THE BOOK--Playing The Percentages In Baseball

Just so you know, you sound like Morpheus from The Matrix

That is the guy (Baker) who interviewed me for the Times article about the Mariners’ defense.  Seemed like a nice, smart, and well-informed journalist.

I’m not too crazy about listing UZR’s after 17 games and then drawing any kinds of conclusions about them. Individual UZR’s, even for a team full of players, are pretty worthless for 15 games or so.

What is more interesting is to prorate everyone’s pre-season UZR projections for the number of games they have played so far which allows us to see how good or bad of a defense each team really has this year, and is likely to have for the rest of the season, assuming no major personnel changes.

I happen to have those numbers.  First number is IF, second number is OF.

AL

Ana 15, -9
Bal -8, 6
BOS 14, -11
Cws -11, -1
Cle -19, 26
Col -2, -34
Det 21, 12
KC -7, -1
Min 4, -4
Nyy -12, 22
Oak 7, 2
Sea 3, 30
TB 11, 1
Tex -9, 5
Tor 6, -3

As you can see, Detroit and Seattle are tied for the best defense in the AL, regardless of what the numbers say so far this year.

NL

Ari -13, -6
Atl 6, -3
Chn 1, 5
Cin 6, -17
Col -2, -34
Flo -15, -3
Hou 4, -21
Lad -15, -28
Mil 3, 9
Nyn -9, 17
phi 16, -1
Pit -6, -14
SD -11, 27
SF -7, 26
SL 19, 0
Was 8, -35

Here is an interesting exercise.  Here is my estimation of the sabermetric and non-sabermetric teams:

Sabermetric

Ari
Bos
Cle
Oak
Pit
SD
Sea
STL
TB

Non-sabermetric

Ana
Atl
Bal
CWS
Cin
Col
Det
Flo
Hou
KC
LaD
Mil
Min
Nyy
Nyn
Phi
SF
Tex
Was

I am leaving out Tor, as I am not sure about them.  Not that I am sure about everyone else. It is just that Riccardi was supposedly a Beane disciple, but I think he is a terrible GM who does not understand and/or use sabermetrics.

Total defense for non-sabermetric teams = -71 or -3.7 per team.

Total defense for sabermetric teams = +60 or +6.7 per team.  Coincidence?

The fun thing about having the weekly UZR updates is that, so far, they’ve been extremely predictable.  In the first week, we saw Gutierrez make a couple ridiculous plays and the rest of the team was fine.

First week update, and Gutierrez lead the league in UZR and the rest of the team was just meh. 

Second week, we saw Gutierrez and Chavez still flashing the range but nothing quite as spectacular as week one, while Beltre made some great plays and Betancourt looked like a potted plant as balls rolled by him.  It was a disaster at shortstop. 

Second week update, and Betancourt went from average to the worst UZR of any SS in baseball, while Beltre went from average to one of the best 3B in baseball. 

Third week, we saw more bad stuff from Yuni, pretty boring week for the outfield, and more goodness from Beltre.  UZR says the same thing. 

Three weeks, three “here’s what our eyes have seen” reactions from the fan base, and three confirmations from UZR.

Cubs?

Forgot the Cubs.  Non sabermetric of course.  Add 6 runs to the “non” group.

I tried to categorize the teams without looking at their defense.  I will admit to one thing, though, and it is a perfect example of “publishing bias” and why everyone should be aware of it.

If I had found that the “non” teams had a better or equal defense than the saber ones, I would not have even mentioned this!

I would think that if you watch a team every day, as David does with the M’s (I assume), you can do a “UZR” in your head, especially if you know how it works and you take pains to be unbiased.  If you are especially skilled, you might even do a better job.

The problem with scouts is not that it is impossible to do a good job evaluating defense with the eyes.  The problem with them is that they DON’T watch the same players nearly every day and that they don’t understand the framework in which they should be mentally recording and processing the data.

In other words, it would be like asking a scout for an evaluation of a hitter’s overall value when he thinks that BA is the cat’s meow and never heard of OPS and does not understand what it represents and why it is a good measure of a player’s overall offensive contribution.  Even if he watches players every day and mentally (and on paper) records exactly what is going on, he is going to watch a player like Dunn, and think to himself, “He takes too many pitches, walks too often, strikes out a lot, and plays with no emotion.  B-!”

MGL- What were your preseason projections for Skip Schumaker?  He has been getting the bulk of the playing time at second, and even conservative projections had him at -15.  Also, Kahlil Greene has been really bad by UZR in each of the past two seasons.  I really don’t see how he could be better than -5.  I don’t see how the Cards could be +19 in the infield, even if you factor in the playing time that Brendan Ryan and Brian Barden have gotten.

Also, why do you think that Arizona is a sabermetrically inclined team?  I don’t know that much about Josh Byrnes, but at first glance they look like the opposite of a sabermetric team.  They have heavy ground ball staff, yet they have two of the worst MIF defenders in the game (Lopez and Drew).  They also signed Jon Garland over Randy Johnson (which isn’t a smart move no matter how you look at it) and they continue to put Chris Young at the top of the order, despite his abysmal OBP.

If you follow my comments on batting order, you would know that there are many considerations regarding who to put at the top of your order, other than OBP.  Young’s OBP projection is around .330, which is not horrendous for a top of the order guy.  Plus, ALL of the sabermetric teams use conventional batting orders because there really is no such thing as a sabermetric manager, at least not yet.

I don’t have a specific reason for putting ARI on the sabermetric list other than I recall reading that they have a sabermetric guy in the front office.  Somebody else that follows the team can probably shed more light on that.

Schumaker’s defensive or offensive projection?  I don’t really have a defensive one, other than translating his OF UZR to projection to the infield, of course.  I have him projected on offense at a .725 OPS, not very good.

Greene I have below average in UZR at around -4 runs per 150.  Glaus -9.  And Albert at +10.

OK, I see one of my mistakes.  I am using “games” rather than PA or defensive innings to figure each player’s defensive playing time.  For example, Brian Barden appeared in 18 games with 35 PA. I assume most of those were as a pinch hitter. I gave him credit for 18 full games at third base and he is a very good defensive player, at least according to Chone’s minor league defensive numbers.  I have him at +12 at 3B.  Same mistake with David Freese!  I credit him with 11 games at 3B at a +11, even though he also is mostly a pinch hitter.  I have Ryan with 15 games at -.3 runs.  Schumaker I have as a zero, after converting his outfield UZR projection to the 2B.  I give Thurston credit for 16 games, although he is a zero in UZR, so that won’t affect anything.

Anyway, thank you Nick for bringing my mistake to my attention.  I need to use defensive innings or something like that, although that is not readily available in the basic stat line I am using to do the team UZR projection/prorations.

That is one reason I love this blog.  Not much gets by our readers! And as I have always said, I love to have my mistakes pointed out so that I can correct them and learn from them.  If I think that I rarely or never make mistakes, either I am delusional or I am perfect.  And I am not perfect.

MGL, thanks for the reply.  Only with a La Russa run team, do you have to account for that many double switches, pinch hitters and defensive replacements, so I wouldn’t feel to bad about your method.  It’s certainly a lot better than just using cumulative UZR for the first 2 weeks of the season. 

One note relating to Schumaker.  He, as you probably know, hasn’t played second in a about decade.  Shifting that far right on the defensive spectrum is pretty much unheard of, and Skip had a little more than a month to practice at second.  If he managed to be average, I would be ecstatic, however the general consensus among most serious Cardinals analysts is that he will likely be in the “Jeff Kent” territory defensively.

OK, I made some changes whereby I go by the number of defensive innings played.  I use, however, a player’s primary position for all his innings though, so if a player played SS, 2B and 3B for 50 innings combined, and his primary position is SS, he gets credit for 50 inn at SS only.  Not perfect but much better than before.

Here are the updated numbers:

AL

Ana 12, -5
Bal -7, 6
BOS 13, -11
Cws -10, -2
Cle -18, 21
Det 18, 10
KC -11, -1
Min 4, 2
Nyy -11, 22
Oak 9, -1
Sea 1, 23
TB 9, 0
Tex -10, 7
Tor 6, -3

NL

Ari -11, -6
Atl 4, -2
Chn 1, 1
Cin 5, -11
Col 0, -27
Flo -15, -3
Hou 5, -20
Lad -10, -24
Mil -1, 6
Nyn -10, 7
phi 24, 1
Pit -7, -8
SD -6, 22
SF -6, 18
STL 12, 0
Was 10, -25

As you can see, it doesn’t change things all that much, although STL goes from 19 to 12 in the IF.  SEA is not second to Detroit but far ahead of the #2 team in the AL.

Notice how you don’t have any exceptionally bad or good teams on defense this year. Basically plus or minus 3 wins.

Skip did not move that far along the spectrum, considering that he played alot at CF, and CF = 2B.

MGL -

How did Texas not make the sabermetric list?  Anybody who trades Soriano for Wilkerson is full saber…

Also, Milwaukee may or may not be doing it.  They made inquiries for my fielding analysis a couple years back.

I presume Mil was more sabr-inclined when Blengino was there.  I have no idea of their current operation.

I don’t know for sure, but I get the feeling the Cubs are becoming more sabr inclined. When you go after Milton Bradley instead of Ibanez or Abreu, that smacks of valuing defense. The addition of Fukudome, who was never going to be a high BA or power guy, but should get on base a lot, and the signing of DeRosa, and subsequent promotion of Fontenot, seems to me at least of leaning more sabr than in the pre Pinella days.

Notice how you don’t have any exceptionally bad or good teams on defense this year. Basically plus or minus 3 wins.

Okay, but this is like when a projection system has everybody hitting below .330. Some team is going to be outside that range defensively, we just don’t know who it will be.

I think that this is an unusual year in terms of true talent team defense in that there is lots of parity. I could be wrong and it is always that way, on the average.  OTOH, as more and more teams focus on defense, I think there will be a shrinking of the spread in true talent defense, no?

Also, with lots of talk about defense lately, I think there is a perception that good and bad defensive teams add or subtract 5-6 wins a year.  Obviously not the case.  It is a relatively minor thing actually.

Well, by cumulative UZR last year, the difference between the worst defensive team (the Rangers) and the best defensive team (the Rays) was about 120 runs, which is about 12 wins.  In 2007, the difference was about the same as well.  I’m not sure if cumulative UZR is the best way to evaluate team defense, but I don’t see any glaring flaw in using it.

I could definitely see that disparity shrinking in the future, with the added value that most teams are placing on defense.

A team’s true defensive value (or a projection for defensive value) and their actual defensive performance, as measured by a stat like UZR, are two completely different things.  Part of the spread we see in actual UZR, like the 120 runs you are quoting from 2007 ad 2008 (which is rare, BTW), actually most of it, in this case at least, is random fluctuation plus measurement error.