THE BOOK--Playing The Percentages In Baseball

According to the Fans, the best fielding teams are: Rays, Cubs, Angels, Dodgers.  Those 4 averaged +3 runs in mgl’s list above.

The 4 worst fielding teams are: Marlins, Mariners, Nats, Pirates.  They averaged -22 runs each in UZR, so that’s good.

***

MGL has Royals as the best-fielding team in MLB, while the fans have them 5th worst.  If you ask Royals fans, the only one close to a standout is Pena.  The others are either close to average for their positions, or worse.

MGL, if you want to compare, here are the results:
ID Pos Rating GP Player
7731 6 69 58 Pena Jr., Tony
5359 4 61 79 Grudzielanek, Mark
7348 8 58 69 Gathright, Joey
7232 7 57 92 DeJesus, David
7907 5 56 118 Gordon, Alex
7028 2 54 37 Olivo, Miguel
7365 9 54 111 Teahen, Mark
7899 8 53 15 Maier, Mitch
8260 6 49 63 Aviles, Mike
6560 3 48 92 Gload, Ross
7832 4 44 24 Callaspo, Alberto
7056 2 32 86 Buck, John
5753 7 31 77 Guillen, Jose
6935 4 27 39 German, Esteban
7634 3 22 24 Butler, Billy

ID is the STATS id.

Pos is his primary position (as of Aug 20).

Rating is set so that average is 50 over all positions (60 at SS/CF, 55 at 2B/3B, 45 at LF/RF, 40 at 1B, more or less).

GP is innings/9, as of Aug 20.

To convert Rating to runs, multiply by 0.7 (0.9 for SS, 0.8 for 2B, 0.7 for 3B, CF, 0.5 for LF/RF, 0.4 for 1B, more or less).

So, Pena is a 69, compared to 60 for the average SS, so he’s +9 ratings.  Times 0.9 makes him +8 runs (more or less), per 162 games.

Two huge differences between the UZR projections and the combined BIS/STATS zone ratings that Justin publishes.  One is that the ZRs have the NL with much better fielders and the Royals as one of the worst fielding teams in the majors.  Data from this post, as of September 5th:

http://www.beyondtheboxscore.com/2008/9/9/609347/fun-with-justin-s-stats

OAK 46
TOR 40
STL 39
PHI 38
HOU 37
MIL 25
CHN 22
ATL 21
TB 20
SD 19
CLE 16
LAA 15
NYN 15
LAN 14
WAS -2
DET -5
COL -6
SF -7
SEA -7
CIN -11
PIT -15
CHA -21
BOS -24
ARI -25
BAL -35
TEX -35
FLA -36
MIN -41
NYA -53
KC -57

If “The Fans” and UZR disagree, I’m not sure how to determine who’s wrong, but isn’t a possible problem with “The Fans” is that it’s not actually one group of “Fans,” it’s 30 different groups of fans. Maybe Royals fans are just turned off by years of failure and have trouble seeing anything positive about the team. Maybe Royals fans are accustomed to seeing above average defense due to excellent Royals teams of the past (hypothetically, I have no idea whether that’s been the case) or the overall high level of defense in the AL Central (the rest of the AL Central is +50 total in MGL’s chart above). Maybe this particular group of Royals fans is just a harsh group of critics.

I think Tango has set up the project in a way to eliminate a lot of the biases (I think breaking fielding down into a bunch of different components forces the voters to really look at things pretty closely and to avoid snap judgments), but it’s unavoidable that a group of people who watch mostly Royals games is going to have a somewhat different perspective than a group of people who watch mostly Dodgers games. If MGL is right that the AL as a whole is better defensively, then NL fans might have a different perception of “average” than AL fans.
Would it make sense to consider some kind of team-by-team adjustment, so that if it seems like Royals fans are being overly harsh overall, all the Royals players get a boost?

FWIW, here are the same numbers for actual YTD Stats ZR (original version as pulled from CNNSI) through yesterday’s games. 

TM IF OF RS
Ari -30 -1 -31
Atl 12 -23 -11
Bal -16 16 0
Bos 17 -50 -33
ChC -1 -3 -4
Cin -4 0 -4
Cle -9 25 16
Col 15 -18 -3
CWS 7 -10 -4
Det 10 16 25
Fla -29 8 -22
Hou 27 -34 -7
KC -21 16 -5
LA 16 7 23
LAA 20 10 30
Mil -7 4 -3
Min -15 0 -15
NYM -31 19 -12
NYY -20 -26 -46
Oak 13 26 38
Phi 23 16 39
Pit -4 -19 -23
SD 5 36 41
Sea 0 -27 -26
SF -5 5 0
StL 28 -6 23
TB 1 12 14
Tex -24 -11 -35
Tor 37 4 41
Was -15 10 -6

I just remembered that I separate the leagues when calculating my RS, so here’s the Stats ZR data when the leagues are combined.

TM IF OF RS
Bal -15 13 -2
Bos 17 -51 -34
Cle -9 23 14
CWS 7 -12 -5
Det 10 14 24
KC -20 13 -7
LAA 20 9 29
Min -15 -2 -17
NYY -19 -28 -47
Oak 13 24 37
Sea 1 -28 -27
TB 1 10 12
Tex -24 -13 -37
Tor 37 2 39
Total 5 -25 -20

TM IF OF RS
Ari -31 1 -30
Atl 11 -21 -10
ChC -2 -1 -3
Cin -4 1 -3
Col 14 -17 -2
Fla -30 9 -20
Hou 27 -32 -6
LA 16 8 24
Mil -7 5 -2
NYM -31 20 -11
Phi 23 17 40
Pit -4 -17 -22
SD 5 37 42
SF -5 6 1
StL 28 -4 24
Was -16 12 -4
Total -5 25 20

Here is what I have for the KC team.

And remember that my numbers use each player’s defensive projection based on their last 3-4 years’ UZR.

Fans are basing their ratings, I assume, on how each player was perceived to play defense this year.

Player, position, games played this year, UZR projection at their primary position (whether they played that position this year or not) per 150 games

Callaspo 2B 62 1.8
Gordon 3B 122 -1.2
Dejesus LF 124 10.4
German 2B 83 -1
Smith 2B 15 -2.5
Gathright CF 98 9.3
Guillen RF 140 -6.9
Grudz 2B 86 3.7
Teahan RF 136 5.4
Aviles SS 88 0
Maier CF 21 0
Gload 1B 121 -1
Pena SS 87 6.6

I don’t see a whole lot of disagreement other than Callaspo and German and I don’t have much data on them anyway, especially Callaspo, so I would definitely tend towards the fans.

I did not include Butler because his primary position in my database was listed as PH rather than 1B.  I have him as -2.8 at 1B with corresponds with the fans as well.

Here is the same list, but using 2008 actual UZR, not projections:

ARI -7 -21 -28
ATL 21 -13 +8
CHN -7 12 +5
CIN -6 -27 -33
COL 2 -21 -19
FLO -33 11 -22
HOU 27 -10 +17
LAN -3 -9 -12
MIL -20 9 -11
NYN -15 27 +12
PHI 11 6 +17
PIT -16 -17 -33
SDN 2 15 +17
SFN -8 6 -2
SLN 27 1 +28

ALA 10 9 +19
BAL -23 8 -15
BOS 18 -15 +3
CHA 20 -1 +19
CLE -9 31 +22
DET -6 3 -3
KCA 6 22 +28
MIN -8 11 +3
NYA 2 -1 1
OAK 11 11 +22
SEA 1 -14 -13
TBA 3 7 +10
TEX -24 -10 -34
TOR -9 -20 -29

Question: How are the Red Sox rated so bad defensively for all projections?  I assume that it is soley because of Ramirez.  Crisp, Ellsbury, and Drew are all above average defensive fielders (I am a Red Sox fans, watch most games, and yes, may be biased).  When Ellsbury plays the corners (LF, RF) he seems to be even better (not sure if this is supported in UZR).

With all do respect to MGL and URZ, the Pirates at -2 overall is truly a misnomer.  I assume because of the higher gb and bip rates you’re crediting them a lot on the backend.

Now that Bay and Nady are gone, Wilson has collected dust almost half the year, Doumit is behind the dish, Sanchez with his bum shoulder all year and lack of range, rookies Moss and LaRoche in corner positions and gloving/running routes like rookies do, and more pitch-to-contact arms than Carter makes pills, if they aren’t on a -50 pace I’d be shocked.

Mike, the 2008 zone ratings have Bay at -10 runs already and Coco in the same range.  Ellsbury is about average and Youk/Pedroia are merely good, not great.

Straight ZR has a problem with Fenway LF and also Houston LF, right? SG has Red Sox outfielders at -51 and Astros outfielders at -32. UZR says -15 and -10 respectively.

---

MGL, what are the numbers for WAS in post #6?

Here is a Google Docs spreadsheet comparing the numbers in posts #5 and #6.

http://spreadsheets.google.com/ccc?key=pso80GqIAQP3kJFOKX7Y4IQ&hl=en

The biggest disagreement between UZR and ZR is the Toronto infield. UZR says -9 and ZR says +37. Both of these are actual 2008 performance if I understand correctly. There’s a similar gap for the Jays’ outfield: UZR -20, ZR +2.

Toronto’s DER is .704 compared to the AL average of .692. That is 3rd in the league. They lead the AL in GB% (TOR 47%, AL 44%) and trail the AL in infield flies per fly ball (TOR 8%, AL 10%). They allow a bit fewer line drives than average (TOR 19%, AL 20%). I would guess that a team with lots of GB and few infield flies would have a lower than average expected DER, though of course it is hard to tell since not all GB, LD, etc. are created equal.

Plus-minus says that the BIP given up by Blue Jays pitchers have indeed been harder to field than average, by 11 plays. But the fielders have been 61 plays above average as a group, leading to the high DER. This roughly agrees with the ZR numbers. (+61 plays versus +39 runs)

If we believe UZR that the Toronto fielders have been -29 runs as a group, but the DER is still high, the BIP must have been very easy to field. This is in spite of the high number of GB and the low number of infield flies. Is there a strong park effect associated with the Rogers Centre?

Anyway, I think this is worth looking into in more detail. When the best fielding team in the AL according to STATS ZR and BIS +/- is the second-worst according to UZR, something strange is going on.

By the way, all the data in my previous post is from the THT team page:

http://www.hardballtimes.com/main/teams/

WAS -9 -20 -29 (actual UZR this year)

PIT was an actual -33 this year, but according to the projections:

Name, 08 Primary pos, 08 games, current projection per 150 games

Laroche 1B 124 4.1
Bixler SS 43 -1.9
Gomez 3B 89 -6.9
Mientky 1B 113 3.1
Sanchez 2B 132 -8.7
Wilson SS 81 2.1
Bay LF 144 -5.9
Bautista 3B 120 -8.5
Cruz SS 8 0
Rivas SS 76 -1.1
McLouth CF 139 -14.7
Morgan CF 47 17.9
Pearce RF 26 .3
Nady RF 134 -.6

That is actually -8 -13 -21 for the whole team and not the 4 -6 -2 I originally posted.  I changed the methodology a little.  Here are the updated numbers for all the teams, again, using current projections prorated to games played for each player.

Basically these numbers represent what we would expect right now if each team fielded a team that consisted of a composite of all their players weighted by playing time, in 2008.

NL

SDN 6 17 +23
PHI 28 1 +29
SLN 8 -5 +3
NYN -17 35 +18
CHN 13 12 +25
COL 19 -20 -1
ATL 19 -13 +6
SFN -7 22 +15
PIT -8 -13 -21
LAN -16 -15 -31
ARI 3 -17 -14
HOU 3 -10 -7
MIL -22 5 -17
CIN -10 -37 -47
WAS -9 2 -7
FLO -51 -5 -56

AL

KCA 11 20 +31
CLE -10 46 +36
OAK 22 18 +41
CHA 22 -7 +15
TOR 17 3 +20
ALA 8 8 +16
BAL -6 10 +4
TBA 4 -8 -4
BOS 18 -37 -19
DET 4 17 +21
MIN -9 13 +4
SEA 10 -19 -9
TEX -25 -25 -50
NYA -24 3 -21

As you can see, that changes things quite a bit.  The numbers are still not rigorous, as I am just using each player’s primary position, but all the games they played.

Actually, here are the more rigorous team total UZR numbers for 2008.  These are as of a few weeks ago, though:

NL

SDN -4 24 +17
PHI 11 5 +16
SLN 15 4 +19
NYN -24 39 +15
CHN -2 9 +7
COL 18 -28 -10
ATL 18 -17 +1
SFN -7 6 -1
PIT -18 -18 -36
LAN 2 -4 -2
ARI 5 -19 -14
HOU 27 -15 +12
MIL -10 7 -3
CIN -1 -26 -27
WAS -18 -16 -34
FLO -22 18 -4

AL

KCA 0 19 +19
CLE -13 28 +15
OAK 11 17 +28
CHA 23 2 +25
TOR 20 1 +21
ALA 9 4 +13
BAL -24 9 -15
TBA 1 5 6
BOS 21 -16 +5
DET -3 5 +2
MIN -11 2 -9
SEA 1 -21 -20
TEX -21 -11 -32
NYA -3 -11 -14

I agree, the Royals Fans’ and UZR match pretty well (I nicked Gathright 4 runs in UZR for his puny arm).  We’ve got pretty decent general agreement.  If you have say a 3 run disagreement for 7 players, and a totally different agreement for 1 player at say 10 runs, that gives you a 30-run disagreement, which is substantial.

The top 4 fielders are the same in both systems, and they both hate Guillen.  It’s in the middle guys where perhaps the Fans are much harsher than they should be, especially for an overall poor team to begin with (showing bias in balloting).

So, relatively speaking within each system, we’re pretty much the same.  It’s just in the extremeness that we’ve got differences.

MGL, you’ve said before you remove pitcher hitting from UZR, right?

The Dial method of zone rating runs looks at each league separately, so both leagues will sum to zero.  If you look at both league together, the NL will appear to have better ratings, because of pitcher hitting.

But it’s no surprise to me that the AL is superior in fielding.  Why not?  They are better at everything else.

I don’t remove pitcher hitting from the UZR data.  I am not sure that I should.  I wouldn’t think that it would make much difference either way.  Especially if the batted ball speeds are accurate.  Even if one league had a bunch of hitters who only hit easy to field balls (such as pitchers), it should not change anyone’s UZR much, should it?  Although since outfielders play shallow against pitchers, maybe it will look like they are making good plays when they catch fly balls from then.  Ground balls it probably does not make all that much difference, although they do get to play deep versus pitchers, I would think, since most of them are slow or simply don’t run out ground balls very hard.

I am not sure that the AL is much better at hitting anymore.  The league that spends more per player is likely to have better players of course at all skills.  But since defense is least understood by the teams, that is probably the one skill that is not too heavily correlated with salary, plus if a team or teams like to spend a lot of money on offense, they probably sacrifice defense.  There is also going to be a lot of random differences between the leagues in any one year or years no matter how much money each league is spending (within reason).

Plus, you should have more DH types occasionally playing in the field in the AL, so that maybe they should have worse fielding.  Then again, you can put a bad fielding player at DH in the AL, but you must use him in the field in the NL, so maybe the AL should have better fielders.  Who knows?

Plus, even though I establish a baseline for each “batted ball bucket” using 5 year data from both leagues, because of park differences between leagues, it is not real clear that even though the AL UZR is better than the NL UZR right now, that this a real effect.  I could just have easily zero’d out each league like everyone does with hitting and pitching and Dial does with fielding.  One argument for that is that the batted balls are from different hitters and pitchers even though the descriptions may be the same (plus the aforementioned park differences).

According to the Fans’ Scouting Report, the AL has slightly better fielders, with around a 0.7 runs gap per 162 G.  So, for a full season, that makes the average AL team about +3 runs and the NL team -3 runs.

Comparing MGL’s Pirates to the Fans: not much agreement.  Fans love Wilson, and UZR thinks he’s average.  Fans are a bit up on McLouth and Bay, and UZR no likey. Rivas, Bixler, Pearce: detested by Fans. UZR has them each as average.

Minky, Sanchez, Bautista, Nady, Gomez: agreed.