Let’s Talk About Eric Hosmer’s Defense
There’s another disagreement regarding the Kansas City Royals and advanced metrics. If you’re still standing, you may take another drink.
This time, it’s Eric Hosmer and the defensive metrics. The Kansas City Star’s Rustin Dodd penned a column over the weekend regarding the disconnect between the perception of Hosmer’s defense and the evaluation of Hosmer’s defense. It’s a well-written and well-researched piece that’s worth your time, but I wanted to dive a bit deeper. And it’s probably about time we had the talk anyway — it’s been an elephant in the room each of the last three seasons, while Hosmer’s won a trio of Gold Glove Awards despite grading as nothing more than an average defender.
So let’s start with the big-picture numbers. We all know that defensive numbers need some time to work themselves out before we can really trust them, but thankfully, Hosmer’s got quite a sample. Dating back to 2011, he’s logged nearly 7,000 innings, which is more than enough for us to feel confident that there shouldn’t be too much noise skewing the sample. So, we grab five-plus years of data, set the innings minimum at 2,000 to give us a pool of 40 first baseman, put Defensive Runs Saved and Ultimate Zone Rating on a per-1,000 inning scale, and Hosmer comes out 32nd, between Logan Morrison and Jose Abreu at roughly -3 runs per year. The best is Adrian Gonzalez, at +7, the worst, Brandon Moss, at -8. Hosmer’s -3 runs over the course of a season is relatively small, but he’s got a reputation for being the best, and these numbers suggest he’s closer to the worst.
But maybe that timeframe isn’t exactly fair, because the numbers see a substantial improvement beginning in 2013, coinciding with Hosmer’s first Gold Glove. So let’s start there. Three-plus years of data, innings minimum at 1,000, 39 first baseman, same methodology as before. This time, Hosmer ranks 21st — closer to the middle of the pack — clocking in at +1 runs saved per 1,000 innings. From below average to very slightly above average, yet still right there behind Lucas Duda and Chris Davis. For what it’s worth, Hosmer’s fielding percentage during that time also ranks 21st — not an error machine, but certainly not error-free.
Dodd’s piece notes Hosmer’s ability to scoop balls out of the dirt, something for which he’s earned a reputation over the years. And it’s true, Hosmer does record significantly more scoops than the average first baseman. In the three-plus post-Gold Glove years, his 143 scoops lead baseball, and even on a per-inning basis, Hosmer ranks in the top five. Part of this could be influenced by an infield defense that puts more balls in the dirt than others, giving Hosmer more opportunities, but the reputation combined with the numbers suggest this is a real skill.
But scoops are just a part of a first baseman’s defense. Even elite scoop artists like Hosmer save just a couple runs per season with this ability. The bigger part of any defender’s game is range, and Hosmer’s manager Ned Yost admitted in Dodd’s piece that “[Hosmer] doesn’t have exceptional range.” And that’s the very reason Hosmer has graded out as the worst defensive first baseman in baseball this year.
Take a look at Hosmer’s 2016 defensive spray chart, with made plays on the left and missed plays on the right:
Compared to a first baseman whose range has graded out as average, like Mark Reynolds:
Or elite, like Paul Goldschmidt:
There have just been way, way too many balls getting past Hosmer this year. The video scouts at Inside Edge credit Hosmer for making just two non-routine plays at first all season. The rest have rolled into the outfield, as a result of Hosmer’s “not exceptional” range. Another way of visualizing this (click to enlarge):
That image — Hosmer lying face-down in the dirt with a ball rolling toward the right fielder — is one the Royals have seen far too often this season, one that doesn’t go for an error but one that negatively impacts the team.
Not that every first baseman would be expected to make this play…
…but Hosmer is slow to react and choppy with his dive, and the ball gets through. That’s far from a routine play, but it’s one that a plus first baseman is expected to make a number of times throughout a year, and that’s the sort of play Hosmer hasn’t been making.
I wanted to do one more thing with some numbers. Inside Edge takes every fielding opportunity and groups it into buckets of difficulty — 1-10%, 10-40%, 40-60%, 60-90%, or 90-100%. The 90-100% plays are routine; there’s little variation to be found there. What I like to do with the rest of the numbers is group them together, as things can get a bit messy with the small samples created by the buckets, and call them “non-routine plays.” It works out so that first baseman run about a 50% success rate on these non-routine plays. During Hosmer’s three Gold Glove years, he converted 36 of 71 non-routine plays — once again, almost exactly average. The Ryan Howards and Justin Bours of the world convert just one-third of their non-routine plays. Goldschmidt converts closer to two-thirds of his. Again, Hosmer finds himself right in the middle, paired again next to Duda.
The advanced numbers, the traditional numbers, the video scouts all say the same thing: Eric Hosmer is a perfectly ordinary defensive first baseman, likely one whose below-average range is cancelled out by a penchant for scooping balls in the dirt. How does that equate to three Gold Glove Awards? I’m not sure, and it probably shouldn’t, but my best guess it that those within the game — the managers whose votes make up 75% of the Gold Glove process — appreciate scoops more than the numbers do. Yost literally says Hosmer has “average range,” but that his scoops “[save] us countless errors, countless runs, because infielders can grab a ball, and they can turn around and throw it in the vicinity and know that Hos is gonna pick it.” What that quote tells me is that Yost believes Hosmer’s scooping ability far outweighs any range deficiencies he might have, though the numbers reveal no evidence of that being true.
As for this year? This year, thus far, is a perfect example of why we want the largest sample possible before working with defensive metrics. Every piece of evidence we have during Hosmer’s three Gold Glove years indicates he’s exactly league average, yet we see his 2016 spray chart. More than likely, Hosmer’s just had a disproportionate number of tough, diving play opportunities, and he’s converted a couple less than normal, and since the hits have piled up, so have the negative defensive runs saved. Hosmer almost certainly hasn’t been playing his best defense this year, but it’s just as likely that his bloated runs-saved total is as much a result of “facing a tough schedule,” per se. After all, we’ve got a much larger sample than just this one year, and that much larger sample points to Hosmer being perfectly average.
August used to cover the Indians for MLB and ohio.com, but now he's here and thinks writing these in the third person is weird. So you can reach me on Twitter @AugustFG_ or e-mail at august.fagerstrom@fangraphs.com.
*editor’s chair*: “That image — Hosmer lying face-down in the dirt with a ball rolling toward the left fielder ”
right fielder…. i believe is what you meant to say.
*editor’s chair pt. II*: Rustin Dodd did not write a “column,” a column in the news biz means an opinion piece. This was a “story” or, if you like, “article.”
The things you get downvoted for :: counts downvotes on this comment ::
A new dish in the Baseball Cafe: Pasta Diving Hosmer!
That defensive spray chart suggests he should audition to be a garden gnome!
Pasta Diving Jeter 4ever
Retired from the menu, but regulars know to ask for it!
Come with basket of….come with a well meaning basket.
Is there a way to quantify the value of being a “great scooping” first baseman. How many times do infielders make a riskier throw knowing that they trust Hosmer more. Not just the scooped plays but the throws that end up decent. Impossible task but not negligible to overall team defense. If you’re throwing to Giambi you probably eat more balls than if you’re throwing to Hosmer.
I truly doubt many infielders don’t make a throw based on who is at first. If they feel they have a chance at the runner, they throw it. These guys can make all kinds of acrobatic throws, as long as they can get one off they do.
I’m not sure I understand what you’re saying with all the negatives, but from watching a lot of Giants games I definitely agree with jlukebeld–it’s clear that Brandon Crawford is more likely to make a desperate attempt to throw out a runner at first if Brandon Belt is playing there than if someone else it.
As I don’t watch as many Giants games as I suspect you do, it’s not clear to me. Can you provide me with some evidence so I can believe what you’re saying?
I don’t have examples ready to send you, if that’s what you’re asking for. In any case it would take a lot of examples to prove it with different first basemen in the game. If you look at pretty much every inning of every Giants game you’ll see it and believe it because it’s just obvious, if you don’t I guess you’ll just have to decide for yourself whether to take my word for it (I would less that jlukebeld is a Royals fan who sees a lot of their games and has the same experience). Or you can find other people who watch a lot of their games (you can ask EricR here although I think he mostly listens to games on the radio). The announcers though would never say anything like this because that would be insulting to the other first person (often Buster Posey, and it’s a cardinal rule of the Giants that no one can ever say anything bad about Posey). There was one game last year where Crawford took a force out at second himself when Tomlinson was playing second instead of Panik, it was obvious that he would have thrown it to the second baseman if it had been Panik, the Giants fans were all saying that it was because Tomlinson was playing second but the announcers didn’t say a word.
Incidentally I didn’t notice Crawford making this distinction until last year, I haven’t seen Duffy make it at all, and it’s only recently that it’s started to look like maybe Panik is making it although I’m not at all sure about this. It may be something about maturity as a fielder or it may be about young players not wanting to insult Posey when he’s playing first by not trying the throw.
Jesus Montero is a great scooping DH, at least when it comes to ice cream.
Are you sure you’re he’s not your kid?
I think that scooping errant throws isn’t valueless, but that the disparity between the best and worst scoopers at 1B is probably pretty small. Much like fielding ground balls, you want to receive the long or short hop. Having your foot anchored to the bag definitely allows the 1B to have better control of which hop they receive and the issue becomes timing, which will improve rapidly with reps.
What is a good sample for defense? Cause frankly I agree. I mean Jackie Bradley and George Springer are grading out as average right now… It really starts to skew the WAR a bit. Im sorry Adam Jones in 2015 didnt just miraculously turn his defense around.
No he didn’t, he turned it around in 2014.
This is temarkably similar to giants fans discourse about brandon belt this year–they only care about his scoops, not his range or his lousy def score.
Sometimes the DEF numbers really make me wonder. Belt is at a +1 DRS and -1.8 UZR this season, but his -6.9 DEF rating is the same as David Ortiz, who has not played a game of defense this year. Belt has made a couple goofy errors this year (dropped foul pop up on a miscommunication comes to mind) but I’m having a REAL hard time believing he’s the defensive equivalent of a 100% full time DH. That makes no sense.
I really don’t understand how they’re calculated.
I’m still standing, so I’ll have another drink.
#RoyalsHatersDrinkingGame
How is scooping figured into fangraphs def rating of 1bman? I have the feeling that there is something screwy going on considering that belt has the 3rd worst def rating of any 1bman, while the other giants’ infielders, panik, crawford, and duffy, are #3, #1, and #3 in the majors at their respective positions in def, but belt clearly bails them out a huge number of times with scoops of their errant throws. Especially crawford and duffy make a lot of highlight film plays to get the ball and then throw the ball in the dirt and belt digs it out and most of the time the announcers don’t even mention what belt did–but if he didn’t they’d be getting errors. Something isn’t being evaluated right.
i think it’s driving the point that mediocrity for a first baseman is being able to at least scoop up balls. when you think about the difficulty in, reacting, ranging and setting up to throw, quickly, scooping a baseball pales in comparison to what other fielders are asked to do. i’m not saying scooping is a cakewalk, but if a first baseman can’t even do that then they are essentially worthless on defense. with that being said, the sport shouldn’t start, or continue, handing out gold medals, er gloves, for essentially doing the bare minimum that any fielder should be able to do. especially when one is put in a position which is already notorious for being the home of defensive mediocrity. in all those situations the hardest part of the play was made by the left side of infield.
But there is a big difference in the ability of first basemen to scoop out the ball. It’s clear for example that Crawford (though not necessarily Duffy or Panik) will sometimes make a decision whether to make a desperate attempt to throw out a runner at first based upon whether Belt is playing first base or someone else, the assumption being that if Belt is playing first it can’t hurt to try whereas if someone else is, it might hurt. Even if the harder part is made by someone else, if the other player makes a brilliant play to get to the ball but then throws it in the dirt so that it gets past the first baseman, the result will be even worse than if he didn’t get to the ball at all. It’s unrealistic to just ignore this and base the evaluation of the other fielder’s fielding on the basis of whether the first baseman was able to make the scoop.
i get that and it is a fair point. i just think that any extra value for that should be marginal at best. like their ability to prevent an occasional errant throw on an ‘impossible play’ should definitely be drowned out from their inability to consistently cover their fielding zone. suppose that crawford for example ranged deep and threw an off balanced throw that belt scooped for a bang-bang play at first that could have been a hit or, erroneously, an error, the fact that crawford was able to set up that photo finish is still more important and the fact that it even has the chance to be an out is still majorly the work of crawford and is still worth the assist attempt even if, say, brandon moss is at first, for the fact that converting balls in play into outs is the most important role of fielding. if suppose crawford threw the ball with absolutely no chance of even getting the runner out and the throw is wide and belt has to scoop it to prevent crawford from getting an error on a throw that should have never been attempted in the first place, that should reflect negatively on crawford’s defense (idk if those non-putout, but would-be-error plays even factor into defensive metrics) and positively on belt, but, and i don’t watch the giants regularly so i cant say for sure, those instances are probably much more infrequent.
In principle missing one grounder because of bad range should count the same as missing one scoop. In fact even that’s only true for hits down the line, which would be doubles if the first baseman doesn’t get it, just like missing a scoop will normally get the batter to second base–a missed grounder which would only be a single would be less serious than a missed scoop. The relative value of these abilities would be simply a question of the relative frequency of these events, which would depend on a lot of variables. I really don’t know about these relative frequencies.
but ranging and picking a hard hit ball is more of play than being stationary and scooping a throw. it is both more physically demanding and requires more coordination due to being a multi-aspect play. and i would venture to guess that picking a ball down the line is tougher than receiving throws because a ball probably typically has more speed coming off a bat than a defender’s arm. and even then one would have to parse which scoops are ‘routine’ and which are truly exceptional. so even though end results of failure to execute a range pick and a scoop are similar, they should be valued according to their physical requirements.
I don’t think so. This isn’t a diving contest where degree of difficulty is an issue. If scooping is incomparably easier than fielding a grounder, there won’t be any cases of one first baseman being better at one thing while another is better at another thing, so rhis discussion would be irrelevant in any case. I’m talking about how do we evaluate cases in which one of them is better at scooping while the other is better at fielding grounders and we have to decide which is a better fielder.
I have to say though that probably the more often a person looks at a game, the more likely he is to take scoops more seriously relative to range. I have the feeling that Giants fan who look at a lot of Giants baseball tend to note each time Belt saves a great play by one of the other infielders by digging their throws out and feel grateful so that they forgive him other shortcomings (some will literally say this) while Giants fans who only see highlights films which are introduced as being someone else’s great play will not notice this so much. Scoops are also more significant for team morale (‘picking up your teammate’) while would explain why the baseball people voting for Gold Gloves take scoops so seriously. But in terms of relative frequency that’s neither here nor there.
If Brandon Belt is head and shoulders above Buster Posey when it comes to scoops, it doesn’t mean that he is necessarily head and shoulders above his 1B peers.
That’s true. I just wonder how much scoops are reflected in Fangraphs Def estimates or even how they are reflected.
Have you watched the Giants this year? Belt is making many more defensive miscues this year than in 2015. That’s an observation from watching the games, but it aligns with the metrics.
Yes, I watch more or less every game. I agree 100%. It isn’t just miscues, he’s less confident or focused or something in the field. For example, last year more or less every time the other team tried a sacrifice bunt, Belt’s first thought was ‘get the lead runner’ and he usually did; this year he generally isn’t even trying. It doesn’t seem to be affecting his scoop (although last week he did miss a bounced pickoff throw) but the rest of his fielding looks clearly worse. Giants fans aren’t commenting on it, I think because they tend to be wildly polarized with regards to Belt so they won’t notice a ‘little thing’ like fielding. The only explanation I can think of is that he’s devoting all his attention this year to remaking himself as a hitter with a much lower K% and a higher avg., which he’s clearly done, and fielding has gotten lost in the shuffle.
Either that, or just maybe other 1B for other teams do a pretty good job of this as well.
Good article. I may be the only one but personally I prefer this sort of article from you rather than the pieces trying to mix Jeff Sullivan’s terse prose with Carson Cistulli’s irreverence.
You’re not the only one.
If only your jailers allowed you to skip over the Sullivan articles! #FreeWLN
One point I would make about defensive metrics:
How much practical value do they have if they are only reliable across multiple seasons?
By that I mean if we like stats because they improve our understanding of the game and the value of different players/skills, what good is a metric that only achieves those things two or three years in the rear view?
idk maybe making long term investments on players?
Yeah, I suppose that would have application for the Jason Heywards of the free agent world, but what about us as fans, watching a Thursday night game in late June?
If the error bars on defensive metrics are so large for 50 or 60 games as to only be able to only tell me that Kevin Kiermayer has been a better CF this year than, say, Nori Aoki then what value are they?
I mean, what good is it to definitively know in 2016 that Kevin Kiermayer was actually a better CF than Lorenzo Cain the previous two years?
For me, a deeper understanding of what’s signal, what’s noise, increases my enjoyment of the game.
For a fan, what good is it to know any statistics of any players?
Do you have similar misgivings about OBP (for example) if a hitter is at .330 one year, .390 the next, and then .340 the next? Just curious, we (collectively) seem to have trouble accepting outlier fielding seasons but no similar qualms accepting that an offensive performance can have outliers too.
I guess I am relying on the industry to tell me when a particular statistic stabilizes.
The consensus seems to be a matter of season(s) for defense vs a matter of months for most offensive measurements, right?
OBP is not really analogous to the defensive metrics because a lot of the variation in the defensive metrics is simply measurement error, while there is zero measurement error in OBP. To make OBP analogous you would have to mis-record batting outcomes such that the reported OBP in a given year might be .340, however the real OBP for that year would be .330 or .390, etc.
We can be confident that most of the variation in defensive measures is simply measurement error because the correlation between the different measures is extremely low (r^2 = ~ 0.4), despite trying to measure the same thing from the same outcomes.
Yes, there is not measurement error in a literal sense, but we also don’t assert that realized OBP equals talent. You have other error terms that affect the actual OBP, like BABIP and HR rate. OBP may feel like it more “accurately” reflects talent level than defensive metrics do, but you can still get situations where a hitter will run hot or cold on luck metrics.
The main point is that we accept that hitters can run hot or cold, with changes in both skill or luck, but we demand a stability in our fielding metrics that allow for no changes in luck or skill. We know to regress a huge offensive season in next year’s projection, without blaming the metric, but we often casually attribute our regressed defensive projections to measurement error.
That is not the same. OBP happened, the defensive metrics are still subjective, and are inaccurate. If defensive metrics can be taken as anything like gospel, then how is it possible that over 6006 innings, one metric thinks the Diamondbacks as a whole have been the 5th best defensive team in MLB, while another equally respected metric thinks they’ve been the 7th worst? While you might say DRS and UZR being in agreement about Hosmer means something, how do you know they’re not just both wrong? Clearly they can be wrong over large samples, because UZR and DRS can’t both be right about the Dbacks, or the Indians, much less individual players
It’s also important to be clear on what you want a particular stat to tell you. The current metrics, as I understand them at least, generally do a good job of describing what actually happened, but a single or partial season’s worth doesn’t indicate true talent levels or provide much predictive value. Taking multiple seasons’ worth of data can come a lot closer to doing that.
This, exactly. Descriptive stats tell you what happened. After a long enough time, if a guy puts up the same descriptive stats over and over, they can become predictive, but the time it takes for this to happen varies. The fact that it takes descriptive defensive stats a really long time to be predictive (as opposed to, famously, pitcher K rate) doesn’t mean it’s any less good as a descriptor. It just means we need to be careful how we use it.
Current metrics don’t actually do a good job of telling you what happened. We knows this because the various measures do a very poor job of agreeing with each other despite the fact that they are measuring the same thing from the same outcomes.
But that’s not the case with defensive metrics. What are the metrics describing about Rajai Davis, or Jean Segura? One thinks they’ve been great defensively, the other, poor. If you lived in the 1950s, and you had two newspapers that printed baseball stats, and these are the only places where you can find this info and you had one say a guy was hitting .230 and another say the same guy is currently hitting .300, how can you accept that either is doing a good job of describing what has happened? Clearly at least one of them is not, and for all you know he’s hitting .270 or anything. If DRS and UZR can be so divergent for individual players, as well as teams, it’s also possible/certain that they are wrong in the same direction at times.
Just a little bit perplexed on the Inside Edge data. Does the graph indicate that Hosmer had to deal with a much higher number of non-routine plays (for an average 1b) than the others to whom he was compared? Or was the lack of range a contributing factor in the ‘non-routineness’ of the plays? Sorry if I didn’t phrase this well enough.
Seems a disconnect between so-called advanced defensive metrics and the eyeballs of those in the game professionally.
I think a scoop is a lot like a diving play in the outfield or a great throw from the hole from an infielder; it’s the first base version of the ‘Wow! Did you see that!?’ play. It’s what sticks out, while range doesn’t and missing a double down the line can be easily credited to the hitter.
So it just seems like Hosmer’s the sort of guy who makes the crazy, highlight reel plays (as far as those can exist for first basemen) but isn’t great overall. I feel like we had a conversation about Hecchavaria at short last year with this same disconnect, and it’s the truth with lots of outfielders as well. There’s a difference between playing consistently great defense and sporadically eye-catching defense, but normally people just assume one follows the other. So Hosmer makes plays that you remember, but he doesn’t make three you don’t. And there’s your disconnect.
I think the baseline for scooping ability for first basemen in MLB is pretty high and the range in abilities quite low, making it hard to excel. Scooping a thrown ball is a skill of being mostly stationary and having good hand-eye coordination for tracking and meeting a thrown ball. This is in a sport (and position within the sport) where the primary skill is in being mostly stationary and tracking a thrown ball.
Left wondering which AL 1bmen should have won in those years. Napoli, Davis, Pujols and Moreland rated okay vis a vis UZR. Seems like slim pickings.
Raffy Palmeiro was a pretty good candidate for 2013-15; I mean, he only appeared in 28 fewer games at 1B in those seasons than he did when he won the Gold Glove in 1999.
Thank you for reading my mind! Could we get a Jay Bruce one, too?
“I think the baseline for scooping ability for first basemen in MLB is pretty high and the range in abilities quite low, making it hard to excel.”
That is exactly right. As August says in the article the difference between an average first baseman and a great or poor one is only a few runs per season. You can argue all you want about the value of “scoops” but that’s what it is.
UZR does not include scoops. DRS may, I am not sure.
Excellent article, BTW. Spot on.