Is it possible to measure defense in baseball?
I do not have an answer to this question, though I've read much on the subject, and find myself struggling with (at least) two loudly competing voices in my head when debating the issue. What I do know is that defense is important.
Again, I'm not sure we can measure HOW important, but a single play or misplay can absolutely swing a game in an instant and any major league pitcher will tell you that having a great defense (or defensive player) can add a great deal of confidence and can even affect his approach against the hitters he faces.
Over the past few weeks, I've been looking at an in-depth comparison of two of MLBs greatest players; Troy Tulowitzki and Mike Trout.Specifically, we've looked at how they match up against their positional peers, mostly on offense. It is obvious to anyone with eyes that both players are exceptional defenders, but how much above the average are they and to what extent does that ability actually translate to helping their teams win ballgames?
I was lucky enough to broach this topic with Jonah Keri at a recent bloggers panel in Denver. He reminded me that there is a name that those smart guys use for the argument I've been making this whole time: standard deviation. Let us look at how Mike Trout and Troy Tulowitzki deviate from the mean in terms of their positional defensive value.
Let's look at the numbers before discussing their validity.
The stat I looked into for this exercise was UZR/150. One of the biggest problems in comparing Trout and Tulo is tha tthere careers don't line up. Trout has much less service time but is still theoretically coming into his prime while Tulo is right in the middle of his and has more numbers over the course of his career to both hurt or help his case.
I think it has some flaws (and we will get into that) but at least UZR/150 attempts to rate defenders based on an equal number of theoretical innings and defensive chances so comparing across many years shouldn't matter. As such, I wanted to compare each player not just to their contemporaries, but to the best players at their positions over the last decade since both Tulo and Trout have claims to being historically significant.
In the last 10 years, Mike Trout ranks at 26th among outfielders in UZR/150 and tenth among just center fielders. Several players ahead of him are retired, but active ones among outfielders include: Josh Reddick, Jason Heyward, Ben Zobrist, Gerrardo Parra, Alex Gordon and Alfonso Soriano ... hold onto that last name for a second.
A few notable center fielders with higher UZR/150 than Mike Trout: Peter Borjous, Carlos Gomez, Brett Gardner, Nyjer Morgan, Andres Torres, and Franklin Guitierrez.
Naturally, Trout has better numbers as just an outfielder because of his time at the corner spots (10.3 compared to 9.4 UZR/150 in center) but is still obviously among the best in the game by this measure, but not THE best and not that close.
By the same measure, Troy Tulowitzki is fifth among shortstops that have played in the last decade behind only Andrelton Simmons (whose numbers haven't had time to balance out yet and may come down) Zach Coazrt, J.J. Hardy, and Elvis Andrus and ahead of guys like Jimmy Rollins and Brandon Crawford.
Zone ratings are more important to outfielders because catching the ball before it hits the ground is their primary goal while infielders are more often judged by what they do once they get the ball. Hardy and Cozart have been consistently amazing at getting to the baseball and should be given all the credit in the world for that, but they are the only two players you can say (based on this stat and because Andrus is only 0.2 ahead) are appreciably better than Tulo in this regard. But as I broke down in part two, those guys can't hold a candle to what Tulo does after he gets the ball and aren't even in the same universe in the batters box, making them much less valuable overall.
According to Fangraphs Inside Edge Fielding tool, Troy Tulowitzki has been tasked with making 16 more "Remote" plays (1-10 percent likelihood of play being made) 11 more"Unlikely" plays (10-40 percent) 12 more "Even" plays (40-60 percent) and 41 more "Likely" plays (60-90 percent) than Mike Trout since 2012.
Keeping in mind that 2013 has been Tulo's worst defensive year to date, Troy Tulowitzki has been forced into 78 more opportunities on plays outside of the routine and the impossible than Mike Trout in the last two years. That is a lot of potential to add value to winning individual games where Trout simply cannot. Trout does have the higher conversion rates except in the "Likely" category but due to the sheer volume of extra plays, Tulo still ends up making more plays of every variety even when he misses time due to injury.
But what about dWAR? I have tons of issues with this measurement, the main one being that it is a counting/cumulative stat, which I don't believe is an effective way to look at defense. My dad always told me that hitting is about what you do and fielding is about what you CAN do. Every hitter is going to get 3-5 ABs per game and over the course of so many games the results will give you a very good indication of overall ability.
This kind of results oriented analysis doesn't work for defense.
Carlos Gonzalez is a perfect embodiment of this conundrum. Putting aside the weird-ass year he is having in 2014, Gonzalez has been a player over the last few seasons who on any given play could be the best outfielder in baseball ... or one of the worst. I have seen him jog slowly to balls over his head to avoid running into the wall and in general just fail to hustle all the time. This may be in large part to help avoid injury and I usually notice it in games when the Rockies are almost certainly going to lose.
But here is the kicker, he didn't all of a sudden lose the ability to play excellent defense. He just doesn't do it sometimes. But (before knee tendinitis) I always knew that if the Rockies were in it, CarGo could absolutely spin a gem at any time regardless of what his overall defense looked like prior to that moment. It can, and should, be argued that this is not the optimal way to approach playing defense, but it is not directly translatable to a player having the same approach at the plate. Every at-bat is precious, but there is no reason to go sprinting and diving when you're down (or up) seven runs in the ninth inning.
Every defensive play is highly contextual and every excursion of great effort in the field is an injury risk in a way that simply swinging a bat is not. This is why I believe that defense should be measured by skill set and not resume. On defense, I want to know what you could do on the next play if the team really needs it and so does your pitcher. The results of the past (and your healthy history) don't necessarily affect that.
For the record, Mike Trout ranks eighth in dWAR behind Carlos Gomez, Juan Lagares, Lorenzo Cain, Craig Gentry, Michael Bourn, A.J Pollock, and Jarrod Dyson.
Okay, what's going on here? A fair point has been made by some that I may lean more toward Tulo on these issues because I have seen him play more than I have seen Mike Trout. I watch more Trout ABs than most would likely realize but it is still fair to say that I have more exposure to Rockies players and therefore maybe more inclined to gloss over weaknesses or make a more in-depth case on their behalf.
But Alfonso Soriano works as a perfect example of someone I have seen play probably even more than Tulo because he has been around longer and my roomate watches almost every single Cubs game.
I have enough eye-test information in my head to know that on no planet is Alfonso Soriano in any way a better or even comparable defender to Mike Trout.
Soriano regularly misplays balls off the bat, takes terrible routes, takes forever to get to the ball because of some of the squarest wheels I've ever seen on someone who used to be known for speed, and makes bad, low baseball IQ decisions. The reason he is rated highly by some of these defensive stats is because coaches figured out how to position him in ways that would cause the least amounts of damage. They effectively fooled the stats by having him take zero chances and play super deep so he can keep the game in front of him at all times.
Former "five tool" player Alfonso Soriano is essentially the anti-CarGo in this regard. Rather than occasionally making bad plays but being capable of making exceptional plays, Soriano only ever makes safe plays. It's a smart team strategy but obfuscates the talent level of each player.
There is zero chance that Alfonso Soriano covers more ground in chasing down flyballs that Carlos Gonzalez, even with the knee tendinitis. The fact that Soriano ranks where he does on these lists is proof that we have a long way to go in objectively measuring defense.
Alfonso Soriano's dWAR in left since 2012 is 5.8, good for fifth best in baseball. Carlos Gonzalez has a dWAR of -16.1 placing him 18th. These are the stats informing WAR, Mike Trout's greatest ally.
Because WAR stats are counting stats they will always undersell the ability of any player who misses time. Obviously Tulowitzki fits this category and has seen his counting stats across the board suffer but none of that changes what he is capable of. Health is a commodity and an important one at that, but we should be careful in conflating that with everything else.
For the record, Mike Trout ranks eighth in dWAR since 2012 behind Carlos Gomez, Juan Lagares, Lorenzo Cain, Craig Gentry, Michael Bourn, A.J Pollock, and Jarrod Dyson. Only Gomez and Bourn have more innings played and only by about 400 and 500 respectively.
Tulo is 12th behind such illustrious names as Clint Barmes and Jhonny Peralta. Every single shortstop ahead of him in WAR has far more innings that Tulo including six who have over 1,000 more innings played.
Judging ultimately based on WAR, or dWAR, for a player like Tulo is disingenuous. It will tell you way more about his ability to stay on the field than it will his ability once he is on the field. We know he has been hurt, we have that info, but an inability to accumulate dWAR doesn't change the defensive impact he will have on today's game or tomorrow's. And it isn't just the health concerns as we've discussed how radically inaccurate the defensive rating are which could be either highly underrating Tulo, Highly overrating Trout or both.
So let's put aside result-based stats for a moment and look into skill sets.
FiveThirtyEight and Tools
Jeff Sullivan (usually of Fangraphs) at FiveFhirtyEight.com is smarter than me. By like a lot. And he recently put together some numbers on "tools," trying to find out who the true five-tool players are. Mike Trout did not make the list but both Troy Tulowitzki and Carlos Gonzalez did. Trout did come out as the "toolsiest" player. More on this in a minute.
Here is how his numbers rate each player:
NAME AVERAGE POWER SPEED ARM FIELDING
Mike Trout 2.8 2.3 2.1 0.2 1.6
Carlos Gonzalez 1.1 1.9 1.3 2.2 2.0
Bryce Harper 1.2 2.2 1.5 2.2 1.4
Troy Tulowitzki 1.5 1.5 1.1 2.1 2.0
Starling Marte 1.5 0.8 2.2 1.7 1.9
Yasiel Puig 1.8 1.5 1.6 1.6 1.3
Alex Gordon 1.0 0.5 1.1 2.5 2.4
Robinson Cano 2.1 1.2 0.1 2.2 1.6
Evan Longoria 0.6 2.0 0.5 2.0 2.1
Manny Machado 0.8 0.4 0.9 2.6 2.6
This led Sullivan to conclude that the only true five tool players in baseball are Tulo, CarGo, Bryce Harper, and Yasiel Puig. Trout missed out because of his arm, which is barely above average.
Both this and Fangraphs Fans Scouting Report rate CarGo as the better all-around defender than Trout and we don't have time to get into the absurdity of sticking him in left field. They also have Harper, Starling Marte, Alex Gordon, and Carlos Gomez (not on this table for offensive reasons) ahead of Trout as defenders. The lack of arm may not seem like that big of a deal, and at first I didn't think it was either, but Sullivan found that a good arm is equal to excellent fielding in terms of overall value.
Each tool also has its own relationship with WAR, which can tell us how important a given tool is to a player's value. The table shows the correlation coefficient for each tool against projected WAR/600:
Average Power Speed Arm Fielding
0.53 0.53 0.13 0.43 0.43
The vast majority of Trout's value comes from him being a phenomenal hitter, which is the most important skill. But defensively, especially realizing how important a good arm can be, he isn't elite even just among outfielders which means he lags even further behind the defensive value of elite shortstops and catchers.
When you add it all up, that offensive prowess lands Trout as the toolsiest player.
Mike Trout 9.1, Carlos Gonzalez 8.5, Bryce Harper 8.5, Troy Tulowitzki 8.3.
So the question is, does being a shortstop make up for the eight-tenths of a point that Tulo lags behind? In my mind it does in terms of value added to his team's likeliness to win and that is why I started this whole process.
We will take one more look at how this defensive information plays into the overarching picture in the fifth and final entry into this series. We'll discuss age, contract, ballpark factors, and more deviation from the mean but I think it is safe to say that Troy Tulowitzki is bringing more defensive value on a game-by-game basis than any non-catcher in baseball and, with the exception of last season, has been since 2007.
So is Mike Trout just that much better at offense that it makes up for it? Or is offense just that much more important? Stay tuned to Purple Row for more.