THE BOOK--Playing The Percentages In Baseball

Looks like I didn’t understand the categories well enough.  RZR is only based on opportunities that are in the “zone of responsibility”.  If a play is made outside that zone, it’s not counted in RZR, but is tracked separately as a count (OOZ).

The typical way we would do a ZR is to take all plays made and divide by the balls in the “zone of responsibility”.

Anyway, I’ll have a new post shortly.

The Fans’ Scouting Report pretty much sums up Freeman as a defensive first baseman. He’s a big, lumbering guy with very good hands.(55 - Reaction/Instincts, 35 - First few steps, 77 - Hands/Catching.) He’s also very flexible and often did the splits when stretching for a throw. That, of course, inevitably drew raves from the announcers every time it happened, and led them to conclude that he’s a great defender.

Yes. As a Braves fan it seemed Freeman dug a lot of bad throws out of the dirt.  He also made a couple of spectacular-looking plays.  I feel confident that Freeman cost the team something with his lack of range. 

Digging balls out could have merely been a result of lots of bad throws.

Alright, this is what I did.  I took all 1B with at least 500 inning played.

I calculated the player’s outs above average using their RZR and OOZ.  This was simple enough to do, and I’ll use an example.  Adrian Gonzalez had 191 in-zone plays made and 17 out-of-zone plays made, for a total of 208 plays made.  That total, 208, is a point of fact (or should be anyway… presumably, things like popups might be removed, who knows).

Anyway, the subjective element is the number of balls hit in the standard 1B zone, which I’ll just call “zone of responsibility”.  (Every 1B shares the same zone.) BIS figures this total as 234 for Gonzalez.

The average 1B makes a play at a rate of .913 outs per ball in zone of responsibility.  234 x .913 = 214 expected outs.  Since Adrian made only 208 outs, we give him -6.

UZR had him at +11 runs, which would be about +15 plays.

Since RZR, which treats all balls in zone as equivalent has him at -6, while UZR, which tracks each ball separately (and includes beyond location, things like pitcher tendencies, and park, and base-out state) has him at +15, this means that Adrian’s balls in zone were biased by -21 plays.

That is, the average 1B, given the quality of Adrian’s chances, would be not an average (0) 1B, but -21 plays 1B. 

Anyway, so this is the implied quality of chances, based on this UZR/RZR process:

QualityOpp
15 Freddie Freeman
14 Aubrey Huff
9 Brett Wallace
7 James Loney
5 Lyle Overbay
5 Daric Barton
4 Casey Kotchman
4 Mark Trumbo
3 Albert Pujols
3 Joey Votto
2 Carlos Santana
1 Matt LaPorta
0 Carlos Lee
0 Michael Morse
-1 Derrek Lee
-1 Gaby Sanchez
-1 Prince Fielder
-1 Todd Helton
-1 Mitch Moreland
-2 Adam Lind
-3 Carlos Pena
-4 Justin Smoak
-5 Mark Teixeira
-7 Ryan Howard
-7 Miguel Cabrera
-9 Paul Konerko
-11 Eric Hosmer
-20 Adrian Gonzalez

Freeman faced alot of easy chances, and Adrian faced a ton of hard chances.

So when you look at their RZR, you would have to adjust your thinking based on the quality of those chances.

So when you look at their RZR, you would have to adjust your thinking based on the quality of those chances.

Why? DRS doesn’t.

With respect to UZR.

If I repeat the same steps for UZR and DRS, this is what I get:
QualityOppUZR QualityOppDRS
15 -2 Freddie Freeman
14 14 Aubrey Huff
9 1 Brett Wallace
7 7 James Loney
5 0 Lyle Overbay
5 2 Daric Barton
4 8 Casey Kotchman
4 2 Mark Trumbo
3 2 Albert Pujols
3 7 Joey Votto
2 -3 Carlos Santana
1 3 Matt LaPorta
0 -3 Carlos Lee
0 2 Michael Morse
-1 0 Derrek Lee
-1 1 Gaby Sanchez
-1 -6 Prince Fielder
-1 0 Todd Helton
-1 2 Mitch Moreland
-2 -1 Adam Lind
-3 -6 Carlos Pena
-4 -2 Justin Smoak
-5 3 Mark Teixeira
-7 -1 Ryan Howard
-7 -4 Miguel Cabrera
-9 0 Paul Konerko
-11 1 Eric Hosmer
-20 -14 Adrian Gonzalez

Freddie Freeman’s opportunities are seen quite differently between DRS and UZR, as are Eric Hosmer.  You could have of course figured that part out just by comparing DRS and UZR directly.

Anyway, DRS thinks that Hosmer and Freeman’s context are very similar.  UZR on the other hand thinks that Freeman faced very easy conditions and Hosmer faced very difficult conditions to make plays.

This method point more as to what each system is thinking.  We see that UZR makes more aggressive context adjustments.  The standard deviation for UZR context adjustment is 7 plays, while it’s 5 plays for DRS.  This would make sense, since UZR considers more parameters.  The more it considers the wider the spread in adjustments.

Anyway, this is always what we are after here: to figure out the context of the opportunities available.

Freeman and Pena were in the top 5 for “Scoops” by a 1B, via Fangraphs.

Good point.

Whereas I’ve been treating DRS and UZR with significant overlap, it’s possible that there is enough differentiation to account for the wide differences we see, starting with scoop, and going from there.

I probably would have been better off looking at something like 3B, where they should be much more overlap.

If you use rPM for DRS scoops shouldn’t be considered at all. If you use the values listed under DRS, frankly I don’t understand what is in those besides rPM.

Can you just clarify what exactly you used for DRS and UZR? For this analysis, you’re probably best off using rPM and RngR+ErrR, respectively.

It’s hard to trust fans sometimes because their perception often gets distorted by the media so easily. Basically any time Freddie did anything decent in the field, Chip Caray sounded like Meg Ryan in the diner. That wouldbe followed by Joe Simpson writing Freddie an on air love letter. So as a fan it gets hard to tell just how good Freddie is in the field.

I dunno. Didn’t notice his range being quite as bad as Chipper or Uggla, but never observed him showing much range either.  But he does seem slick with the glove. Just think it is like a Larry Bird kind of thing. Larry had all the pump fakes and sleights of hand, and so forth to get his shot off while stationary. He could look athletic while not moving, but looked unathletic on the move. I would say that is what Freddie is. No range but nice hands.

I used the DRS and UZR columns.

Could you rerun with rPM and RngR+ErrR? UZR includes GDP ratings for 1B, which is a 2-run swing from best to worst. Like I said, I have no idea what is included in DRS that’s not in rPM - all the other columns for the DRS breakdown are blank for first basemen. Whatever it is, it’s about a six-run swing from best to worst.

My file is at home, so I can only do that tonight.

However, anyone here can repeat my steps if they want to get a head start, as specified in post 4.

You have to also consider that if you do what Colin says, rPM is not re-adjusted to an average of 0 for the season, while RngR+ErR is.

On Hosmer, the 5 or 6 of us Royals fans noticed his low numbers earlier in the season. I have MGL’s UZR data on him to see where the differences lie. I have not gotten to it yet.

There is plan on comparing video with the data to see what is happening. I am suspecting to see that he plays right on the baseline because he rarely needed to throw the ball to the pitcher if he fielded the ball.  He just takes the ball himself.  I think he would have better numbers if he moved closer to 2B, but I am waiting until the video is looked at.

Jeff Z, that’s what I’ve heard others say about Hosmer - he tends to hug the line.  I’ll have to take their word for it, because I seldom get to games in person and TV doesn’t often show defensive setup.  But yeah, his numbers have been perplexing as well, since he’s had a rep as a good defender since high school.