The Importance of Defense
One of the things we’ve been talking around, but have never said outright, really, is how wrong we used to be about the importance of defense. The statistical community, 5-10 years ago, believed that defense was a very minor factor in the game, and that you could evaluate the worth of a player by looking at his offensive line, and then making a minor adjustment for defensive value. Players like Darin Erstad and Rey Ordonez became sabermetric whipping boys as the main analysts of the day railed on the franchises that gave them playing time that they apparently didn’t justify through their at-bats.
However, as we’ve begun to come up with ways to quantify defensive performance, at least at a blunt level, we’ve realized that we were wrong. Badly wrong. Defense is a huge, huge part of baseball. To steal a quote from Bill James, much of what we call pitching is really defense. For all intents and purposes, the Chicago White Sox are world champions because they were a team of terrific defensive players who weren’t so awful at hitting or pitching that it would offset the strength of the team.
Clearly, if you’ve been reading our posts for the past few years, you’ve seen a change in the way we evaluate players. I stumped for Pokey Reese (well, a healthy one, anyways) for two years. The support of the signing of Jacque Jones is largely based on his defensive abilities and the effect he would have on the team’s pitching staff. However, it seems to me that the importance of defense is still underestimated by most fans. In fact, while I think our readers represent the upper tier of intelligent fans, I’d say that I would expect that most of you underestimate the importance of defense as well. We see this manifest a lot in the comments, when people suggest doing whatever it takes to bring in “The Big Bat”, including terrible ideas like shifting Richie Sexson to left field.
I’ve been trying to think of a good way to explain just how important defense on a team level can be in a way that would make sense and still be understandable. While there’s some good defensive work being done, most of it deals with mathmatical theory of regression analysis and other non-interesting reading material that I’ve slagged through. But, tonight, Chris Dial introduced his “method” of evaluating defense over at Baseball Think Factory. It’s not that tough a read, and it makes a lot of sense, especially if you’re familiar with some of the previous work that’s been done in this field. Now, there’s still enough problems with individual defensive stats that I’m not linking to this to back up any claims about a specific player, and I’m not saying that his numbers are the answer. I haven’t even given them more than a cursory glance. However, there was one table that jumped off the page at me that I’d like to share, and hope its importance sets in.
Position AvgZROps Runs/play Runs/Season AvgZR* Runs/Season
Perfect Average
1B 281 .798 224.24 .870 195.09
2B 507 .754 382.28 .822 314.23
3B 430 .800 344.00 .783 269.35
SS 532 .753 400.60 .835 334.50
LF 348 .831 289.19 .861 248.99
CF 462 .842 389.00 .888 345.44
RF 365 .843 307.70 .873 268.61
Essentially, what this table tells us is how many runs an average player at each position saves over the course of the season. The baseline it’s compared to is zero, so don’t freak out because the numbers are huge. Essentially, what this is saying is that if you played a season with just 8 fielders and left everyone in their normal spot but eliminted the shortstop, you’d give up an extra 335 runs. If you punted the left fielder, you’d “only” lose 249 runs.
The difference between the positions should jump out at you. Center Field is the most important defensive position in baseball. It is way, way more important than right or left field. Think of it this way; an average center fielder will have the opportunity to save his team 70 more runs than his average right field teammate. Holy crap. 70 runs!
Defense isn’t something to be looked at as a minor factor in player analysis. We certainly haven’t discovered the key to perfectly evaluating the abilities of each player, but we should all be past the point of thinking that the vast majority of a player’s value is tied to what he produces with his bat. What he can do with his glove is far from insignificant, and needs to be considered.
In other words, just say no to Richie Sexson, left fielder.
Comments
74 Responses to “The Importance of Defense”
JPMouton — I disagree with your last two posts. Your argument is close to being in direct contradiction to Dave’s premise in the original post. I trust your next argument won’t be — “there is no skill really in catching that fly ball to the CF”.
All fly balls to CF are not created equally. Neither are all throws to 1B, or pitches to C. All ML players should be consistently making the routine plays. We are trying to measure the skill of the defender to turn *all* plays (including the tough plays) into outs and giving his team a consistently higher chance of winning a game. So these skills, although not measured here, *are* important *and* meaningful when evaluating a player’s value.
46 –
The catcher is the only player of the 9 who can see the entire field in front of him, including all positions. He traditionally “runs” the defense by calling plays, setting fielder positions, calling the pitches, directing throws and verbally directing players during plays when their backs are turned (such as when receiving a ball from the outfield). This is why baseball types refer to the catcher as an important defensive position…for generalship. It’s not necessarily that he prevents more runs by fielding tough plays at his position once the ball is put into play.
Any more, this traditional role is mitigated by the fact that the benches/coaches provide a lot of direction from the bench.
sorry: “these skills” … meaning the ability of the 1B and C to catch a non-batted ball
I know that ZR doesn’t take into account throws received. That’s why I was asking whether or not this was a reasonable shortcut in evaluation – I’m pretty much wondering if it the difference between, say, a Ryan Klesko and a John Olerud manning the bag is due to fielding or receiving ability, or a combination therof. If receiving is at all important, completely taking it out of the equation is not going be a reasonable way to construct an infield model.
My thoughts on catcher: Aren’t we first comparing a living, breathing player to a glove stapled to a sandbag to get all these numbers for the rest of the positions? Now run that exercise minus a catcher and see if anyone gets out (hint: no). Catcher defense is a different beastie than standard fielder’s defense, which is probably why it’s not mentioned in Mr Dial’s model. Apart from plays at home and those odd pichter-catcher-first double plays, a catcher deals entirely with balls *not put in play*, wheras there aren’t too many plays a fielder makes on similar pitches (foulouts and tags applied on a catcher’s throw being the only two that spring to mind).
That said, it’s not so drastically different from fielding that you need to rely on things like ‘game calling,’ etc to evaluate it. But when we don’t have the tools to properly evaluate play by play defense, how do we measure the fielding part of a catcher’s duties?
Yeah Derek, I realize that the throwing arm for a C is incredibly important. And Incredibly valuable(someone did a study showing how dangerously close Pudge caught up to Piazza in various years accounting only for the arms), but my point is that OTHER than throwing arms, C defense is greatly overblown. And that players that are rumored to be amazing defenders that don’t have ridiculously high CS%(like Mike Matheny, Jason Varitek got a GG this year, etc) have un-earned defensive reputations. The difference between a great defensive C and a crappy one resides solely in his arm, and a team gets like 140 SB attempts a year on them.
51-My point isn’t nescessarily that there is no skill in catching the throw at 1B. I’m saying that there is no noticeable difference between any 2 players. The difference between an “amazing” reciever at 1B and a terrible one is going to end up being statistically insignifigant.
Also, anyone else think the NY Post is making this up:
http://www.nypost.com/sports/31152.htm
One of the key skills for a 1B may show up in DP efficiency – percentage of DPs completed in DP opportuniites, adjusted for GB/FB tendencies of the pitchers on the staff.
Because of the speed of the play and the intereference at second base, there should be a higher fraction of errant throws to 1B to complete the double play. A 1B who is better than his counterparts at fielding errant throws should contribute to increased DP efficiency.
I haven’t seen any detailed work on this. I do recall seeing some data one time that showed that whenever Olerud switched teams, the fielding percentage of the infielders on his new team increased and the fielding efficiency of the infielders on his former team went down. I also did some work a couple of years ago that suggested that during during the Guillen/Ordoñes -> Boone -> Olerud years, the Mariners were one of the most efficient teams in baseball at converting DP opportunities. Are Some Fly Ball Pitchers Also Good Double Play Pitchers? – Part 2
Good stuff, everyone.
A few basic points:
Obviously, this metric is useless when it comes to evaluating catcher defense, which is why Dial didn’t even try. Sometimes, it’s just best to say that our information is so limited that we admit that we have no idea how to quantify something. So it is with catcher defense. That is an entirely different subject, and one on which we have a long, long way to go.
I had thought it would be obvious that a Zone Rating system was not measuring a first baseman’s ability to receive throws. I disagree with JP; this is obviously a skill, and it does vary from player to player, and it does make a difference. This system does not try to capture that, so we should take the first base ratings as less important than all the others.
Also, Chris’ methods are wide open. He’s been posting about defense for years and years, and he’s laid out more information on the subject for everyone to see than anyone else around. Assuming that, just because you didn’t see the answer you were looking for in the one article i linked, that he’s trying to make this a proprietary system is a bad, bad assumption. Everything Chris does is open source.
One intangible about 1B is the confidence he gives the infielders. Impossible to measure I suppose but I think if you are at SS and make a tough dive and are off balance knowing that you are throwing to John Olerud has to make you feel better about the throw than if you are throwing to say Frank Thomas. Does that subconscious confidence make you more likely to get the runner because you don’t hesitate on the throw?
I suppose the counter is that even if that is true how often does it happen? Once or twice a season is no big deal, once or twice a month might matter
Think of the tag-up advance from 2nd to 3rd in these terms: It’s only critically important 1/3 of the time: when the fly caught is the first out. A play at the plate always is important, so maximize your best arm’s chances by putting it in CF.
Dave, I could agree that it might be a skill. But realistically, the difference between someone who is great at it and someone who is not good is going to be a handful of plays a year in terms of errors (because in manycases, when the 1B misses the ball, it is going to be an error on the player who threw the ball) or “missed balls”. The sheer number of putouts that a 1B makes(1100+ a year) would hide the number of missed balls a 1B is going to have. By treating errors(the only way to miss a throw right at you that I can think of) akin to saves(that is an error is like a blown save that a MR gets in the 7th inning, when its not really a save situation) the player does not get credit for a play that he is not really creating, but can still get negative credit when he screws it up. So, basically, my point is that it isn’t really screwing up the 1B stats to take out the putouts on the bag.
Yeah, the attacks on Chris are pretty unfair. Laziness on your part does not equal fault on his part.
Also, Dave, as a non-Seattle fan I really loved your article at The Baseball Analysts. I don’t watch that many Seattle games(I only saw Felix pitch 3 times last season, but I picked 3 good ones, his debut, his home debut, and the Randy Johnson matchup), but he is eagerly awaited throughout the baseball world, not just in Seattle, so the more I get to read about him the better.
JP,
I think you’re either vasly underestimating the amount of opportunities a first baseman has to catch non-optimal throws, or you’re missing the point. I’m not sure which one.
There are something like 1300 throws to first base a year, on average. If one out of every 50 is what might be considered a “tough play”, that gives us 26 chances a year for a great fielder to make a play that a bad fielder will not, just on receiving throws. If you want to say that the great fielder will only make 20 of those 26 plays, and the bad fielder will make 5 of them, that’s still a difference of 15 outs, or close to 10 runs, over the course of the season.
10 runs is far from statistically insigificant.
Oh, and thanks. Glad you liked the Felix piece.
You are absolutely right. This is one of the reasons that Beltre was more valuable to the team than his offensive stats showed (and his offensive stats were better than any other third baseman the M;s have had in years). He saved runs on several occasions. A run saved should count for as much as an RBI.
I’m not nescassarily saying that there aren’t 26 “tough” plays a year, nor that the difference between great and terrible is 15 outs a year. What I’m saying is that they don’t show up in the stats when you just include a 1B putouts. Because only 1/50th of the plays are actually signifigant, the difference between a great defenseive player and a mediocre one(who converts 12 of the 26 outs) is no longer 8/26, its 8/3000, which makes the deviation between great and crappy look smaller than it is. The other problem is that of the 6 plays the great defender doesn’t make, a good number of them are errors on the player who threw him the ball.
What I am NOT saying is that 1B defense is not important. I of course don’t believe that. What I am saying is that I don’t see much of a way to judge a 1B defensively without excluding the “receptions” he makes of other players plays.
How many errant throws did Sexson recover last year? I’d say his “reception” of other positions’ balls in play is extremely important.
Ichiro to CF is ludicrous. Reed is better, now. Ichiro to CF would have made better sense in 2001, except we had Cameron. Then Ichiro grew to love RF, and whatever makes Ichiro’s mojo, Ichiro gets. At his age, moving back to CF (Ichiro played CF in Japan) doesn’t make all that much sense. Aging OFers with good bats move to the corners.
“Yeah, the attacks on Chris are pretty unfair. Laziness on your part does not equal fault on his part.”
Really? There were attacks on Chris Dial? Looking through all the comments, I really don’t see any. For my part, I saw some (important) numbers that were pulled out of a hat. It’s probably a legitimate hat, but I would like to know where I can find that hat. So, I asked for a little help. For my part, I got nowhere with various google searches. I also got nowhere with searches of Baseball Think Factory, which returns 5 results for “runs per play” and none of them describe how to come up with the runs/play number or give a cite for where it comes from. Then, after Dave’s last comment, I was hoping it would be obvious by looking through a listing of Chris Dial’s entries to find information on this. After the 6th or 7th page of game threads that he posted, I decided that I was barking up the wrong tree.
Honestly, I’m just frustrated at this point. I would like to know more about this and I’ve spent a lot of time looking for it with no good results. I don’t see how that makes me lazy, and I don’t see how that means I’m attacking anyone. If anyone has a link to anything relevant about the runs/play number, I would appreciate it.
To be fair to yourselves, USS Mariner, since I first began visiting here or your other sites, I was never under the impression that you thought defense was unimportant.
My impression of your opinions was always that you just didn’t know how to measure it. And I didn’t get the impression that you thought the inability to measure it made it insignifigant.
I say this because defense has always been my very favorite part of the game and I never really thought it was being slighted here. Elsewhere, yes. And, with the exception of shortstop, say, I think the “non-stathead” community undervalued(-es) it, too.
Re: first base. Olerud was an eye-opener for me. His play at first was terrific and made me re-think the position as a place to put your defensive liabilities.
I think Dave’s one-in-50 toss-out might be more like one-in-20, or even lower. I don’t think a lot a games go by without one difficult catch at first. And anytime you handle the ball, there is a chance of bad things happening. First basemen handle the ball a lot.
I agree that the table is an excellent first step towards appreciating the value of defense. The fact that there are 70 runs up for grabs between an average centerfielder and a perfect centerfielder is valuable to know, but an accurate assessment of defensive value would be more easily identified by knowing the standard deviation for defensive performance.
I just went to ESPN and took the zone ratings for all qualifying centerfielder for the last 6 years. The mean of the distribution is 0.883, close to the 0.888 value in the table Dave posted. The standard deviation is 0.026. Taking plus or minus one standard deviation as the definitions for a good and a bad centerfielder, a good centerfielder saves 20 runs a year compared to a bad centerfielder. During that six year span, the difference between the best (Jeremy Reed 2005) and worst (Ken Griffey 2005) seasons by qualifying centerfielders is 54 runs.
To calibrate these differences, the difference between a ‘good’ and a ‘bad’ centerfielder is the difference in earned runs allowed by Bartolo Colon and Jeff Weaver in 2005. The difference between Reed and Griffey is the difference in earned runds allowed by Roger Clemens and Chris Capuano in 2005.
[I have a plot of the histogram if you would like it posted and can tell me how to post it.]
OLERUD (# 68) – Here’s a column that says the same thing, and superbly backs it up:
http://espn.go.com/mlb/columns/schwarz_alan/1377032.html
Another question regarding catcher defense – is it really true that at the major league level, all catchers set the defense? I see a lot of coaches in the dugout giving signals to the fielders themselves, and also relaying signals, which could be through the catcher. Could it be that a lot of catchers are just relaying signals from the bench rather than really deciding on the defensive positioning? Kind of like the fact that Peyton Manning is out there calling plays and audibles all the time, but most QBs just get the call from the sideline and run what they’re told…
Zero –
There is a lot of signal relaying these days from the bench, but remember…the catcher calls (most) of the pitches. Good ones do so based on (1) what the pitcher is good at, (2) what the batter is bad at, (3) how the defense is set, and (4) the situation, including the batters tendencies and the defense’s tendencies. It is in this detail, repeated hundreds of times throughout the course of a game, that a “good” defensive catcher should seperate himself from a “bad” one.
Tough to measure directly, but there are some indirect measures out there (like how defenses/pitchers do when certain catchers are in the game).
I feel like that metric vastly undervalues 1B due to him recieving throws from other fielders. I know other people have hit on that, but it just seems wierd to me. I’m also curious as to how much things like assists and holding runners are valued in this system. It doesn’t seem like it’s very much.
I’m one of those guys that has been really angry with the stats guys for a long time due to undervaluing defense. I’m glad people are starting to see the light finally!
I wonder how Manny looks if you look at him with the lense of his terrible defense. I can’t imagine you’d really want him on the field unless you had to have him out there.
Dave re: #59 on Chris Dial and methodology, I assume nothing but only consider contextual issues relevant to any piece of work. Your clarification gives context, for which thanks; hence (next to) no issue.