The fog of glove: how to measure defense?

Yeah I don't know enough about the way the defense is broken down to say why Alomar is rated as average, but defensive metrics are measuring things that might be hard to discern with the naked eye (like whether or not on average a similarly positioned fielder even gets to a ball to make a routine play). Also spectacular or memorable hitting by average dudes happens all the time and some people do try to convince other people that this average hitter is really some secret clutch superstar and ohmigod we better sign him next year, etc.

Also Gold Gloves should be a yearly measure. There is a lot of fluctuation in the defensive statistics (wtf Mike Trout) but I still think that's gotta be a better yardstick than some dude's overall impression of a player's defense based on maybe seeing him a couple of times plus consideration of reputation.

Which is exactly Joe's point: advanced metrics, with room for first-hand observation.

I know that range is one of the hidden things that Fielding Average doesn't measure. In terms of Alomar, his range was the precise thing that was so much fun to watch. That play in the '92 Series where he was out in short right field--that was the kind of thing we were accustomed to.

I meant one specific average hitter consistently doing spectacular things, not the whole group--I might not have been clear there.

I'm not even sure what a spectacular hit really is anyway. Maybe Juan Pierre hitting a home run.

I was thinking of spectacular one-game accomplishments by otherwise undistinguished hitters--Freddie Patek or Dioner Navarro hitting three home runs, Phil Weintraub knocking in 10--but you're right, that's not a very precise analogy to making great plays in the field.

I'm looking at Alomar's fielding stats, and he does well in range factor/game...except for his time in Toronto.

With the Padres: 2nd, 2nd, 1st
With Toronto: 5th, --, --, --, 4th
Leaves Toronto: 4th, 3rd, 2nd over the next three seasons

Where he does well in Toronto is in Fielding Pct: 5th, 1st, --, 3rd, 1st.

I give up.

did he play on turf anywhere else?

As I sat in the dentist's chair this morning, that occurred to me too. Maybe turf boosts fielding average (truer hops) but reduces range factor (ball scoots faster, players are more cautious about diving). Or maybe the Jays had a higher-than-normal strikeout/flyball staff. I honestly don't know--I'm sure people have tried to figure these things out with regards to Alomar.

I wonder if people in L.A. and St. Louis experience the same disconnect with Jim Edmonds, whose dWAR figures seem to be fairly ordinary. I'm someone who only knew Jim Edmonds the Human Highlight Reel.

Alomar played on grass in Jack Murphy Stadium and (of course) Camden Yards, so maybe there's something there. You'd have to check how the changeover affects other players, and I'm sure many people have.

wyers wrote a good thingie on WAR and defense a few months back: http://www.baseballprospectus.com/article.php?articleid=20569

we have GG "finalists"

http://www.baseballnation.com/2013/10/25/5029232/gold-glove-announcement-al-nl

http://www.sbnation.com/mlb/2013/10/25/5027950/rawlings-gold-glove-award-finalists

Might get a glimpse of the Fog of Glove tonight with Ortiz at first base.

OK, Nolan Arenado? I def watched some Rockies this year but he flew under my radar.

http://sabr.org/latest/2013-rawlings-gold-glove-award-winners-announced

i never get to actually watch much baseball, so no idea if these were good calls or not - but Hardy? is he actually that good at D?

Neyer:

Hardy was fine, but doesn't have the range of fellow finalist Alcides Escobar, who probably should have won this one.

http://www.baseballnation.com/2013/10/29/5044554/2013-gold-glove-winners-fielding-awards-sabermetrics-statistics

hardy wasn't as good this year as he always is, but he's always good. he actually is a defense-first SS, despite the bombs.

which surprised me when we got him. and when i say "good as he always is" i mean "basically flat out incredible, one of the best in baseball." if only he could obp.

Minnesota giving up Carlos Gomez to get JJ Hardy then turning around and shipping Hardy after a bad season to B-more for a bag of balls is looking like a double bad deal at this point.

the one you should be ??? about thermo is jones, he's never been more than just acceptable out there and now that the GGs split the OF awards it's especially weird, he keeps taking these awards for CF when CFs have always been the winners. it's like crowning him king of the american league outfield.

SABR Defensive Index rankings thru mid-Aug:

http://sabr.org/latest/sabr-defensive-index-rankings-released-through-august-17-2014

One of the best arguments for the validity of new defensive metrics I've read (even though the argument is between the lines). I've often mentioned that I've been very slow on the defensive side of sabermetrics, but this frames the issue in a way that makes sense to me.

http://joeposnanski.com/joeblogs/a-little-bit-above-normal/#more-2457

Although it does deal with the easy part of the question: convincing a skeptic that Torii Hunter's defensive metrics say he's not anywhere near what he used to be isn't all that difficult, as common sense says the same thing. Convincing me that Roberto Alomar in his prime wasn't anywhere near what I thought he was, that's tougher.

the full SABR Defensive Index™ rankings, through games of July 12, 2015.

http://sabr.org/sdi/2015-07-12

Do you guys think putouts per inning or putouts per nine innings is a decent way of assessing an outfielder's defensive value over an entire season?

out of date

Obviously errors and assists are not a part of the stat, but other than that, I don't know what is missing.

I'll throw out the example that's bothering me: Matt Kemp 0.6 WAR for the year, Gregory Polanco 2.5 WAR.

Kemp .265/.312/.443 23 HR, 100 RBI
Polanco .256/.320/.381 9 HR, 52 RBI

In right field this year:

Kemp 1282 innings, 269 putouts (8 errors, 10 assists)
Polanco 1220 innings, 247 putouts (8 errors, 13 assists)

I'm assuming Matt Kemp's lower WAR is due to defense, but he has a higher number of putouts per inning than Polanco. What is being measured that accounts for the difference in WAR?

not really sure. maybe Kemp had a lot more balls hit towards him that he didn't get to?

there's also the base running component of WAR tho, that would probably better explain the gap.

Looking at the fangraphs stats, Polanco does have more OOZ putouts - 97 vs. 71 for Kemp. They have almost exactly the same percentage of putouts per ball-in-zone.

26 more OOZ putouts gives a guy with much worse hitting stats four times the WAR as the other guy?

Polanco does have 27 stolen bases and Kemp has 12. Kemp has 38 more total bases, though.

Polanco has a 5.3 baserunning, while Kemp is at 0.9

Kemp 1282 innings, 269 putouts (8 errors, 10 assists)
Polanco 1220 innings, 247 putouts (8 errors, 13 assists)

these are not good stats to evaluate defensive value. especially errors. scorekeepers don't generally give out errors for plays where a really slow outfielder (or one who takes bad routes or gets bad jumps) comes nowhere close to catching the ball. if they did, kemp would have like 160 errors! (sorry matt kemp)

you're using baseball-reference WAR, i think? i'm much more familiar with fangraphs, but i'd just dig into the glossary/FAQ for bWAR and figure out how they calculate defensive and baserunning value.

Polanco has a 5.3 baserunning, while Kemp is at 0.9

Kemp was the better offensive player by far. Thirty-eight more total bases, I think, means more than these numbers. So does forty-eight more RBIs.

scorekeepers don't generally give out errors for plays where a really slow outfielder (or one who takes bad routes or gets bad jumps) comes nowhere close to catching the ball.

But I mentioned just above that Kemp has pretty much the exact same percentage of putouts per ball-in-zone that Polanco does according to fangraphs.

By the way, fangraphs WAR for these two is Kemp 0.4, Polanco 2.3.

there does seem to be more of a disparity between the two than meets the eye initially. but i think you were on the right track before: going by FG's revised zone rating, the two converted outs equally as often for balls in their zone (.888 for kemp and .882 for polanco). that doesn't tell us much though, since balls that are in the zone, especially for outfielders, generally aren't difficult plays and there is not going to be a huge difference between the best and worst players. (kevin kiermaeir, who by some metrics had the best defensive season of anyone in about 10 years, had a revised zone rating of .928, just 3.5 points higher.) so there isn't much to be gained by being really good at catching routine fly balls, since just about anyone can do this. where outfielders can add value is by getting balls out of their zone -- balls that normally shouldn't be caught and would tend to go for doubles and triples. despite seeing significantly less action -- polanco had about 40 fewer balls hit into his zone than did kemp despite comparable innings -- polanco managed to turn 97 balls hit out of his zone into outs, compared to 71 for kemp. so again, on the balls that actually matter -- the ones that would otherwise go for extra bases -- polanco made 26 more outs.

this is obviously a very simplified answer, but it's the basis for advanced fielding stats. other things, like strength of the pitching staff, home field, etc are factored in too

Do some balls out of the zone fall in front of the outfielders for singles? I agree 26 hits is a lot, but if half of them are singles...

Ultimately, I think you've got quite a long ways to go to pull back 60 points in slugging percentage, 14 more home runs, and 48 more RBIs. I'm not sure that 26 more hits allowed does that. Or even comes close? Of course, it not only has to pull it back enough that they're even, but enough that Polanco ends up with a significantly higher WAR.

you gotta stop saying the r-word

kemp was not that much better at the plate. he hit for a .750 OPS to polanco's .701, good for a wRC+ of 109 to polanco's 94. meaning kemp provided 11% more value at the plate than the average hitter, and polanco 6% less. taking into account base running (which favors polanco), kemp was worth 8.1 runs above average offensively to polanco's 1.0. maybe about half a win. polanco was worth about 25 more runs in the field, by FG's calculations. sum those, and you have about a 2-win difference

sorry, that should say kemp provided 9% more value than the avg hitter

But I cannot see where 25 runs comes from. I can see 26 more hits allowed. Actually, I'm not sure that Kemp allowed 26 more hits. Kemp played about seven more games than Polanco and actually had 22 more putouts total. Maybe those are outs most players make but Kemp was there, recorded the putouts, and that has some value.

Most hits are hits to the outfield and not infield hits. So maybe those 26 hits are average hits. Even if they're not, they don't equate to 25 runs. If they're average hits, they maybe equate to something like ten runs.

As for RBI's, Polanco is a leadoff hitter and it's not a fair comparison. He had 83 runs scored and Kemp had 80. Polanco, with his 52 RBIs, did have good numbers with RISP this year. As did Kemp, as hitters with 100 RBI, you would think, generally do!

yeah, i can't exactly explain where the 25 runs came from, that would be a question for dave cameron or somebody. there are more layers and adjustments that go into it than the stuff i mentioned, though that's the main idea/backbone of it. but to answer your original question, errors and putouts are not a good measure to evaluate fielders, because the extent to which they take into account range is limited. as far as these two players go, polanco was marginally worse at the plate (though some of that was mitigated by superior base running), but far better in the field. it's certainly possible that the metrics overvalued his edge on defense, but the fact that both systems came to the same conclusion makes it more likely than not that polanco was a more valuable player overall this year

I don't understand wRC, but if you use James' runs created stat, Kemp had eight and a half more runs created than Polanco (80.3 vs. 71.8). So, that's about .85 of a win if we're using ten runs for a win. To account for a 1.9 difference in WAR, then, Polanco has to to make up 2.75 WAR in difference.

By having 26 more putouts.

Polanco has to make up 2.75 WAR difference in defense, I meant

And I'm sorry if I'm being stubborn, I understand that "more layers and adjustments" involves ways of trying to measure all of those outs - maybe how hard they were hit, how much time the fielders had to get to the balls, etc. But, again, there is such a lot of ground to be covered to account for 2.75 WAR given Kemp's better offensive numbers, same defensive numbers on balls in the zone, and maybe not that big of a difference on balls out of the zone.

Or, sorry, to account for 2.75 WAR just given the defensive differences

well the 26 outs are kind of a lot, given that polanco saw much less action in general. the other adjustments could include things like park factors, whether their respective pitching staffs give up a lot of ground balls vs fly balls, idk, other stuff. but if you're going to make an effort to understand the game, you have to be willing to admit that defense is more important -- and more complex -- than previous generations have understood. it's not difficult to envision a situation where two averagish hitters -- which polanco and kemp were this season -- could be separated due to differences in defense and base running. it's really no different than, say, mike trout being worth a couple of more wins than miguel cabrera, even though the two are both great hitters

but again as to why in this specific case it turned out this way, i couldn't tell you the exact reasons. but i do think that skepticism that a case like this could actually happen is grounded in an old-fashioned belief that defense is of negligible importance

http://www.fangraphs.com/library/defense/uzr/

here is a decent primer on UZR, the defensive metric FG uses, as well as why we should be using these in the first place

I certainly do not believe that. I'm just stating that the difference that I am seeing defensively is 26 outs. Polanco didn't see much less action - 1282 innings vs. 1220. And I'm not sure that matters anyway because Kemp played more, recorded outs, and was therefore valuable in those 62 innings that Polanco didn't play. But again, 26 extra hits for the other team, if they're average hits, should be worth about 10 runs or so. Given Kemp's 8.5 advantage in runs created, Polanco should then beat him by something like 0.15 WAR and not 1.9.

it's not difficult to envision a situation where two averagish hitters -- which polanco and kemp were this season -- could be separated due to differences in defense and base running.

Well, I think they're separated by about something like 0.85 WAR offensively overall (Kemp over Polanco), so that includes base running.

xp

I mean, the 0.85 number, as I mentioned, comes from runs created, which is a stat that seems to make sense but maybe even James didn't know why? I don't know why. At least it includes total bases so, presumably, base running.

when i say he saw less action, i'm referring to the fact that he played the same number of innings but had far fewer balls hit into his zone. (polanco played 1280 innings, not 1220, so they were essentially neck and neck.) this obviously was due to no fault of his own; likely it's due to the fact that his pitchers tend to allow more ground balls. this is a prime example of something that could be adjusted for, which would make polanco look even better in comparison

also, you might be underselling the value of a hit: http://www.insidethebook.com/ee/index.php/site/comments/run_values_of_events/

also, i don't think 10 runs created necessarily = 1 WAR. (this is obvious: kemp "created" 80 runs last season by james' (now-antiquated) method, but he certainly wasn't worth 8 wins on the offensive side.) 10 adjusted runs above average is roughly equal to that, but the two are different things

the way we're framing this makes it sound like polanco is some defensive wizard. he's not -- he's thoroughly average, both at the plate and in the field. hence his WAR total of about 2, which is that of an average player. kemp is (or was this year) a marginally above average hitter and an atrocious defender. polanco wasn't 25 runs "better" in the field so much as kemp was 25 runs worse

10 adjusted runs above average is roughly equal to that, but the two are different things

this should read above replacement, not average

the response to passan about schwarber's defense is that he has average-ish range and a good arm; passan's response was that that's reasonable in a vacuum, but is he really a better defender than almora or heyward? because that's what the stats are saying

is Schwarber particularly good at positioning or something? also i guess this might make sense if he's being strictly compared to other LFers, who are a pretty slow-footed and weak-armed lot to an extent.

https://www.fangraphs.com/library/defense/def/

sounds like they're being compared to others in the same position AND others in other positions?

he's got a high UZR so i think that means he does well compared to other LFers ... i am not great at reading stats though

is Schwarber particularly good at positioning or something

that might be a factor. dexter fowler's outfield defense was poor before and after he played for the cubs (esp after). whoever is shifting the OF in chicago around seems to be doing a good job of making the most out of mediocre defenders.

fewer BIPs, less significance for defense

Unfortunately, data before 2003 is unavailable, but the stark drop in balls in play over time should make it clear that defense has become relatively less important—perhaps ground balls and hits on the infield were less prominent and this is just a shift back in the aggregate, but the increased rate of extra-base hits would lead us to believe the contrary, that infield defense was even more important in the last millennium.

In addition to fewer balls in play, though, teams have also moved to shift far more often. As Jeff Zimmerman outlined in The Hardball Times Baseball Annual 2014, shifting is in vogue. It makes sense, after all, since between 57 and 60 percent of pulled balls result in ground balls compared to 40 to 43 percent on balls hit to center and 22 to 27 percent on balls taken the opposite way. With advanced scouting and more data available, teams can better identify hitter tendencies and adjust with less risk. The fact that 24 hitters pulled at least 200 balls in play this year makes it even easier to do for the high-pull hitters (as Zimmerman notes, the top-20 players hitting into a shift accounted for over 2,500 plays in 2013, dropping their collective BABIP by 37 points in those situations).

https://www.fangraphs.com/tht/infield-defense-changes-with-three-true-outcomes/

Best fielders in baseball per DRS:
1. Matt Chapman
2. Harrison Bader
3. JaCoby Jones
t4. Nick Ahmed/Lorenzo Cain/Miguel Rojas/Andrelton Simmons

Worst:
1. Charlie Blackmon
2. Adam Jones
3. Asdrubal Cabrera
4. Amed Rosario/Eduardo Núñez/Nick Castellanos/Miguel Andújar

— Baseball Reference (@baseball_ref) August 16, 2018

Amed Rosario may be a terrible fielder but at least he has a 636 OPS. -1.1 bWAR this year

directional outs above average: https://baseballsavant.mlb.com/directional_outs_above_average

kinda wonder if bader's numbers toward LF are skewed by marcell ozuna's shoulder injury -- recently saw bader come wayyy into left to take a sac fly attempt

Athletics vs Yankees at 3rd Base

Matt Chapman: 29 Defensive Runs Saved (most at 3B in MLB)

Miguel Andjuar: -25 Defensive Runs Saved (fewest in MLB)

This image shows difference in how often Athletics/Yankees get outs on grounders near 3B pic.twitter.com/5xGiQwdSti

— Sports Info Solutions (@SportsInfo_SIS) October 3, 2018

this seems like a more relatable, easier to understand way to communicate defensive stats

The Mets right fielder just launched himself full speed into the netting at Guaranteed Rate Field, which extends all the way down to the foul pole, to make a great catch. Extended netting is going to change some outfielders' approaches.

https://atmlb.com/318BlNz

Not full speed, but a dangerous play if the net hadn't been there.

guy is hitting .331 and he's just the "Mets right fielder"? c'mon

I leave all the little details to you, Morbs.

Defense has improved significantly since 2016, to the tune of thousands of extra hits taken away this year--I believe because of improved positioning that puts fielders further away from home plate. https://t.co/D9I0NbBrYY

— Rob Arthur (@No_Little_Plans) June 4, 2021

Just saw the N.L. GG nominees--Arenado's not up this year?

What happened to the good old days, when GGs were based on reputation and how well you hit?

can't really speak to the other two, but ke'bryan hayes is an absolute defensive wizard

which is just as well because he hits like an old-school shortstop

Evidently they look at this pretty carefully now. (Scroll down to "Why isn’t Arenado a finalist?")

https://theathletic.com/4974733/2023/10/18/mlb-gold-glove-finalists-list/

Hey, I think there might be a game on...