Chronicles of the lads is working on converting the Probabilistic Model of Range numbers to runs. Start with the previous link and work forward in time.
Update: It’s open source basebal research!

Here is the table lising 2004 centerfielders on the field for 1000 balls in play.

The first thing I notice is that Andruw Jones is very good and Bernie Williams is very bad. So in this case the system appears to be getting the end points right. The thing that really surprises me is the equality of Biggio and Beltran. Here’s a table comparing them just with the Astros:

Beltran was much better with the Astros than Biggio; his poor fielding was a result of his time with Kansas City last year.

So the question for the Mets is, which Beltran will show up in centerfield next season? Neither is better than Cameron, and one is a lot worse.

There’s a great discussion of the Probabilistic Model of Range going on at Dodger Thoughts. I was going to leave the following comment, but I keep getting an error so I’ll leave it here.

I think it’s important to realize that I’m not measuring total defense here, I’m simply trying to measure range. So turning double plays is important, but I’m not trying to measure that here.
As for differences between my system and UZR, the two are not exactly alike. The idea is the same, to look at the probability of fielding a certain ball, but I know MGL adjusts for parks differently and his zones are very different from my slices. And for all I know he’s using a different set of data as well.
This is only a start. There’s a long way to go before we feel comfortable with these numbers but I do believe we’re going down the right path. All your comments are very helpful.

Tom G likes what the Probabilistic Model of Range is telling him about the Phillies defense and the composition of the pitching staff.

Edward Cossette and others have raised objections to my characterization of the Nomar deal (see comments). Edward writes:

To me, your data only confirms the veracity to Theo’s reasoning that defense was the reason for the trade.
As others have pointed out, Epstein didn’t have the luxury of hoping that Nomar was just “rusty.”
Indeed, isn’t that the whole point of using stats to make decisions, i.e, to remove the “gut feeling” aspect of evaluating players?
It’s great that the numbers show Nomar got better after the trade and may in fact have been “rusty.” But it’s even better to have a GM that saw a problem and did something about it.
Meanwhile, you have absolutely nothing but pure supposition to support your argument “that defense was an excuse to move a player the Red Sox no longer wanted.”
That’s kind of weird for a stats guy isn’t it?

I’m willing to admit that Edward has a point. So I’m going to step back from my earlier statement and look at the numbers again.
There are two things I look at as a stat guy. One is the number, the second is the context. A month and a half is a short time frame for an evaluation. Anything can happen in 100 or so AB (look at Jeter’s April hitting numbers). And anything can happen on 100 ground balls or so. Nomar came back rusty. You can see it in his June hitting numbers. But by July he had recovered his swing. Why wouldn’t his fielding numbers come back also? He did show range improvement in July, but his numbers were still poor. What was the context for believing the small sample size of poor fielding numbers were valid?
And I will admit that I haven’t looked at context either. One is the context of his injury. I don’t know how well the injury healed. It was good enough that he could hit well, but not good enough that he could play everyday. Obviously, the Cubs thought the injury healed well enough that they were willing to take Nomar in trade. It’s possible that the Red Sox thought that Nomar’s poor range would not improve due to the tenderness of his foot. That proved to be incorrect.
The other is the long term context of Nomar’s fielding. Were his poor fielding numbers the continuation of a trend? This one I can research. Yes, it was the continuation of a trend. In both 2002 and 2003, Nomar ranked near the bottom of the pack in PMR for shortstops on the field for 1000 balls in play. In 2002, he ranked 31 out of 36. In 2003, he ranked 28 of 38. Was his range costing the Red Sox outs? Yes.
Nomar’s offense, however, was making up for his defense. He did earn 52 win shares over 2002-2003. So, with Nomar’s offense fine, did Theo really believe that Nomar’s defense was costing them that much? Remember, the difference between a really great defensive shortstop and a really bad defensive shortstop over a full full season is 2 or 3 wins. And while Cabrera was good, he wasn’t great. So you’re talking about maybe 1 win defensively with Cabrera playing instead of Garciaparra. That doesn’t seem to me to justify a trade on defense, especially when it’s not a long term solution.
As it turns out, Nomar earned 1.7 defensive win shares with the Cubs, and Cabrera earned 1.7 with the Red Sox. Overall, Nomar had 6 win shares with the Cubs, Cabrera 5 with the Red Sox. The tangible evidence says the Red Sox would have done about the same with Nomar or Cabrera at short. The tangible evidence says defense wasn’t that important. The tangible evidence says the Red Sox drew the wrong conclusion from 1 1/2 months of fielding data.
The intangible evidence says it was a great trade. Theo trades, runs allowed per game go down, runs scored per game go up and the Red Sox win the World Series. And every once in a while, Soriano swings at a low outside pitch and hits a home run. It’s about process. Maybe this trade was about changing the personality of the team. That’s fine, but I’d like to be told that rather than some fluff about defense. Maybe it was just that the Red Sox didn’t want Nomar long term and tried to get what they could for him. There’s nothing wrong with any of that, but they would have been rid of Garciaparra by the end of the year anyway.
Here’s what I believe. Theo didn’t go to ownership and say, “We have to trade Nomar because he’s killing us defensively.” I believe ownership came to Theo and said, “Get what you can for Nomar, and find a way to justify it.” And yes, that’s pure speculation. But I know Theo is a very smart guy and knows about sample sizes. I know he has a very good handle on the value of defense vs. offense. And knowing that, the explanation for the deal does not make sense to me.

Here’s the table for major league second basemen in 2004. Again, fielders are included if they were on the field for 1000 balls in play.

Like the shortstops it wasn’t a good fielding season for the second basemen overall. This table does give some credence to the idea that Jeff Kent is a better fielder than conventional wisdom says. I hope someday to improve this program to a point where it’s similar to whatever DePodesta uses.

This list should also make Phillies fans happy. They appear to have two of best in Utley and Polanco. And while Nick Green didn’t add much to the Atlanta offense, he ate up balls at 2nd last season.

At the other end of the scale, the Yankees look like they actually upgraded their range at second replacing Cairo with Womack. And if defense is so important to the Red Sox, I wonder how long Mark Bellhorn will last at 2nd.

It’s time to start looking at individual players. We’ll start with the position to get the most opportunities, the shortstops. As the following table shows, it wasn’t a great season for these middle infielders.

One hypothesis for the overall poor play by shortstops in 2004 is the aging of the big players. Vizquel, Jeter, Garciaparra and Tejada are not youngsters anymore. A-Rod moving out of the position hurt also. All of these players will be a year older in 2005; it will be interesting to see if there is further decline in the position as a whole.

It looks like the Nationals got a decent vacuum cleaner at short with their signing of Christian Guzman. With all the talk about Rich Aurilia being old and broken down, he did a very good job fielding. It also appears that the Angels got a nice upgrade replacing Eckstein with Cabrera. If Eck fields like that for the Cardinals, don’t expect that team to be number one in defense again next season.

Pokey Reese, who was supposed to spend most of his time at 2nd base before the Nomar Garciaparra injury, had the best range at shortstop in the majors in 2004. Nomar was down near the bottom. This gives us a chance to evaluate the Red Sox shortstops.

So if we go back to the Garciaparra/Cabrera trade, we can now see it in it’s full light. It wasn’t that the Red Sox defense had been bad all year; it’s that it was bad with Nomar at shortstop. With Reese injured, Boston figured they needed another fielder at the position. However, Boston may have jumped the gun. There’s some evidence that Nomar was just rusty. Compare Nomar with Cabrera after the trade:

So after the trade, Garciaparra had better range than Cabrera. Yes, Cabrera was able to play more. The uncertainty of Nomar’s future health was certainly a factor in the deal. But given Nomar’s play the rest of the way, Boston could have done without the trade and been just as good on defense, with Crespo or Reese (once he got off the DL) spelling Nomar occasionally. I felt at the time that defense was an excuse to move a player the Red Sox no longer wanted. This data does nothing to change my mind on the matter.

I’ve been working on the software for this during the week, and have acquired updated ball in play data as well. I’m now ready to go full bore with the study.

First, however, an update of a couple of tables shown previously. The good people at Baseball Info Solutions have been busy recording batted ball information this winter to complete the database, and that new data is included in the following table. This should replace the table found here.

The order of team changes a bit, but not much. It still looks like a poor defensive season overall.

The other chart to update had to do with performance behind pitchers, and that’s updated in the extended entry.

Continue reading →

Update: I have improved data for the models, so I’ve updated the tables in a new post. They are in the extended entry. The order changes a little, but not enough to make a big difference.

Something easy for me to do with the software I’m developing is to look at the defense behind particular pitchers with the Probabilistic Model of Range. The following chart lists every pitcher with at least 300 balls in play against him for a particular team.

The thing that struck me when I looked at this table was Curt Schilling at the top and Derek Lowe very close to the bottom. On the same team, with pretty much the same defense, Schilling received 28 more outs that expected (that’s a whole nine innings worth of outs) and Lowe missed almost as many, -25. So what’s going on? Here’s a closer look at the pitchers on the Red Sox.

If there’s a pattern here, I’m not sure what it is, except that Wakefield and Lowe had the lowest K per 9 in this group. What we really could be seeing is how pitchers effect the balls in play. It could simply be that the balls put into play against Schilling are easier to field than the balls put into play against Lowe! Voros McCracken’s theory is that a pitcher has little to do with a ball in play being turned into an out. Tom Tippet found that that’s not really the case, although the effect by the pitcher is small. (Links to both found here.) Maybe we’ve found a way to quantify that contribution.

Of course, it could be all luck. Lowe had a very positive number last year. Schilling was on the plus side, but only by about 6 outs in 2003. Looks like a whole new line of study is opening.