Archive for the ‘Sabermetrics’ Category

No Gold Glove for Kevin Kouzmanoff

Did Kevin Kouzmanoff deserve the Gold Glove award for best NL fielding third basemen of 2009? That’s the only question I really care about. As a Padres fan, I certainly root for the Padres and their players, but when it comes to things like player analysis, I try to be as objective as possible, setting aside my Padre fandom in a search (usually, a futile one) for the truth.

The basics

Anyway, let’s compare Kouzmanoff with Ryan Zimmerman, the Washington Nationals third basemen who won the Gold Glove award. First, some basic numbers courtesy of Baseball Reference:

Player Innings Chances PO Assists E DP FP RF/9
Kouzmanoff 1186.7 311 94 214 3 24 .990 2.34
Zimmerman 1337.7 459 117 325 17 28 .963 2.97

 

Two things stand out. Kouzmanoff made a lot less errors than Zimmerman, but Zimmerman made a lot more plays. The difference in errors is important for one thing that is often overlooked: Kouzmanoff gave up very few extra bases on wild throws, while Zimmerman presumably did. However, as far as turning batted balls into outs, despite the error difference, it appears that Zimmerman is far superior, as evidenced by his 2.97 Range Factor (assists + put outs/innings *9) against Kouz’s 2.34.

Context counts

The difficult part of fielding analysis, however, is context. That is, determining truly how many opportunities each player had to make outs, because they were playing at different parks, behind different pitchers, and against different hitters. They certainly did not face the same exact set of batted balls, in such a different environment, so looking at their fielding stats, without adjustments, may serve us no good.

So, what do we want to do? We want to determine how many opportunities each player had. The Hardball Times’ RZR can help us here. Baseball Info Solutions watches video of all batted balls and determines whether or not they were in the third base zone. All other grounders are classified as out-of-zone. Here are the RZR numbers:

Player BIZ Plays RZR OOZ
Kouzmanoff 223 168 .753 34
Zimmerman 282 205 .727 101

 

BIZ—Balls in zone
Plays – Plays made on BIZ
RZR – plays/BIZ
OOZ – plays made out of zone

What we see here is that, yes, Zimmerman did have more in-zone opportunities, in part because he played more innings, and probably in part because of some differences in context (such as the Nationals and Padres respective pitching staff tendencies). We also see that Kouzmanoff made a higher percentage of plays on in-zone chances, in large part because he was so good at not making errors.

However, Zimmerman has a huge lead in out-of-zone plays, 101 to 34. Now, he also had more BIZ, so we’d assume he had more OOZ opportunities, but that doesn’t come close to making up the difference.

The more precise, the better

One of the problems with an RZR-type system is that it throws all in-zone balls into one bucket, and all out-of-zone balls into another. What if, by chance, Zimmerman had a lot of balls hit right at him, and a lot hit just outside of his zone? That would probably skew our analysis. So, we can look at something like UZR, which is more precise, cutting the third basemen’s area of responsibility into many small slices, rather than one big one (it also has adjustments for park and pitching staff and so on).

Simply put, it’s better. Last year, Zimmerman’s UZR per 150 games was 20 runs above average, a very high number. Kouzmanoff was at +10.7 RAA per 150. Obviously, a good rating, but Zimmerman clearly has the edge.

More data, please (redux)

One thing about advanced fielding stats is that we really shouldn’t look at just one year. There’s too much variability, year-to-year, to put too much reliance into a season of data. Looking at more than one year gives us a better idea about a player’s true fielding ability. Kouzmanoff, in his 418 career games at third, is about +3 RAA (per 150). Zimmerman’s RAA/150, in his 590 game career, sits at +12. Again, a relatively clear advantage to Zimm.

Scouting and the wisdom of the crowds

In fielding analysis, scouting is undoubtedly pretty valuable. I don’t have scouting reports available on either player, but just about everything I’ve ever read about Zimmerman’s fielding, from non-stathead types, has said that he’s great with the glove. Kouzmanoff, on the other hand, is more of a mixed bag, rarely ever receiving too much praise for his glove-work.

Tom Tango runs his Fans’ Scouting Report every year, in which fans rate the fielding prowess of the players they watch everyday. This year Ryan Zimmerman ranked 6th among third basemen in average rating, while Kouzmanoff came in at 13. In 2008, Zimmerman was 4th, Kouzmanoff 29th.

The Fielding Bible asks a panel of ten experts, including the likes of Bill James, James Dewan, and the BIS video scouts, to rate fielders at each position every year. Zimmerman came in first place here, while Kouz came in 11th overall. Last year: Zimmerman 9th, Kouzmanoff 12th.

A little common sense

After going through all of this, I don’t see any reason to suspect that Kouzmanoff was truly a better fielder than Zimmerman in 2009. Zimmerman’s advanced numbers were clearly superior, he has a better track record defensively, and he’s better from a scout’s perspective. Kouzmanoff had a fine year with the glove, and he’s come a long way in the fielding department in the last few years, but that doesn’t make him better than Zimmerman. A low error total doesn’t either, because we’re not looking for fielders who don’t make errors. We’re looking for fielders who make outs.

There is always that chance that he was better, though. We can never really tell for sure. The numbers could have been off five or ten runs for each player, in either direction. But, considering what we do know about both of these guys, I think it is pretty clear that Zimmerman deserved the award over Kouzmanoff. If anything, it is far from an outrageous decision.

More data, please

With the offseason upon us, there is going to be a lot of player-valuation analysis for potential free agent signings, trades, and other roster moves.

It’s a good time to mention that more data is virtually always better. I saw some chatter today where Chone Figgins was being evaluated as a 6+ WAR player. Figgins did post 6.1 WAR this year, according to FanGraphs. However, I don’t think there’s any reason to believe he is a 6 WAR player.

Prior to this year, Figgins has been worth about 2.8 WAR per 729 PAs (the number he received this year). His last four years look like this: 2.7, 0, 3.1, 2.4. Figgins is a good player, no doubt, but valuing him at 6 wins could be a major mistake. There are multiple reasons to suggest Figgins is not at that level.

1. His past performance, as mentioned above, is a good indication that this year was a bit of a “fluke.” A great season by a good player playing over his head.

2. Regression – It’s a good idea to assume that every time a player has a great season, he’s going to come back to earth a bit next year (points one and two are a bit interrelated). Consider a hypothetical player, who put up five straight 6 WAR seasons. Still, even in his case, we’d project the player at, maybe, 5 or 5.5 WAR, depending on the exact situation.

3. His age – Figgins will be 31 next year and his game revolves around, in many ways, his speed. There is a good chance that asset will be on the decline through the rest of his career.

None of this is to say that Chone Figgins can’t become a 5-6 WAR player, year in year out. He definitely could. But if we’re just looking at the numbers, and we have no special reasons to believe he’ll overcome the odds, then we’re better off taking the safer, more logical bet: That he will revert back to his old self, to some degree,  and be a ~3-4 WAR player going forward.

Jake Peavy trade analysis

At the trade dealine, Jake Peavy was dealt to the White Sox for four pitching prospects. Peavy, who has been injured since June, may not pitch again this season until late August, at the earliest. Anyway, before we look at the details of this trade, I think it’s essential to touch on Peavy’s career in San Diego.

The guy was a great asset for the Padres, and a wonderful player to watch. Peavy finished his Padres’ career with a 3.29 ERA in 1342 innings, racking up 1348 strikeouts to just 435 walks. According to FanGraphs, Peavy was 25.6 Wins Above Replacement level from 2002-2009. That performance, on the free agent market, was worth about $95 million. The Padres paid a mere $21 million for it. His time here wasn’t perfect, as he came up short in some of his biggest starts, but overall Peavy was tremendous for the Padres.

At this time, however, it does make some sense to deal Jake. Peavy makes $48 million over the next three years. The Padres are a team not expected to compete for at least one of those years, if not all of them. Peavy is getting older. It adds up. Other than trading him when his value is relatively low, its probably a good time to trade Peavy.

That said, was it a good deal? First, we need to estimate Peavy’s production and value over the rest of his contract. ZiPS projects Peavy’s park-neutral ERA at 3.65 – note that we’re ignoring his injury for now and just going by his past production. So, I’ve got Peavy as about a 4 WAR pitcher, in a full season. Here’s the chart:

Year 2009 (1month) 2010 2011 2012 Total
WAR .7 WAR 3.5 WAR 3 WAR 2.5 WAR 9.7 WAR
FA Salary $3.4M $18.5M $17M $16M $55M
Actual Salary $2.6M $15M $16M $17M $50.5M
Surplus Value $.8M $3.5M $1M -1M $4.5M

 

Peavy is ‘only’ worth $4.5 million in surplus value through 2012, according to my calculations. That is not considering his option for 2013 (and $4M buyout) or the draft pick the Sox will pick up when/if he leaves for free agency. There are two factors making that figure look smaller than you might have expected. One, Peavy is getting older, he likely isn’t going to be as good now as he has been in the past. Two, now he starts making the big bucks. Again, $48 million over the next three years. That’s not bad for a sure-fire ace. But for Peavy, coming off injuries and approaching 30 – well, it’s not bad, but it isn’t necessarily a bargain.

According to John Sickels, the Padres picked up:

Aaron Poreda: B+ prospect
Clayton Richard: C+ prospect
Dexter Carter: C prospect
Adam Russell: unranked

Those four prospects are worth about $11 million in surplus value. So, the Padres actually made out relatively well in this deal. $11 million – $5 million = $6 million to the plus side. That is, if you trust my evaluation of Peavy, and the judgment of these prospects, and all of the other assumptions made here. There are a lot of factors involved in this type of analysis, making it much more of a fun little exercise than it is a true evaluation of the trade.

Also, even if this is a “win” for the Padres, it doesn’t necessarily mean it is a loss for Chicago. The White Sox could increase their playoff chances enough to offset the apparent gap in the deal that favors San Diego.

Anyway, the Padres got great production out of Jake Peavy while he was making very little money, relatively speaking. Now, when he’s going to get paid (and should gradually decline performance-wise), they ship him off for a haul of decent, if not overwhelming prospects. Yeah, that may not be how you want to look at things from a fan’s perspective, but I guess it is a business. For a small market team like the Padres to compete, moves like this may be a necessity. They can afford players up to a certain point, but once that point comes it is time to cash in and move on.

It was fun watchin’ ya while you were here, Jake.

Mat Latos strong in second start

Mat Latos was solid again in his second major league start, as he went 5.7 innings while allowing five hits and two runs. He struck out five and walked two, and allowed two solo homers. Latos threw 94 pitches, 58 of them for strikes. Let’s again check out the PITCHf/x data and see what it tells us about Latos’ outing. Here’s a look at his pitch selection, along with average velocity:

Fastball: 61 pitches (94.2 mph)
Slider: 25 (81.5)
Changeup: 8 (81.3)

Here’s his pitch movement graph:

latos start 2 movement

And here’s velocity:

latos start 2 velocity

Latos mixed in the off-speed stuff a little more often in start two, as he threw 65% fastballs (his first start was around 80%). When looking at any of this data, I wouldn’t get too caught up in any slight differences start to start. For instance, according to the data, his average fastball velocity dropped about one mph from start one to start two. While that could be the case, it’s also very possible that there are park effects influencing things or a pitch or two got misidentified.

Anyway, the main points remain: Mat Latos has great velocity, and appears to have decent secondary offerings. There isn’t much more we could expect from his first two starts. Latos is quickly becoming one of the few bright spots in this dismal 2009 season, and he’s one reason to keep following these Pads in an otherwise lost year.

Quick glance at Latos through PITCHf/x

Top Padres prospect Mat Latos made his Major League debut Sunday, striking out four and walking just one in four innings of work. I didn’t see a pitch, so I figured I’d take a quick look at his PITCHf/x data (though others have already done an excellent job at that). Here’s his movement graph:

latos

You can see the fastballs clustered up top (with a change up or two), and the sliders and curves toward the middle of the graph. If anything, this graph shows just how much Latos relied on that fastball (about 80% of his pitched recorded by PITCHf/x).

Here’s a peek at his velocity, pitch-by-pitch:

latos2

Latos was relentless with the fastball early, throwing it almost exclusively (though the camera missed a few early pitches). The most impressive thing here is the velocity – he averaged about 95 with the fastball, and almost hit 98 a few times. You can see that his fastball velocity gradually decreased as the game went on, as expected, but his last two pitches still reached 95.

We can talk about movement and pitch selection and mechanics all day long – and I’m sure we will – but when a guy can throw 95+ as a starter, and has a decent idea of what he’s doing, well, there’s reason to believe the hype. One start certainly means very little, but it does show us that this guy can really bring it. That is a good thing.

The future

Tango gives us the links.

Seriously, this video stuff is awesome. The New York Times has an article on a new system that’s being tested by Sportvision that will track the movement of the ball, fielders, and base runners. You might imagine how valuable that information will be.

I’m looking forward to all the questions and answers that will come from all of this new technology.

First base fielding, THT style

We move onto first base in this slow moving series. Thanks as always to The Hardball Times for the data.

Top 5

Player Team RAA
Pujols STL +11
Konerko CHA +10
Kotchman ATL +9
Youkilis BOS +7
Garko CLE +7

 

Bottom 5

Player Team RAA
Fielder MIL -11
Helton COL -11
Hernandez CIN -10
Giambi OAK -7
Davis TEX -7

 

Adrian Gonzalez’s fielding

By the numbers, Gonzalez never seems to rate well. This year he’s average by this stat. On balls in his zone, he’s made 92 plays on 114 chances (3 plays better than average). On balls outside of his zone, however, he’s made just 23 plays (-4 plays). One half year of data doesn’t tell us much, though. Here’s his UZR (BIS data) from 2006 through 2009:

Year UZR
2006 +7
2007 -7
2008 -3
2009 +3

 

Overall, like UZR, most metrics see him around average. Who is right? The fans who see him as one of the best first basemen in the game or the numbers which peg him right around average? I really don’t know. It’s important to note that these stats don’t consider picking errant throws and some don’t consider double plays (UZR does, however – the THT-based numbers above do not). We may be safest taking a middle ground, and assuming Adrian is a +5 fielder, while fully acknowledging that he really may be average (or a bit below) or +10. There’s a pretty lot of uncertainty here.

Third base fielding, THT style

Here are the top 5/bottom 5 third basemen, according to the data at THT:

Top 5

Player Team RAA
Hannahan OAK +15
Zimmerman WAS 15
Rolen TOR 14
Crede CHA 12
Beltre SEA 12

 

Some familiar names there like Zimmerman and Beltre. Beltre’s loss will be a big one for Seattle (I hear they might be interested in Kouzmanoff…). I had no idea Hannahan was this good. UZR has him at +14/150 so far for his career.

Bottom 5

Player Team RAA
Lowell BOS -15
Young TEX -14
Feliz PHI -12
Fields CHA -10
Peralta CLE -8

 

Some interesting names here, including Lowell and Feliz, who are generally regarded as great defenders, and Peralta and Young, who have both recently switched over from shortstop.

Technical note: I’m now reporting all fielding numbers to the nearest whole number to not give the impression that they are accurate to the decimal. I could put them into buckets (-5 – +5, +5 – +15) or something like that to further emphasize the uncertainty, but I’ll leave that much up to you.

Kouzmanoff’s fielding

Kevin Kouzmanoff has received some praise this year for his great fielding percentage (.989), which is second in MLB (behind only Geoff Blum – minimum 300 innings). This is of course an accomplishment, but the goal of fielding is not to avoid making errors. It is to make as many plays as possible given your opportunities.

Kouzmanoff has made 86 of 123 plays in his zone this year. That’s about 70% – major league average is about 72%. He’s about two plays below average here. He’s made 20 out of zone plays. Given his opportunities, we’d expect him to make about 22 OOZ plays. So he’s also –2 here. That’s a total of –4 plays, or about –3 runs.

UZR has him at –1 run this year, and –1 for his entire career. This year (by UZR), he’s +3 on error runs, and –4 on range runs. It should be noted that Kouz’s lack of errors does result in less extra bases by the opposing team (on overthrows and things like that). UZR does take that into account, while other metrics don’t.

However, that has a relatively small impact on his overall fielding ability. He’s been an average fielding for his entire career. Hopefully his low error totals won’t trick anybody* into thinking that Kouzmanoff is a gold glove caliber third basemen.

*unless they are interested in trading for him : )

WPA Inquirer

Dave Studeman introduces a new WPA tool at The Hardball Times. Very cool.

It gives you the win probability for any situation in a game, and you can also adjust for run environment. The important thing to remember is that the numbers that get spit out are overall averages; you have to keep in mind the specific situation (pitcher, hitter, on deck hitter, base runners, etc.). Also, with any sac bunt analysis, you have to remember that many sac bunt attempts turn into hits, errors, double plays, etc, which changes the numbers.

Anyway, fun stuff. I’m sure we’ll be using it in the future.

Stolen base research

Along with misinterpreting a stat in my last post, I also somehow managed to wipe out another short post I had made on the same topic. Good grief! I guess this truly means I’m back ; )

Anyway, before looking into the Pads’ 2009 base running stats, I discovered a few recent base stealing articles on the ‘net. One, Brian Cartwright’s latest in the BP Idol contest, is kind of a running (retrosheet) history of the stolen base. Gives a couple nice charts at the end that show how stolen base percentage has been on the rise, while stolen base attempts have been in decline.

The second, by Dan Turkenkopf at THT, gives us a couple of alternative ways to measure the value added from base stealing (including Net Stolen Bases, developed by Rich Lederer and slightly altered by Dan, which I used in my last post).

Good stuff.

Follow

Get every new post delivered to your Inbox.