (in the Hardball Times Annual), where he actually pulled out his stopwatch, and did just that. His results were even more startling than I expected.
While David uses slices and hardness of ball hit, and MGL uses the same plus distance to add better granularity, Robert's addition of hang time to the above is a huge leap forward.
Well, Matt Stair's only asset in right is his arm, which isn't measured here; but Larry Walker worse than Ben Grieve? Shawn Green was really really bad every time I saw him; but Sammy Sosa better than Richard Hidalgo? JD Drew between Vlad and Jay Gibbons? Millar scoring as high as I would expect O'Leary to?
Wow-- are you sure you ran it right? :-)
I believe that Derek Jeter had a worse OPS than Rey Ordonez in Apr 05. That doesn't make OPS "inaccurate". It means the sample size that OPS needs is more than 100 PA.
Same thing here.... if you see someoe who is out of line, it's possible that the sample size is not sufficient. If that means 120 games is not enough for fielding, then so be it.
Or of course, your eyes are deceiving you. No Cub fan will believe that of course.
I hate being one of these people, but I just can't accept the results of something that says Kevin Millar is among the league's top right fielders. God could come down from Heaven and tell me this, and I would not believe it.
I think the outfield results are missing something...
Anyway, thanks David - I always look forward to seeing these results.
OF is missing distance as a parameter.
***
David, if you have access to your old STATS data, what was the breakdown in the 1 to 4 (routine to difficult) balls in play, for IF and OF? And, what were the out-conversion rates?
The problem with rating outfielders is that so many of their putouts could go eitherway. A popup could be caught by a second baseman or a right fielder...or a center fielder instead of a right fielder...I know eyes can be deceiving, but there's no way that Sammy Sosa is a better outfielder than Ichiro Suzuki. Sosa is a circus, and Ichiro has won gold gloves and is being compared to Roberto Clemente...
Question for David Pinto:
Looking at your numbers, you have the Red Sox as one of the best fielding teams in the MLB, about 26 runs better than average (which is ~-20 plays). On the other hand, their DIPS says that their pitching staff is about +18 runs better than expected (or in other words, than it would have been with an average defense). Something's got to give. Are you saying that the Red Sox arms were THAT bad (which I doubt, not only because of how big the number is, but also because they were in the middle of the pack in terms of assists, plus pretty good with DPs)? Or is there something else I'm missing here that could explain why the Sox are rated as such a good defense by you and why that wasn't true in actuality?
Is park factor being taken into account here? It seems to me that a HR in Wrigley might go for a double or triple in, say, Safeco or Petco, where Giles seems lower than he "should" be. I would think park factor would affect the corner OF spots the most. Also, I don't see Milton Bradley up here.
But ... but ... Craig Wilson plays the bounces off the right-field wall so well.
"Looking at your numbers, you have the Red Sox as one of the best fielding teams in the MLB, about 26 runs better than average (which is ~-20 plays). On the other hand, their DIPS says that their pitching staff is about +18 runs better than expected (or in other words, than it would have been with an average defense). Something's got to give."
Unless I'm misunderstanding you, those two numbers are in agreement. If the Red Sox gave up 18 runs fewer than DIPS projects, that implies that the fielding was 18 runs [i]better[/i] than average (of course, it could just be variation or whatever).
Is there a chance that Millar is so high because of the following factors:
*small sample size?
*short porch/Pesky Pole at Fenway cuts down on balls that might other wise fall in for singles, go for doubles, etc?
*Johnny Damon catch fly balls between the two, talking away an out that Millar easily (and I use that term lightly) could have made?
"Unless I'm misunderstanding you, those two numbers are in agreement. If the Red Sox gave up 18 runs fewer than DIPS projects, that implies that the fielding was 18 runs [i]better[/i] than average (of course, it could just be variation or whatever). "
My bad. What I meant is that the Sox have a much better DIPS than ERA, by 18 runs, which means that their overall defense should be -18, while David has their defense at +26. That's a 44 run difference, and one that I'd like someone to explain. I appreciate the work being done here greatly, and plan to use the numbers for various different things, but I would like to know why this happened.
Multiple people have commented that Millar doesn't deserve such a good rating. My intuition says the same thing. I think Millar is an okay RF, but he's not one of the top 3 in MLB. In particular, I don't think he's nearly as good as Kapler, yet he gets a better difference than Kapler. Millar and Kapler played all of their 2004 for the Red Sox. However, their predicted DER's are quite different. Millar gets a 0.65, at the low end, and Kapler gets a 0.73, towards the middle. My question is: why? Did Kapler play RF in "easy" parks and Millar in "hard" ones? You'd think the reverse would have been true...