Vindicating Derek Jeter’s fielding at short (sorta)
June 24, 2008 23 Comments
Introducing OPA! OPA! is my new (still in the works) fielding system for use with Retrosheet, one that I’ve been meaning to create for a while now. Last week, I teased the beginnings of OPA!, at least the ground ball part. This week, a more full exploration of ways in which we can rate infield play without the benefit of knowing where the ball went.
First, the framework. You may be wondering what OPA! stands for. Other than my goal of making it the most festively-named fielding system out there (next time you go to a Greek wedding, they won’t be shouting UZR! or FRAA!), OPA! is short for OPAAA, or out probability added above average. Consider a ground ball. Any ground ball will do. The infielder’s job is to turn it into an out. He can either succeed or fail at this job, but several things must happen in order for him to succeed. He must have:
- Good range: he has to get himself and his glove in the neighborhood of the ball
- Good hands: he has to actually get the ball into his glove
- Good arm: he has to then throw the ball to first (or second?) and put it somewhere in the neighborhood of the first baseman’s glove
- The first baseman has to catch the ball
All of these things must happen in order for a ground ball to become a ground out. One of the major problems that I see with some of the major fielding systems is that they treat all of these as one giant package. Either the play was made or it was not. Sure, the point of the game is to make the play, but let’s think about the following situations. A ground ball to short where the SS gets to the ball, fields it cleanly, makes a throw right to the first baseman… who drops the ball. Sure, the 1B will pick up an error for his efforts, but the play not being completed, the SS gets no credit when he did everything right!
One of the things that spawned the new generation of fielding stats was an understanding that fielding percentage, indeed, the entire concept of an “error” was flawed. An error means that the fielder did something right, namely that he got to the ball. Yes, he booted it, but we don’t have a debit for those guys who are too slow to even get to the ball to begin with. So, an error actually penalizes one of the skills that you hope a player has. But, the type of error given (fielding, throwing) does tell us where things went wrong. It’s time to develop that line of logic more fully.
The average ground ball to somewhere on the third base side of the infield has an X% chance on average of being turned into an out. We can play with the parameters around pitcher handedness and batter handedness and if I had more detailed data, hit location, but there will be some number that emerges. The very act of the fielder ranging to the ball and at least stopping it from going to the outfield adds some additional percentage chance that the ball will become an out. Letting the ball through destroys what chance there was to make an out. (I’m sure most of you have figured out by this point, but if anyone’s still lagging, I’m basing this model on the idea of WPA.) If the third baseman makes the play, we ought to credit him with the out probability he adds based on his range. If the ball goes through to left field, we should assign the 3B some blame, along with the shortstop. How to chop up that blame was neatly explored last week.
But, now let’s take a look at what happens if the third baseman gets to the ball (range), but boots it (hands). He’ll be charged with a fielding error, and the out probability that he built up by getting to the ball is now gone. To more accurately reflect what happened though, we can put his range OPA in the “range” basket and debit his “hands” basket. (And if the first baseman drops the ball, we can debit his “hands” basket, while leaving the third baseman’s contributions alone.) Now, we have a much more fine-grained idea of where a player’s strengths and weaknesses are.
That’s the theory. For the numerical spaghetti and some 2007 results (including a few things about Jeter), keep reading.
Here’s the really short version of what it took me two weeks to write the proper code to do:
- Take 2007 and isolate all ground balls
- Figure out the rates of expected outs by play state (after it leaves the bat, fielder got there, clean pick, good throw, 1B catches) controlling for who fielded it, and batter and pitcher hand.
- Create a separate look at double play grounders, in which we isolate the two plays that will hopefully happen, and account for the fact that it’s harder to turn the second leg of a double play.
- On each play, code for whether the play was completed with no problems or where the play broke down (ball went through to the OF, it broke down at the “range” stage; fielder was charged with a fielding error, “hands” stage; no error, but the batter reached base OR fielder gets a throwing error, “arm” stage; 1B is charged with an error on the catch, “catch” stage) and if it broke down, who was at fault.
- Aggregate it all together, including a total “outs added above average” column.
I have to admit, the code is still in the debugging stage, but I’m getting results that make enough sense to publish. Let’s take a look at some early results for shortstops. To control for the number of chances each player received, I gave him credit for a ball in his area if a) he fielded it or b) if he bore more than a 20% blame on the ball getting through, using the division of responsibility chart from last week. I limited the following to gentlemen who had at least 100 balls hit in their general direction.
In 2007, the best OPA! per ball among shortstops in the “range” category:
- Adam Everett: .0281 outs added per grounder above average
- John McDonald: .0243
- Eric Bruntlett: .0197
- Troy Tulowitzki: .0168
- Bobby Crosby: .0138
The bottom five in range:
- Mark Loretta: -.0207
- Nick Punto: -.0231 (wow, he couldn’t do anything right last year)
- Jeff Keppinger: -.0275
- Aaron Miles: -.0303
- Derek Jeter: -.0319
I promise that I come not to bury Jeter, but to praise him (a little). However, he had the worst range of any Major League shortstop in 2007. In fact, in raw numbers, Jeter had an OPA! on range of -20.55, which was more than double the second worst gentleman, Hanley Ramirez (-9.64). The problem of course is that getting there is half the fun… and a good chunk of the play (so a lot of the out probability). A shortstop with no range isn’t a very good shortstop.
But let’s take a look at some of the other skills that a shortstop can possess, such as picking the ball cleanly once he’s gotten to the ball (“hands”). The top 5 and bottom 5 from 2007:
- Troy Tulowitzki: .0209 outs added per grounder above average
- Marco Scutaro: .0183
- Felipe Lopez: .0181
- Bobby Crosby: .0177
- Jeff Keppinger: .0166
- Brendan Harris: -.0161
- Jack Wilson: -.0181
- Aaron Miles: -.0212
- Adam Everett: -.0258 (yes, this is the bottom 5 list)
- Ben Zorbist: -.0265
Adam Everett, near the bottom? One concern that I have about this fielding system is the matter of a “range penalty.” We saw above that Everett had the best range in baseball at short last year (before he got hurt). He probably got to balls that no other shortstop would have. But some of those probably tested the very limits of his range and he had to kinda stab at the ball to even have a chance to pick it. In doing so, he might have looked like he “booted” a few ground balls. Again, those errors are the product of his spectacular range. Ideally, I’d like to factor this “range penalty” into my system, and correct for it. Not quite there yet.
Moving on to arms. Let’s look at which shortstops excelled on fielding a ground ball and then making their first throw. In some cases, that would be to second (on a double play ball) or to first (on just about everything else… yeah, there’s the occasional throw home to nab the guy there and those are factored in.)
- Ryan Theriot: .0355
- Tony Pena: .0311
- Omar Vizquel: .0290
- Adam Everett: .0258
- John McDonald: .0234
- Aaron Miles: -.0181
- Eric Bruntlett: -.0190
- Felipe Lopez: -.0265
- Ben Zorbist: -.0366
- Brendan Harris: -.0477
Bruntlett might be another range penalty guy in that here, a player would be debited if he gets to the ball, fields it cleanly, and either “puts it in his pocket” or throws but not in time. Sure, that could be a sign of a bad arm, but suppose that a shortstop has a habit of ranging deep into the hole and stopping balls from going into left field. I couldn’t make that play either.
What about turning the double play? That second throw on a double play, particularly coming across the bag for a shortstop, is a recipe for a collision with a runner who just might be bent on taking him out. In fact, I’ve shown it’s the one “gritty” and “hard nosed” play that players seem to have any skill for. It’s also one with a lower out expectancy, but once the ball is in the hand of the shortstop at second base and he makes the phantom tag of the base, there’s a certain number of plays at first we would expect him to complete on average. Prorated for number of potential DPs to complete (so, there’s already been at least a something-6 forceout, even if it’s 6 unassisted), the number of
The top 5 of 2007:
- Tony Pena: .1767 outs above average per double play ball
- Yuniel Escobar: .1634
- Ryan Theriot: .1555
- Ben Zorbist: .1536
- Cesar Izturis: .1340
The bottom 5 of 2007:
- Felipe Lopez: -.0559
- Julio Lugo: -.0559
- Mark Loretta: -.0559 (yeah, all three the same)
- David Eckstein: -.0726 (but he’s so “hard nosed!”)
- Miguel Tejada: -.0908
The fact that Zorbist was among the worst in making the first throw and then among the best in making the second throw was weird enough that I looked into whether there was any correlation between the two skills. The correlation was .514, which is decently strong. Maybe just a fluke.
Finally, what about catching the ball on a throw? The shortstop doesn’t take as many throws as the first baseman (obviously), but he does take a few, particularly on the 4-6-3 and 1-6-3 double plays. Catching the ball might not seem like much, because it’s rare that a fielder actually fails to catch a ball thrown at him, but one dropped ball could destroy a lot of built up out probability.
The shortstops who do the best catching a throw:
- Eric Bruntlett: .0104 OPAAA per catch opportunity
- Royce Clayton: .0097
- Jeff Keppinger: .0092
- Jack Wilson: .0089
- Michael Young: .0088
The bottom five:
- Felipe Lopez: -.0136
- Ryan Theriot: -.0158
- Jason Bartlett: -.0198
- Yuniel Escobar: -.0368
- Cristian Guzman: -.0507 (yikes!)
So, what of Jeter? You might notice that he avoided all of the bottom 5 lists after range. In fact, had I extended some of the top 5 lists to ten entrants, Jeter would have shown up. In fact, he ranks ninth overall in arm, ninth in turning the double play, and eighth in receiving throws. So, there are some places where he’s above average, particularly around throwing and catching. It’s fielding that’s his problem. But all told, was he the worst fielding shortstop in baseball in 2007? Here’s the top 5 and bottom 5 of cumulative OPA! divided by grounders for which the player had some semblance of responsibility. Starting at the top:
- John McDonald: .0671
- Troy Tulowitzki: .0470
- Tony Pena: .0467
- Omar Vizquel: .0427
- Adam Everett: .0332
And the bottom:
- Hanley Ramirez: -.0463 (poetic)
- Jack Wilson: -.0524
- Brendan Harris: -.0707
- Aaron Miles: -.0714
- Ben Zorbist: -.0758
Derek Jeter ranked #31 last year out of 43 qualifiers. Certainly he’s not a gold glove defender (according to the numbers), but he’s also not the worst of the worst. Finally, just to make sure, we need to remember that while Zorbist, Miles, and Wilson were awful ground ball fielders at shortstop last year, they were only part-timers with none of them logging more than 250 grounders. To some extent McDonald (can’t hit) and Everett (injured, can’t really hit either) had a similar story on the other side. Ramirez was a full-time shortstop, and so he did a lot more to damage his team by being so bad and yet so immovable from that spot between second and third. Who helped his team the most/did the most damage to his team over the course of the year?
- Troy Tulowitzki: 36.18 OPA! (also got more GB his way than anyone else in baseball… think the Rockies have a strategy?)
- Tony Pena: 29.04
- Omar Vizquel: 26.38
- John McDonald: 25.44
- Jose Reyes: 19.93
- David Eckstein: -11.76
- Michael Young: -15.50
- Carlos Guillen: -19.26
- Brendan Harris: -25.33
- Hanley Ramirez: -29.65
Jeter, for the curious came in at -8.37, good for 35th place out of 43 qualifying shortstops. Derek Jeter cost the Yankees eight ground ball outs last year compared to an average shortstop. So, all this talk about Jeter being the absolute worst in baseball doesn’t hold in my system. He’s 9th from the bottom. Consider him vindicated.
Now, I haven’t (yet) looked at liners or pop ups, and I suppose that could tilt the balance in another direction, but I have one more question for those who have been paying close attention. Last year, the two shortstop Gold Gloves went to Jimmy Rollins and Orlando Cabrera. You’ll notice that they haven’t been mentioned at all in this article until the last sentence.
Rollins ranked 18th in range, 16th in hands, 11th in arm, 13th in turning DPs, 29th in catching throws, 12th in OPA! per ball and 8th in cumulative OPA! Not bad numbers by any means, and he was usually above average, but he was middle of the pack. (What else Troy Tulowitzki had to do, I don’t know…) Cabrera was 27th in range, 23rd in hands, 24th in arm, 18th in turning DPs, 17th in catching throws, 20th in OPA! per ball, and 22nd in cumulative OPA! Same story. Tony Pena, and perhaps John McDonald, even though he was a part-timer, deserved better.
As with anything I do in Sabermetrics or life, it is not perfect and it is a work in progress. I would, of course, appreciate feedback using the lovely comment button below.