March 13, 2009
It might be a stretch to say that "defense" is John Dewan's middle name, but then again it easily could be. The author of the highly acclaimed The Fielding Bible has delivered an even more impressive second volume, making Dewan the industry's most influential voice when it comes to defensive metrics. A co-owner of Baseball Info Solutions, Dewan moderated the Baseball Analytics panel at last weekend's MIT Sloan Sports Analytics conference. Afterwards, he sat down with Baseball Prospectus to talk about why Carlos Gomez is a better defensive outfielder than Nate McLouth, why shortstops love Justin Morneau, and what it means to be a Molina.
John Dewan: Well, it has certainly been my frontier. I've been working on defense since I started in this business in the mid 1980s. When I first had a chance to work with Project Scoresheet data, I focused on defense. When I had a chance to work with STATS Incorporated data, I focused on defense. When I started working with Baseball Info Solutions data, I worked with defense and did The Fielding Bible, and now The Fielding Bible II. And I guess that the reason I focus on it and like it so much is that I see importance to it and it is hard to measure. And it hasn't been measured. So it presents a challenge, and I think that there are some very interesting results.
DL: Can you give a good example?
JD: I'll give you an example which came out in my book: When we did our analysis of defense by team, we took our various metrics and broke them into Defensive Runs Saved. What we found is that the best defensive team in baseball last year was the World Champion Philadelphia Phillies-their defense saved 78 runs over the course of the year compared to the average team in baseball. The worst team in baseball was the Kansas City Royals, who allowed 48 more runs than the average team. If you put those two together, you get about 130 runs difference between the best and worst defensive teams in baseball. If you look at that same analysis on hitting, you'll see that the Rangers scored about 900 runs, the Padres about 640, so the difference is about 260 runs. That is exactly twice what the defensive impact was, which is almost like saying that defense is exactly half as important as offense, which no one would ever give that kind of credit to. And I'm not sure I believe it entirely myself, but it is one analysis. There are other analyses to do, but I think that what we're getting to is realizing the magnitude of the importance of defense. We're starting to move in that direction, and it's not just me. There are all kinds of people doing defensive analysis, whether it's David Pinto with his Probabilistic Model, or if it's Mitchel Lichtman with UZR, or the folks at Baseball Prospectus who are doing some new stuff this year. There are also the guys at The Hardball Times. So a lot of people are doing work on defense, and I think it is enlightening everybody in the field.
DL: Is there any one position that causes you to bang your head against the wall, wishing that you could do a better job of measuring of it?
JD: I did knock my head against the wall trying to evaluate catching. One thing that is interesting about catching is that the throwing arm is overrated. What a catcher is capable of doing is so dependent on the pitcher. In my own mind, I think of the pitcher as being responsible for 65 percent of whether a runner steals, while the catcher is responsible for the other 35. And the dichotomy comes out in the Gold Glove voting. When Gold Glove voters vote for pitchers, they never ever consider how well they hold runners. But when they vote for catchers, it is almost the only thing that is considered, how well they hold runners. The irony of it is that pitchers are more responsible for holding runners than are catchers.
With catchers, what we tried to do this year was put some kind of value on their ability to call a good game. How much do they affect a pitcher's ERA? Are certain catchers able to do a better job calling a game than other catchers? Of course, the statistic Catcher's ERA has been around for quite awhile, but what we've done is try to quantify it a little bit better. We've turned Catcher ERA into Earned Runs Saved for the catcher, and we found a comparison that was really pretty incredible. It involved Pudge Rodriguez, who moved from the Tigers to the Yankees. If you compare Brandon Inge catching for the Tigers, as opposed to Pudge Rodriguez catching for the Tigers, you find that every single pitcher that Pudge caught was better when he caught them than when Inge caught them, down to a 30-inning difference. Then you move over to the Yankees and look at Jose Molina. Molina caught every Yankees pitcher better than any other Yankees catcher, including Pudge Rodriguez. In the book, we have an article called, "Molina Is To Pudge As Pudge Is To Inge." It's a comparison of the three, and I think you can easily say that, based on the numbers, Molina was better than Pudge defensively last year, and Pudge was better than Inge last year. I think it's a pretty clear-cut analysis. Nevertheless, the numbers can get real squirrelly. The numbers can jump. There are so many factors at play. Andy Pettitte threw a game with Pudge Rodriguez catching and allowed something like 11 hits and nine runs-I forget the exact numbers. But it was an incredibly poor performance, and Pettitte said, "I want Molina catching me," and Molina caught him for the rest of the year. Well, how much of that game was because Pettitte just didn't have it that day? And how much of it was that Pudge Rodriguez didn't call for the pitches in the way that he would have liked? Well, it could easily have been that Pettitte just didn't have it that day. So there is a lot of noise in the data. We're trying to filter that noise out as best we can, so I built in a credibility factor. We've come up with Runs Saved on the part of catchers based on their handling of pitchers. It's not based on the true, complete Earned Run Numbers Saved I just referred to; it's based on a credibility weighting where I don't give as much credit, because there is so much noise in the data.
DL: How much variance have you seen in performance year to year? For instance, Jorge Posada's level of defensive play last season was likely negatively impacted by his injury.
JD: That is the exact thing we looked at. Was there some consistency? Could we see players who had some consistency? Just like any other statistic, players have good years and bad years, so are we getting good years and bad years randomly mixed, or do we seem to be having a trend where the catchers who are generally better, are really better? And we thought we were getting consistency. We were finding there was a trend for certain players to do a better job calling a game.
DL: Did any of the results surprise you?
JD: Well, what was really cool was the fact that if your name is Molina, you generally do better than average. The three Molina brothers are incredible in terms of how well they do. Now, I was looking at the entire spectrum of catching, which in the case of what we're doing in The Fielding Bible, we're combining handling pitchers with their ability to save stolen bases. So we have Stolen Bases Saved turned into Runs Saved, and we have Adjusted Earned Runs Saved based on their handling. And the Molinas all turned out pretty well.
We did a little historical work, and one player who has always had a horrible defensive reputation is Mike Piazza. And Piazza, sure enough, when you look at his Stolen Bases Saved, adjusting for the pitchers' ability to hold runners-that's something we've done with Stolen Bases Saved-even when you adjust for that, Piazza is horrible. He's just one of the worst all-time at holding runners. But when we applied the model of Earned Runs Saved, we found that he was clearly above average. He was a better-than-average catcher at handling pitchers.
DL: Looking at the book, I was surprised to see how poorly Jason Varitek rated in Earned Runs Saved over the past six seasons.
JD: Yeah, that actually surprised us. Jason Kendall is near the top, Pudge Rodriguez is near the top, and we were a bit surprised by Jason Varitek. But the bottom line is that his Catcher ERAs-and we do an adjustment based on park factors, and in his case it doesn't matter as much, though historically park factor adjustments are important for a guy like Mike Piazza especially. But Jason Varitek just consistently, when you looked at how he compared with other Boston catchers, he didn't do as well. And our system not only compares against Boston catchers, but it will take into account pitchers who pitched for the Red Sox, and how they did when they were traded in midseason and caught by catchers on other teams. There are a variety of factors that come into play, and Varitek did surprise us by not coming out as well as we had expected.
DL: Much like Varitek, Brad Ausmus has long had a reputation as an outstanding defensive catcher. Does Ausmus's defensive value make up for his lack of offensive production?
JD: Well, when you look at the numbers, the answer is clearly no. But the numbers don't measure everything. Some of the problems with the numbers are that we're trying to measure pitcher handling, and we have a little bit of a handle on it. We're trying to measure stolen bases and Runs Saved, and we have a very good handle on that, but there are still so many components of what a catcher does, like working with a pitcher and working with the team, and those are areas where both Ausmus and Varitek are team leaders. Both have aspects to what they do that we don't know how to measure yet. So if I were a push-button manager, neither of those guys would play for me. But the managers who are managing those guys know a little bit more about them, and what other aspects are important. What I would tell [Terry Francona] is that it doesn't look like [Varitek] is helping his pitchers' ERAs. If that's what you think is happening here, it's not.
DL: Holding baserunners aside, just how much do pitchers help themselves defensively?
JD: Very much. Greg Maddux has won, what is it, 18 Gold Gloves? And based on our numbers, aside from the fact that he can't hold runners to save his own life, he is the best defensive fielding pitcher in baseball. He has saved more runs with his agility around the mound, getting balls that other pitchers don't get-it's just incredible how well he's done. But he's also lost a lot of that because, to my mind, holding runners is part of a pitcher's defense. So I don't think he deserved all of those Gold Gloves, though when I evaluated the players in 2008, I did feel that Maddux should have won it. His problem holding runners was minimal compared to his tremendous coverage around the mound.
DL: Along with Maddux, who are the best-fielding pitchers in the game, particularly those who typically don't win Gold Gloves?
JD: Last year was a travesty, because Kenny Rogers was so far and above everybody else in terms of his defense on the mound that it wasn't even funny. What we're getting is a little bit of what happens with position players: if you don't hit, you aren't going to win. With Rogers, I think it was clearly a case of, if you don't pitch well, you're not going to win a Gold Glove. The defense gets lost if the overall product isn't there, and that's what happened to Kenny Rogers.
The best defensive pitcher in baseball right now, given that Rogers and Maddux both just retired-although I wouldn't put Maddux among the best when you consider holding runners-is Mark Buehrle. Mark Buehrle is the best pitcher in baseball over the last six years in terms of holding baserunners, and he also covers his position very, very well. He has the most Runs Saved of any pitcher other than Kenny Rogers.
DL: Is defense at first base underrated?
JD: I think it is. It's still the least important defensive position, but what happens is... I'll give you one example. Bill James invented a system that we implemented at Baseball Info Solutions called Misplays/Good Plays. We introduced the concept in the book to keep track of 55 different types of misplays and 27 types of good plays. One of the types of good plays is the first baseman handling a difficult throw from an infielder, basically taking a bad throw and converting it into an out. The leader last year was Justin Morneau; 44 times-and just think about that, 44 times-he took a bad throw from an infielder and turned that into an out. He led all of baseball. And here is something even more remarkable: Todd Helton played just a little over half the innings that Justin Morneau played, and he had 42 saves of throws from his infielders. So, there is a skill that Todd Helton has, even compared to Justin Morneau, that if you took it over the course of a season, there might have been 20 or 30 more plays that he made, saving throws from his infielders, than the second-best guy in baseball. These are plays where a guy would have been on first, whether they ruled it an error or maybe given the guy a base hit. Getting those outs is of significant value for a first baseman. And if the second-best guy is that far from the best guy, you realize that some of the average guys aren't making 44 or 42. They're making closer to 15 or 20.
DL: As Helton is essentially saving his infielders errors, are his teammates' defensive ratings impacted by his play? For instance, Troy Tulowitzki might have six or eight more errors on poor throws in a given season, if the Rockies had a lesser first baseman.
JD: I've never really thought of that, and that's a very interesting observation. The shortstop is getting credit for having made that play, but he made it thanks to a good play from the first baseman; he's getting credit because he has a good first baseman. So yes, there is a combination of elements there that, in theory, you could separate out.
DL: Can you address the importance of middle infielders having the ability to turn double plays?
JD: What's clear is that there are second basemen and shortstops who are better at turning double plays than others, and the guy that jumps most to mind is Jeff Kent. Year after year he has been one of the worst players at turning double plays. What we've done is try to attach a numeric value to it in terms of Runs Saved. So, what would happen if the double play wasn't turned? What is the run situation and how many runs are you saving by turning that double play? If you look at our three-year data, you'll see that Jeff Kent is right at the bottom of the list in terms of how many runs he's saved, or cost, his teams. He has cost his team, in double-play situations, nine runs over the last three years, which is actually not that much, but we are quantifying it because it means something. The guy who had the most Runs Saved was Mark Grudzielanek, who had 11, so the difference between the best and the worst second baseman, in double plays, is only 20 runs over three years.
JD: The Fielding Bible panel did choose Phillips as the award winner, but I have an article in the book called, "What Makes Chase Utley so good?" In that article, I admitted that I actually voted for Chase Utley. I think that Brandon Phillips is probably a more athletically gifted second baseman, even as gifted as Utley is, but Utley has what I refer to as ultra positioning. Chase Utley moves around the field, depending on the hitter, more than probably any other second baseman in baseball. As a result, he is fielding a higher percentage of balls hit in the second-base/first-base hole by left-handed hitters, and he's getting a higher percentage of balls hit up the middle by right-handed hitters. That made a huge difference, as he saved over 30 runs collectively for the Phillies last year by doing that.
DL: As good as they are, both Utley and Phillips scored low in the double-play rankings; Phillips at minus three, and Utley at minus two. Is that at all meaningful?
JD: Minus three and minus two are just a little below average. Then, looking over the three-year period, Phillips is minus five, and Utley minus one. So, based on this data, Utley is pretty much an average second baseman at turning double plays, while there is some indication that Phillips is a little below average. Still, the difference between minus one and minus five really isn't all that great. Every second baseman has to have the ability to turn a double play, and as we showed here with Jeff Kent, even the worst will come up with nine runs lost over three years. But, to be honest with you, I don't think we're yet measuring double plays as well as we can; a reason some of the numbers don't show to be as diverse is because we're not measuring it well enough. I think we're on our way to measuring it better, but we need to do things like look at the location of the ball when it's hit, and to determine how hard, or easy, it was to start a double play. Right now we're just looking at all situations, and not considering the type of ball, so there's more work to be done in that area.
DL: Any final thoughts?
JD: The one thing I'd bring up that was kind of fun, was the analysis of Nate McLouth and Carlos Gomez; McLouth won a Gold Glove, and Gomez didn't. Carlos Gomez had the most defensive misplays in center field, which is a characteristic of young players that we've found; other young players up there are Delmon Young, B.J. Upton, and his brother, Justin Upton. All of these players have more defensive misplays. But Carlos Gomez covers so much more ground, that it just shows through on the number of runs saved. The difference that we found between Nate McLouth and Carlos Gomez was amazingly straightforward. Simply, Gomez is covering ground in deep center field, where fielding a ball is much more valuable, than Nate McLouth, who covers more ground in shallow center field, where making a catch means that you're saving a single. Gomez, meanwhile, is saving doubles and triples. It looks to be that the biggest problem for Nate McLouth is that he should play deeper. He has good skills and a lot of good fielding plays in our system, but when we break it down between shallow, medium, and deep, which is something we did in the book this year, he's plus on shallow balls, and minus on medium and deep.
To listen in to a segment of John Dewan moderating the Baseball Analytics panel from the MIT Sloan Sports Analytics Conference on March 7, 2009, check out this exclusive Baseball Prospectus Radio podcast:
Click to download mp3