Sunday, November 26, 2006

The Defense of Sarge Jr.

I've tracked down all the defensive data I can find on our new center fielder. Mitchel Lichtman (MGL) has posted his last 3 years of UZR data on BTF. The Defensive Bible has his plus/minus ratings for 2003-2005, but not 2006. For 2006, there are leaderboards with the top 10 at each position. Matthews did not make the top 10 in 2006, so the only thing I can say about him is that he's less than +9 for the most recent year. David Pinto's PMR, posted on and translated into runs by the halosphere's own Blackhawk of Waterloo, is available for the last 2 years. Matthews did not play enough CF to qualify for the list in 2004. Finally, I've got my own zone rating based data, which last year was very close to Chris Dial's numbers from the same source. They aren't as similar any more, as I have incorporated park factors for outfielders, which you can see in the old posts. I will have to publish my final park factors sometime, what I'm using differs slightly from the last batch I published as I've increased the PF for the green monster, averaging my multiyear data with more accurate (but only one year) data courtesy of Joe Arthur. In addition, I am only using the park factors that rate as significant at the 95% confidence level. Texas is one of those, its a tougher field than most OF, and my guess is that the ball jumps off the bat there better than most parks, resulting in tougher to field chances, and also the hitter's environment we know as the Ballpark in Arlington.

Anyway, what do these methods say? All numbers in actual playing time, not per 150 or 162 games.

UZR: 2006 +3, 2005 +15, 2004 +7
ZR_pk: 2006: -2, 2005: +5, 2004: +1, 2003: 0
Dewan: 2005: +2, 2004: +4, 2003: +3
PMR: 2006: -1, 2005: +10

I can plug all those numbers into my projection formula, which weights 2006 as 1, 2005 as .8, 2004 as .6, and 2003 as .4. Results are regressed at 50% for 300 chances. Finally, I subtract 1 from the projection result because he's 32. And I get:

UZR: +8
ZR_pk: 0
Dewan, PMR: +3

So we have, at worst an average defensive outfielder, and likely a good one, as UZR is probably the most accurate of the play by play defensive metrics.

Then there's the scouting report of the fans, from Tangotiger's website. I looked at the "range" components that the fans track. Sarge Jr grades as:

Instincts: 78
1st step: 84
Speed: 81
Hands: 82

Without doing a detailed comparison, these all look well above average for a center fielder. But for 2005, its not so good.

Instincts: 60
1st step: 69
Speed: 72
Hands: 53

Again, I haven't tallied up all the centerfielders, but this looks to be the profile of an average centerfielder (though above average compared to all outfielders).

I'm not sure I buy that he was faster at 31 than he was at 30, especially when UZR had better numbers for him in 2005 than in 2006. Perhaps his 1st step and instincts really did improve, as 2006 was the first year he had the center field job all to himself, instead of being used at multiple positions or in a 4th OF role.

My CHONE projection has him hitting .268/.338/.423. The total batting line is 2 runs below average. If anything, it seems on the optimistic side. I've got him hitting for about as much power , with similar K/W numbers as he did last year. The only difference is that he's not a real .313 hitter. His previous career high was .275, and career average is .263. I've got him at .268, so if anyone wants to tell me he's really an above average hitter, using prior years linear weights ain't gonna cut it. I want to know how he does it. Produce his projected batting line, and then we'll see. My projected batting line is out there.

Baserunning, he's essentially average, as are most players. I give CF a +2 position adjustment. I've seen higher, but +2 seems right to me, right about even with 2b & 3b, less than SS or C, more than 1B and corner OF.

With +8 on defense he's +8 against average, and +25 vs replacement level. At 4 mil per win, that's actually 10 million. He's about 1 win above Figgins in center, all due to improved defense. Even if he's closer to average on defense, he's a slight upgrade, as Figgins can be an adventure out there.

One more win last year and game 162 would have meant something, and I can assure you Shields and K-Rod would have been out there as that game vs Oakland went into extra innings.

I don't hate this deal anymore. I still think its a little too long given his age, but welcome to LAnaheim, Sarge Jr.

Thursday, November 16, 2006

The CHONEs, 2007

Here are the most deadly accurate (not) projections known to mankind, the CHONE (Comprehensive Holistic Objective Numerical Estimations) projections.

The shorthand was named in honor of our favorite jack of all trades, Mr. Figgins. The long version of the name was suggested by the Baseballthinkfactory poster known as Dag Nabbit.

This is the first complete release of the data. All players are park adjusted and projected to a mythical stadium that is completely average, and somewhere between the American and National leagues, just off the interleague play highway. I have left out runs and RBI, as that requires a separate step where I try and predict what spot of the lineup the player will primarily hit in, plus information about his team.

I have included baserunning and fielding projections, a position adjustment, and a figure for total runs over replacement level. Players who didn't play too much, or mostly in the minors, may not have a defensive or baserunning projection. These players will have a zero in that field. I have projected almsot everyone who played major league baseball in 2006. In the future I will have CHONEs for pitchers and selected minor leaguers.


Update: Selected minor league players added.

Saturday, November 11, 2006


This week I got the Bill James Baseball Handbook 2007. At this point, having player career stats isn't that big a deal, since you can find them so many places online, with and Retrosheet being my favorites. But Bill's book has some goodies that you can't find anywhere else.

One is leaderboards for average pitcher velocity. I'm happy to find that they include relievers this year, in the past they only had pitchers with 162 innings. They left Daniel Cabrera in the 2006 book because he just missed the innings, even though he had more 100 mph+ pitches than anybody.

K-Rod averaged 94.8 with his fastball, good for 8th among AL relievers. Among the starters, Escobar (93.1) was 6th, and Magic Santana was 8th (also 93.1). Felix Hernandez led starters at 95.2, and Verlander and Beckett were also close to 95.

In the (inter)National league, Brad Penny led with 93.9. Only he and Matt Cain would have cracked the top 9 in the American League. Zumaya (98.6) is well ahead of the top NL guy, Billy Wagner (96.3), further evidence that you need to use MLE's when looking at National League stats.

Then there's baserunning. First to third, second to home, 1st to home, plus miscellaneous bases (advancing on wp, balk, sac fly, etc.). Figgins gained 4 runs on the bases, (not including steals), the most of any player.

They had baserunning numbers last year, so I can use them to make a projection. The correlation between 2005 and 2006 isn't that great, so I've got to apply a good bit of regression. Almost all players project to within -2 and +2 runs for 2007. 7 players get the top rating, +2, including Figgins and the OC. Pat Burrell has the worst rating, -3, and seven others get a -2, including our old friend Bengie Molina.