Why Does UZR Hate Eric Hosmer's Defense?
by Jeff Zimmerman on Nov 3, 2011 11:51 AM EDT
in Kansas City Royals News
Email
PrintThe Royals brass and scouts have bragged about Eric Hosmer's defense before he was called up to the Majors. Once there, he made some beautiful plays and mastered his swinging arm scoop at 1B. After playing for a bit, defensive metrics began to be be released and the numbers had him as one of the worst defensive 1B in the league. Today I am going to look deeper into one of those metrics, UZR, and why it does not think Hosmer's fielding is so shoddy.
UZR Background
UZR (nice read up here) was created a few years back by Mitchel Lichtman (MGL). The forumula behind UZR takes BIS (Baseball Info Solutions) batted ball data to see how often a team and player makes a play. The type of hit (bunt, ground ball, liner, flyball) and the location on the field are collected. Here is the map they use to plot the data:
Using linear weight, the data is then converted into the number of runs saved compared to the league average.
Hosmer and UZR
The data that caused the uproar is the following from Fangraphs (listed are some of some of the qualified 1B from 2011):
| Name | Team | Double Play | Range | Error | Total UZR |
| Adrian Gonzalez | BOS | 0.2 | 7.1 | 3.8 | 11.1 |
| Mark Teixeira | NYY | -0.4 | 6.0 | 3.0 | 8.6 |
| Joey Votto | CIN | -1.2 | 6.5 | 2.1 | 7.4 |
| Todd Helton | COL | 0.6 | -0.3 | 3.7 | 4.0 |
| Aubrey Huff | SFG | -0.3 | 3.1 | 0.7 | 3.6 |
| Albert Pujols | STL | 0.0 | 3.3 | -0.9 | 2.4 |
| Carlos Pena | CHC | 0.1 | 1.4 | -0.8 | 0.7 |
| Paul Konerko | CHW | 1.2 | -3.4 | 1.7 | -0.5 |
| Justin Smoak | SEA | -0.2 | -0.1 | -0.5 | -0.8 |
| Miguel Cabrera | DET | 0.5 | -2.3 | -2.0 | -3.8 |
| Ryan Howard | PHI | 0.2 | -4.9 | -0.1 | -4.8 |
| Prince Fielder | MIL | 0.0 | -0.4 | -4.7 | -5.1 |
| Lyle Overbay | - - - | 0.5 | -5.5 | -0.8 | -5.8 |
| Eric Hosmer | KCR | -1.0 | -7.3 | 1.4 | -7.0 |
| Freddie Freeman | ATL | -0.3 | -13.1 | 0.8 | -12.6 |
Of the qualified 1B, he ranked 2nd to last in the league behind the defensive greats, Ryan Howard and Prince Fielder. The component stat that sicks out is his lack of range.
I went and asked MGL for the data on all the ground balls batted towards Hosmer. I wanted the ground ball dated because I wanted to look at his range and positioning on ground balls hit at 1B. MGL graciously gave me the data. The data was in the form of zone, result of the play (Out, Hit, Out by 1B) and the chances of the average 1B making the play. I took the data, lumped them into the zones and here is the data for Hosmer:
| Zone | Total GB's | Outs by Hosmer | Average Outs by IB | Hosmer Out Diff |
| R | 54 | 0 | 1.4 | -1.4 |
| S | 61 | 4 | 4.9 | -0.9 |
| T | 55 | 1 | 12.3 | -11.3 |
| U | 62 | 16 | 26.5 | -10.5 |
| V | 64 | 45 | 45.5 | -0.5 |
| W | 59 | 54 | 51.6 | 2.4 |
| X | 14 | 12 | 9.8 | 2.2 |
| Y | 9 | 0 | 1.0 | -1.0 |
Hosmer is not making enough plays on average in zones U and T compared to the average 1B. In those 2 zones, the average 1B would make 20 more outs (over 2/3rds of a game's worth) over the season. He does better near the bag with a about 5 more plays. Hosmer should probably look at taking a step or two towards 2B and see if he can still make the plays on the 1B and get to a few of the balls hit towards 2B.
I have always believed that he played too close to first base and the statistical evidence supports this claim. I am not able to take it to the next step and put video to the plays, but fellow writer Connor Moylan plans on it. He is currently looking over the plays that Hosmer was most likely to make and didn't. Also, Mr. Moylan is looking at his positioning on the field.
Hopefully you now have a better understanding of why Hosmer's UZR numbers are so low. It is probably not because he is worse fielder than Ryan Howard and Prince Fielder. Maybe it is because Hosmer does not position himself correctly on the field to maximize outs. We will have a better idea when Connor is done going over the video.
Big Thanks to MGL for the use of the data behind UZR.
121 comments
|
7 recs |
Do you like this story?
Comments
Great stuff, Jeff
It tends to confirm some of the stuff we talked about earlier this season. I look forward to Conor’s follow-up with video.
Dumb question that I’m sure has been asked and answered many times, but anyway: how much of an MLB’s first baseman’s positioning is up to him and how much up to the coaches? Anyone know anything about the Royals’ process on positioning in particular?
Making watching baseball as fun as doing your taxes.
My Twitter feed.
Before getting tweaked, read up on regression.
Like the coaching staff has enough balls to tell Jesus what to do.
- .-. ..- … – / – …. . / .—. .-. - .. . … …
by Jeff Zimmerman on Nov 3, 2011 12:12 PM EDT up reply actions
and, we don't know about the process
we trust the process™
"Things could always be worse." - Buddy Bell
They are all super Christy and have to bow down to him.
Sporadically musing on the Royals at both Royals Review and Royalscentricity, pop culture at Inconsiderate Prick, SVU at Munch My Benson and on Twitter at Old Man Duggan
by Old Man Duggan on Nov 3, 2011 7:04 PM EDT up reply actions
Rusty Kuntz was like their positional guru for a year or two when he was on the staff
I don’t know if anyone does it now. But the behind the scenes stuff they showed on RSTN a few years back showed Kuntz barking at guys all the time to change positions. I know the Brewers have been big on defensive positioning this year to make up for their – ahem – lack of skill.
Relive Royals History at royalsretro.blogspot.com
Kuntz is an outfield guy.
He was positioning infielders? I imagine Eddie Rodriguez is in charge of that stuff.
maybe Hos's low UZR in T and U is more a result of
KCs below average starting rotation and its high era and bb rate. if Hos is holding runners on base a higher percentage of the time, he would not be able to get to zones T and U (unless his name is A. Escobar) as often as he would with no runner on base.
Its my understanding that UZR attempts to adjust for the players positioning, but how well does it adjust things for a firstbaseman who is playing during an infamous “hochaver-melt-down” inning (or any inning Davies was pitching) and is stuck holding runners on for 20 minutes straight.
how does a firstbasemans UZR correlate to staff ERA and BB rate?
by DickHowser4ever on Nov 4, 2011 9:30 AM EDT up reply actions 1 recs
UZR accounts for situations where Hosmer would have been holding a runner on base
It separates out 1B fielding plays when the 1B would be holding the runner (e.g., runner on first, none on second) from when the 1B would be playing back.
Conor's follow-up
Has Conor posted his follow up? All I know after watching all the games! Hoz played well both defensively and at bat! There were many bad throws to the HOZ on many occasion and on those bad throws he still made the out. How does this take affect into the UZR ratings? Many throws that if he would have missed those catches at the time would have been counted an error on the players who threw the ball to first base. Question…do the coaches tell their players where they want them to position themselves or does the player make his own call? I for one…don’t consider Hoz to be as horrible as they predict on their data and question the fact…what games where they watching? Cause to me clearly we have different values as to HOZ playing defensively! I feel he was awesome both at bat and defensively.
by Fans-of Eric Hosmer on Nov 8, 2011 12:29 PM EST up reply actions 1 recs
Good work
I am sure this question will come up, but, yes, UZR takes into account whether there are runners on base, the number of outs, whether the batter is left-handed or right-handed, the batter’s speed, and whether there was a crazy shift on the play.
Yes, it is all taken to account with the % chance of making each play.
- .-. ..- … – / – …. . / .—. .-. - .. . … …
by Jeff Zimmerman on Nov 3, 2011 12:14 PM EDT up reply actions
Could it be that Hosmer is not getting off first base well when he is holding runners on?
If so, it might not be a positional problem, he just needs to work on getting in a good fielding postion when he is holding runners on.
Go Royals!
interesting
Kila's slash for Apr 20 to May 4, 2011, right before he was sent down: .276 / .344 / .448
by SagehenMacGyver47 on Nov 3, 2011 1:13 PM EDT up reply actions
This may be asking too much
but it sounds like there a way to determine whether those grounders in the U & T zones (which Hosmer didn’t get outs on) were hits or outs (fielded by the 2B presumably). Does Hosmer’s UZR differentiate between a hit and an out in these areas?
If the Royals 2B is making a high number of outs in zones U & T, Hosmer’s low rating is much less worrisome? Extrapolating further, could Alcides’ good range up the middle be pushing the Royals 2B more towards right field, and thereby pushing Hosmer towards the line.
Not sure if this answers your question
but UZR is set up so that players are only “penalized” for plays no one makes.In other words, a player makes cannot hurt another player’s UZR by “stealing” the play from him.
Making watching baseball as fun as doing your taxes.
My Twitter feed.
Before getting tweaked, read up on regression.
by Matt Klaassen on Nov 3, 2011 12:33 PM EDT up reply actions
Thanks, that mostly answers it
My inversion of punctuation marks, aside – i just wanted to make sure Hosmer wasn’t getting short-changed because pitchers, 2B, and RFs (Frenchy!!!!) were making unusually high amounts of plays on balls in “Hosmer’s zones”.
But then aren't you helped if your 2B makes the play, thus saving you a "no one made the play"?
or hurt if he doesn’t b/c that’s an additional “no one made it”?
I don’t want to start a big UZR discussion, but if there’s an easy answer….
Kila's slash for Apr 20 to May 4, 2011, right before he was sent down: .276 / .344 / .448
by SagehenMacGyver47 on Nov 3, 2011 1:16 PM EDT up reply actions
That's interesting
We joke about Getz not being able to make a play more than two strides to his left, but maybe this reflects it. ??
There's also an issue if you could have made a sweet play, but your awesome teammate made the sweet play instead
Nobody gets penalized, but you could have earned bonus points. It shouldn’t happen for 1B and 3B tho because they play in front of the middle infielders.
Hmm...
Maybe. Here’s one relevant section from MGL’s primer.
Although it’s just a general point you are making, I bet this is what happened with Hosmer. Mistake-free Getz + those two diving plays Giavotella made prevented Hosmer from putting Keith Hernandez to shame this year.
Making watching baseball as fun as doing your taxes.
My Twitter feed.
Before getting tweaked, read up on regression.
by Matt Klaassen on Nov 3, 2011 1:39 PM EDT up reply actions
If your 2B doesn’t make a play in the hole that another would, it makes you look bad. Presumably, if the 2B makes the play its not an opportunity for Hosmer in the above chart, right? So how many opportunities in R, S, T, U did Hosmer have relative to others? Tons of error there, but it might give you an educated guess.
cool,
I remember reading that now. They acknowledge the problem and note that there’s no easy way around it. As you suggest, the neighboring fielders are almost certainly not the entire problem – Frenchy doesn’t help and Getz had above average range per UZR, so I imagine he’s at least average to the first-base side.
It would be really interesting to try to isolate the effect of neighboring defenders’ ranges on a specific player’s range..
Kila's slash for Apr 20 to May 4, 2011, right before he was sent down: .276 / .344 / .448
by SagehenMacGyver47 on Nov 3, 2011 2:28 PM EDT up reply actions
So, Moore and Beane should trade stadiums?
The Oakland Royals + Kila
The Kansas City A’s + Mazzaro
Each takes back the homegrown rejects.
Very interesting stuff
Chaim Mattis Keller New York City's # 1 Royals fan!
it looks to me like the Gold Glove committee just used UZR for awarding the first basemen
"Things could always be worse." - Buddy Bell
all those managers and coaches
needs to get their heads out of their spreadsheets and watch a game!
Making watching baseball as fun as doing your taxes.
My Twitter feed.
Before getting tweaked, read up on regression.
by Matt Klaassen on Nov 3, 2011 12:57 PM EDT up reply actions 2 recs
Very interesting
Great work putting this together, and thanks to MGL for releasing data that I’m sure was expensive to gather.
I’m having a hard time understanding how it’s possible to square this with the claim from the Fielding Bible that Hosmer was equally bad going to his left and his right. I was under the impression that that system and UZR were based on the same underlying data. Is that wrong?
Thanks for posting that
In a quick review, I don’t see any differences that could explain how one system has him above average on the 1B side of his zone and the other has him below average to his left, but I’ll read closer to see if I’m missing something.
Fielding Bible might take positioning into account more accurately
If Hosmer is shifted more toward 1B than most, that would explain some of the difference.
If he were shifted more toward 1B, there'd only be positive zones to his left (except the one foul ball)
The only way to get the negative zones to balance to his right and left is if he was shifted toward 2B, but it sounds like all the evidence we have says he’s playing unusually close to the line, not unusually far from it.
Separately, does BIS include player position data?
There are many more articles on the two systems
They score credits/debits differently and would have a different level for average (partially in part because UZR uses more than one year of data to establish it).
RUSTY KUNTZ NOW
Great work. Someone should seriously forward this to the Royals. They may already know it but you never know.
Relive Royals History at royalsretro.blogspot.com
I'm not trying to be snarky here, but
what do anonymous scouts say about his fielding? My impression on watching him on TV is that he isn’t horrible, but I never “played the game”.
"Things could always be worse." - Buddy Bell
He certainly passes the eye test
What Jeff is showing is that any deficiencies he has seems to be due to positioning or perhaps a lack of range, which would be much harder to pick up during a TV broadcast – or even in person unless you were watching him the entire time.
Relive Royals History at royalsretro.blogspot.com
Teaser: Connor has find a little chunk of info that is a little interesting, but I will save it for him.
- .-. ..- … – / – …. . / .—. .-. - .. . … …
by Jeff Zimmerman on Nov 3, 2011 1:04 PM EDT up reply actions
This may be a dumb question...
…but the Zone-Y data confuses me.
That’s foul territory, right? Good 1B’s make outs in foul territory by catching pop-fouls. But, you’re only looking at groundballs in this post… so I’m confused how it is that the average 1B was recording an out over there.
Shouldn’t the “groundballs in foul territory” data always be zero? Am I missing something?
those on the ones on the bag and eventually roll foul
- .-. ..- … – / – …. . / .—. .-. - .. . … …
by Jeff Zimmerman on Nov 3, 2011 1:05 PM EDT up reply actions
I was about to ask the same thing.
Gentlemen, you can't fight in here. This is the war room!
by KeepItCopacetic on Nov 6, 2011 1:24 AM EDT up reply actions
The Book Blog (home of MGL) takes up the discussion
http://www.insidethebook.com/ee/index.php/site/comments/does_eric_hossmer_play_too_close_to_the_bag/
- .-. ..- … – / – …. . / .—. .-. - .. . … …
UZR is not a complete picture of defense for a 1B
As it does not take into account the 1B’s primary role, receiving throws from other fielders. This is the source of divergence between scouts and stats… Hosmer, on face, has proven to be very adept at “scoops,” or saving his teammates from errors. For a 1B – a scout would say that Hosmer is superior defensively because of this, even if his range is diminished compared to other major leaguers.
by Cleveland on Nov 3, 2011 1:14 PM EDT reply actions 1 recs
Of course, you could also use a stat to measure scooping ability.
And, Dewan lists Hosmer on this year’s Scoops leaderboard. (For those who don’t want to follow the link, he is tied for fourth on the list.)
fourth in total
what is relevant to this discussion is scoops above/below average, which Dewan lists there, too. Hosmer isn’t in the top five.
Making watching baseball as fun as doing your taxes.
My Twitter feed.
Before getting tweaked, read up on regression.
by Matt Klaassen on Nov 3, 2011 1:42 PM EDT up reply actions
At the risk of nitpicking...
…you mean that the relevant issue is “scoops-runs-saved above/below average.” And you are correct. However, in terms of raw scoops above average (which would not account for mishandled throws), Hos would still be fourth in the league… but that’s as much a measure of the number of bad throws he received as of his ability.
In my own defense, though, I wasn’t trying to be relevant to the overall discussion. I was just responding to Cleveland’s suggestions that only scouts (and not stats) can tell you anything about the receiving abilities of a 1B.
I would be curious to know Hosmer’s 2011 “scoops outs saved above average.” There’s no way it could make up for the extra 20 outs that UZR says he missed on groundballs.
I would like to see the final season numbers
as those were only through August 23.
Do you know of a public (and free!) source of these numbers?
investigations of the "primary role" of first basemen
shows very little divergence between players.
Making watching baseball as fun as doing your taxes.
My Twitter feed.
Before getting tweaked, read up on regression.
by Matt Klaassen on Nov 3, 2011 1:41 PM EDT up reply actions
that article only uses data on plays in which errors were made on the throw,
and clearly states that ‘the requisite data is not readily available’ to make an accurate judgement….Flawed.
Yes, working with the available evidence is a "flaw"
rather than a “limitation.”
Are missed scoops usually scored as something other than an error?
Making watching baseball as fun as doing your taxes.
My Twitter feed.
Before getting tweaked, read up on regression.
by Matt Klaassen on Nov 3, 2011 1:58 PM EDT up reply actions
also
I hasn’t read the article in a while, the “requisite data” phrase is about data needed for him to use it with something like UZR. But the meat of the article is about how he works around that problem using the WOWY method. That is where the results come from.
Yay mobile posting.
Making watching baseball as fun as doing your taxes.
My Twitter feed.
Before getting tweaked, read up on regression.
by Matt Klaassen on Nov 3, 2011 2:56 PM EDT via mobile up reply actions
perhaps Hosmers defense should be rated based on
how much the fielding percentage of his infielders increases with him in the lineup.
by DickHowser4ever on Nov 4, 2011 9:19 AM EDT up reply actions
% of plays made compared to avg., as a "rate" stat
R 0%
S 82%
T 8%
U 60%
V 99%
W 105%
X 122%
Y 0%
Kila's slash for Apr 20 to May 4, 2011, right before he was sent down: .276 / .344 / .448
by SagehenMacGyver47 on Nov 3, 2011 1:20 PM EDT reply actions
how about Polk points?
"Things could always be worse." - Buddy Bell
by buddyball on Nov 3, 2011 1:34 PM EDT reply actions 2 recs
Positioning Never occured to me
But this shows how far defensive stats have to come.
What defensive stat shows how many balls would have passed by Eric due to the crappiness of throws? This is one of Hosmer’s greatest strengths—he ‘saves’ so many errors.
I think we can all agree that Hosmer’s a better defender than Butler at 1b. And certainly better than Fielder. His arm’s really good too, which doesn’t seem to count in UZR.
I think "Scoops" referenced above covers a lot of that
And Hosmer does very well in that. I guess its a matter of how much do you value that in a 1B defense overall? I don’t know the answer to that.
Relive Royals History at royalsretro.blogspot.com
I'd value it highly
Think about it: every single ground ball out with no one on ends up in the 1B glove. That is a huge amount of plays, much more than ground balls near 1st.
Yea
But the average first baseman also gets like 95% of those throws. Well, those not named Billy Butler.
Relive Royals History at royalsretro.blogspot.com
95% is too high I think.
You’d be surprised. And if it is that high, that 5% is an awful lot of baserunners that didn’t advance/score.
95% is way too low.
Assuming 20 throws to first per game, your firstbaseman would have to miss one throw every game all year long. That guy would be a DH or a janitor pretty quickly. In fact, if every single error committed in the MLB this year was a firstbaseman dropping a throw, you still wouldn’t have enough errors to account for a 5% rate of failure.
The actual percentage of successful throws to first base is probably closer to 99.9%.
you could have an IF of tballers and the 1b isnt having to make 20 scoops a game...
most games will have like 10 GB outs coming to 1b at most….and the vast majority of throws arent scoops
Fire Everyone
by billybeingbilly on Nov 3, 2011 11:03 PM EDT up reply actions
Yodazilla isn't just talking about scoops...
every single ground ball out with no one on ends up in the 1B glove.
Yodazilla is talking about every single throw to first…and claiming that less than 95% of those are caught.
You’re right about the 20 throws per game, though… dunno what I was thinkin’ there. But, you can re-do the math with 10 throws per game and the point still stands. Firstbasemen do not miss 5% of the throws headed their way.
If you believe these stats, he's not better than Fielder
If you believe he’s better than Fielder, you don’t believe these stats.
Or you could say that it's interesting
but needs more data?
Glad I came, just wish I hadn't stayed so long.
Rock Chalk Talk
I guess it would only be accurate to say
If you believe he was better than Fielder this year, you don’t believe these stats.
If you believe these stats, you don’t believe he was better than Fielder this year.
More data about future years could change opinions about their overall performance levels or true talent levels.
or three
one year of defensive data showing a difference of only 1.9 runs is not enough to draw firm conclusions either way.
We're never going to get more information about how they performed this year
Also, isn’t this article limiting the analysis to the range, where there’s a 6.9 run difference between Hosmer and Fielder?
I think you are asking the wrong question
Does anyone put much value into whether Casey Kotchman outhit Prince Fielder over a two-month period?
I remember a lot of discussion on how people like Posada, Crawford, Uggla and Ibanez hit the first few months
and what that said about their hitting ability
I don’t know if anyone would bother to make the argument the Mark Trumbo was a better hitter than Dan Uggla over the first few months, but I’m sure every subjective estimation and hitting metric would be in agreement.
Also, I know 3 ears of data is equivalent to a full season of hitting data, but I always just though that was a way of comparing sample sizes. Does 600 opportunities have some significance as a sample size that I’m not aware of. When we use a single year of defensive metrics in calculating WAR and in writing defensive guides like the Fielding Bible, I got the impression that it was at least somewhat OK to draw conclusions from a single season of defensive data.
The larger the sample size is obviously going to provide more accurate results
For defensive stats, three years is the generally recognized sample size before you start making conclusions about true talent level. Over a full season, 1B generally get about 200-250 opportunities to field grounders, so 600 would represent roughly three seasons.
The larger sample size reduces the noise in the numbers. Just like with hitters, players are going to have random years where their fielding numbers do not necessarily match up with their talent. But because the reliability of fielding metrics accrue at a slower rate than hitting (where three years of fielding approximates one year of hitting), the effects are going to be more pronounced in any given single season.
Some years, a player might not get to a few balls that he normally would for any variety of reasons — he had a nagging injury for a month, he let himself get out of shape at the end of the year, he lost focus during his divorce. And in some years, the distrubtion of batted balls is going to vary widely — he might get an unlucky number of “bad bounces,” or he might not get as many tough plays where he can pick up credits as the previous year. The noise from these things, however, gets reduced when looking at more than one year.
MGL, the guy who developed UZR, suggests that when looking at any single year, you want to regress it by half toward zero.
A single season of WAR uses a single season of UZR because that is the time period arbitrarily selected. It is the best estimate of the player’s defensive value during that period, even if it does not necessarily represent his true talent or is as accurate as his hitting stats. That is why in all of the articles about using WAR in discussing MVP candidates, the authors often go out of their way to note that the defensive component is not nearly precise as the batting component, so when you have two guys with similar WAR, you should favor the guy with the better hitting component.
Gopher, I appreciate the exchange
I don’t think I’ve ever had a debate with you on this board where, by the end, I felt like I had the better end of the argument. I always learn something and usually change my mind. Thanks for taking the time today to make several lengthy posts to respond to my questions and provocations.
So there's nothing really special about 600 opportunities?
Except that’s a scale that we’re used to working with in baseball so it’s familiar? I’m asking because I know that some other sample sizes (like 30) represent key inflection points in statistics where the size has gotten large enough that we’re able to make some assumptions about the quality of the data.
No
the guys much smarter than myself have said that 3 years of defensive data is enough of a sample size to safely make assumptions about the data without the need for further regression. I think the 3 years of defense = 1 year of hitting came later as a way to explain the difference on how long it takes for the defensive data to become, for lack of a better word, reliable. That is not to say that less 3 years of data is useless, just that you would need to regress it more (see the example about taking one year numbers and halving them).
The 600 number comes from me based solely on looking at how many opportunities 1B get each year according to the BIS data, seeing that the range is frequently between 200 and 250, and extrapolating it out. The number is different for each position — 2B is closer to 350-400.
except positioning matters too
This particular defensive stat attempts to show how Hosmer performed overall in regards to range. It is not attempting to answer why he performed at that level. It might be — and probably is — largely due to positioning, but that is a separate question.
UZR includes components for errors (penalizing those that make more errors while fielding throws than the average and rewarding those who do not) and ability to turn the double play (which is 90% of the time that a 1B’s arm matters).
could the young guys being at the bottom just show how much experience matters at 1B
if reaction time is a top priority, positioning has to be as well. for the young guys who haven’t been in the league as long, they don’t know the hitters tendencies as well as the veterans do. because of this, the veterans have a better idea on where to position themselves when even a foot or two of space could be a big advantage against a line drive or sharply hit grounder. veteran savvy sounds stupid, but it probably plays a factor at 1B.
I think you could be on the right track.
A step or two can make a huge difference in fielding a ball at the corner. Not just because of more time but getting yourself in a better position to play a ball off the turf.
Glad I came, just wish I hadn't stayed so long.
Rock Chalk Talk
I think you are right
I don’t even think I would have noticed him play close to 1B, except he seemed to never have to throw the ball the pitcher covering. He always just took the ball himself.
- .-. ..- … – / – …. . / .—. .-. - .. . … …
by Jeff Zimmerman on Nov 3, 2011 2:43 PM EDT up reply actions
I think that's the only possible explanation
But I don’t understand why 1B would show a different age peak for the aging curve than we see at other positions, where players are in the decline phase for their entire time in the majors.
it is a bad idea to draw conclusions
just from two young guys ended up at the bottom in one given year. The other rookie, Mike Trumbo, was fourth best, and in the previous year, rookie Ike Davis was second best. And one season of defensive data is a small sample size that should not be given too much weight.
not that it means much, but hosmer/freeman were both 21 this year, while trumbo was 25 in his rookie year and davis 23. i think added experience is somewhat a factor when it comes to 1B. we’ll just have to see how it plays out
I don't disagree with you that the learning curve plays a big factor at 1B
The guy who developed UZR has noted that 1B does not show the same aging curve as the other positions where speed is more important.
I just don’t think you can draw conclusions about these two young players in particular just because they ended up at the bottom of the league standings in one year.
i'm not trying to draw defined conclusions, because its not possible yet
just throwing things out there that may be a factor in their sub par numbers right now. the learning curve could be a factor, or they both could be below average at 1B defensively.
Thanks for posting this
It does clear some things up from the back-and-forth I had with Gopherballs yesterday. I’m still reading thru all of the stuff, so I have more work to do. Right now, I’m at the point of: Out Differential by Zone {magic happens} then you get the Range component of the UZR. I’m guessing that each zone has a different conversion factor to turn these outs above or below “average” into a number for range?
Also, I’m glad to see that the system doesn’t presume that an “average” 1B will convert GBs into outs 100% of the time in any of these zones. “Average” never equals perfect.
Lastly, about the “Error” component of the UZR – I guess I haven’t reached this point in my reading yet. Is there a quick and easy explanation of this number?
Tension is the enemy. - Charlie Lau
by aHorseWithNoName on Nov 3, 2011 3:31 PM EDT reply actions
please keep in mind I perhaps oversimplified how UZR is done in yesterday's discussion
the information in the links is what you should go on
Yeah
I’m starting to get that. Also, from my previous post, it should be noted that I am aware that more than Out Differential by Zone (on GBs) goes into making the Range piece of UZR (I just included that in “magic happens” because I don’t know what those things are…yet).
Tension is the enemy. - Charlie Lau
by aHorseWithNoName on Nov 3, 2011 3:53 PM EDT up reply actions
I thought he was positioned badly this year
and the fielding metrics seem to as well. (Doesn’t DRS use video?) He seems to be reasonably athletic and does not get a terribly slow read on the ball. His famous “swipe scoops” cost him more than they helped him I would bet. I can remember at least 3 occasions where he made the swiping scoop-when the throw was not bad enough that he had to do it- and scooped the ball right into the infield. I am sure there must have been more.
"Trying is the first step to sucking" -Jimmy Chance
Just for reference
Swiped from a comment from another thread, here is the three-year UZR/150 leaderboard for 1B (minimum 2500 innings plus Billy Butler at 2445):
A. Gonzalez +5.5 runs
Kotchman +4.1
LaRoche +3.9
Loney +3.2
D. Lee +2.9
Votto +2.8
Sanchez +2.4
Teixeira +2.1
Huff +1.9
Pujols +1.9
Helton +0.6
Overbay -1.7
M. Cabrerra -2.1
C. Pena -2.9
Fielder -3.7
Konerko -4.9
Butler -5.1
Howard -5.5
None of these look unreasonable to me.
remember this for only "old man" Teixeira
His career UZR/150 is +5.4.
Would a team
Want their first baseman positioned close to the line if the had a RF with poor range and a great arm? I.e., is he positioned there to complement Francouer? Shouldn’t these numbers be in context of the whole defense? Would the team have been better off with him missing more balls on the line and getting more in the hole? How much better?
by billexgordler on Nov 3, 2011 6:56 PM EDT via mobile reply actions
good question
I wonder if Frenchy was telling Hos where to position himself (like Pujols calling his own hit & runs) so that he (Frenchy) could get more opportunities to throw out base runners? (I’m being sarcastic of course)
Tension is the enemy. - Charlie Lau
by aHorseWithNoName on Nov 3, 2011 7:13 PM EDT up reply actions
Perhaps he needs to be closer to the bag
because hitters tend to drill more balls down the line off our SPs and their 84mph “Fastballs”
This is my signature line. It is full of awesome and win.
Actually, what about holding runners on and playing behind runners?
The Royals staff had a high BABIP and walked a ton of guys leading to more baserunners. More baserunners means that Hosmer has to hold players on and play behind more runners then an average first baseman, So that would have an affect on his range right?
Go Royals!
UZR adjusts for baserunners
If there is a runner on first, Hosmer is not getting compared to plays made by others with no runner on first.
What method does UZR use to ajust to playing behind runners on first as opposed to holding them on?
Seems like it would be pretty tricky to be able to make those ajustment without quite a bit of error. It seems to me that first base positioning and the small sample size could make using UZR as a predictive measure for first base to be pretty problematic.
Here is a player I chose at random to look at.
Here is Mark Teixeira’s UZR
11.5, 10.4, 3.3, 2.2, -1.6, 15.3, 0.6, -2.9, 8.6
He has a three year period where his UZR average is 1.3 and three years around that his average UZR is 12.4
Even a three years data at a time wildly fluctuates. A players skills should diminish over time, but with UZR it seems to be more random then anything else. You would be better off just predicting everyone as average, and just have UZR be the descriptor of that SSS of that particular season. It seems pretty worthless as far as a true talent predictor, even three years at a time.
Go Royals!
UZR uses 6 years of data
so it is comparing a player on a certain play to every time it happened to any player at the same position in any game during those 6 years. Sample size is not an issue.
UZR adjusts for baserunners and outs — with less than two outs, a runner on first, and no one on second, UZR expects that the 1B will be holding the runner. How often does a 1B not hold the runner in that situation? 1%? It is going to make neglible difference, if any.
Here are Teixeira’s UZR/150 for each three-year period of his career. They are pretty far from “wildly fluctuat[ing].”
2003-05 +9.5
2004-06 +4.9
2005-07 +1.4
2006-08 +5.2
2007-09 +4.9
2008-10 +4.1
2009-11 +2.1
Not that you will read it (or care), MGL specifically addressed the issue in the UZR Primer at Fangraphs:
Consistency from Year to Year
Another issue that often comes up is the significance of consistent or inconsistent year-to-year UZR’s. It is human nature to look at Player A, who has a UZR of -11, +14, and -3 over the last 3 years, and be confused and unsure of that player’s true defensive skills. At the same time, we see Player B, who is +1, 0, and -1, and we feel confident that he is an average defender. Don’t be fooled by these illusions. There is virtually no difference between the two players’ stats. The fact of the matter is that both players have a 3-year UZR average (albeit non-weighted) of zero, and therefore both players are likely around average defenders. It makes very little difference what those yearly samples look like. They are merely 3 arbitrary sub-sets of a 3-year sample of data. Keep in mind I didn’t say that it makes no difference – only that it makes very little difference. If you are uncomfortable treating both of those players the same, then you can weight those 3 years (which is correct to do anyway) using a weighting scheme such as 3/4/5, and then you can ignore those year-to-year "inconsistencies."
If you want to see real wild flunctuation, take a look at just about any hitter’s OPS or wOBA from month to month. That is basically what you are doing by looking only at one year of UZR data.
Well, I looked at his wOBA, which has a difference of about .175 points from worst to first, and
there was only one two month period in 2005 that had a difference like his UZR (giving UZR a difference of 30 from worst to first) had from year to year. However, interestingly enough, the data was pretty simular to month to month differences in his wOBA. When looking at the three year differences in UZR verses the year to year data of wOBA the differences in percentage was not that far off as percentages go. Although the wOBA data followed ageing curves while the UZR data isn’t seeming to follow it nearly as well. How many plays are made by a first baseman? Maybe that is why there are a lot of fluxuations, because the data is really not the same as two months but more like 6 weeks of data for a first baseman and you need to take data in four year sets to get it the same as batting stats?
Go Royals!
Now convert to runs like UZR
Even looking at Teixeira’s year to year offense, you still get wider variance than his 3-year defensive numbers.
Teixeira’s Batting Runs Above Average by Year
2003 4.2
2004 25.7
2005 44.0
2006 22.6
2007 35.2
2008 47.2
2009 40.3
2010 24.1
2011 21.8
So on offense, his best offensive year is better than his worst offensive year by 43 runs. Even in what should be his peak years at ages 26 to 30 (2006 to 2010), there is still a difference of 25 runs between his best and worst offensive years, and a year to year swing of at least 10 runs in all but one circumstance (2008 to 2009 when it was 6.9 runs).
For Teixeira’s 3-year defensive numbers, the difference between his best and his worst is only 8 runs, and the period to period swing is never more than 4.6 runs (and usually no more than a couple runs).
UZR has limitations, but calling it worthless merely because of year to year fluctuations is silly.
Comparing runs for batting and runs for defense is comparing apples to oranges.
It is the percentage of run difference in the range of both UZR and batting runs that you would have to convert to in order to compare them. I was looking at the percentage change, and Teixeira’s wOBA was actually tighter percentage wise for two month stretches then the year to year UZR data. But Teixeira may well be an outlier and have his data closer together then an average first baseman. When you went a month of batting data to one year of UZR then it was about the same percentage, so It looked like a year of UZR data may not have been as much data as two months of hitting data. I don’t know if it is measurement error, error or data points, an abberation for Teixeira, or some combination. If you take three years of UZR data and measure it next to year to year hitting data, it seemed to be in line. Of course, when you average data you tend to eliminate some error, as whatever measuring error was done one year could be reversed the next year.
What I don’t think that UZR is good as predictor of UZR and wOBA is at predicting wOBA for the next season. It seems to vary more wildly then wOBA for a single year of data (which you would expect with less data), and does not seem to follow ageing curves very well either. What would you project Teixeira’s WAR to be next year? It seems to be that in using UZR for projections, you should not include it with hitting projections, but instead include it separetly with its own greater error range for 50% percentile, 60% percentile and so forth. Also because UZR has such a different level of uncertainty, you should regress a players UZR a lot more to the mean then hiting data when doing projections, or when saying how much a player is worth as far as true talent wise at least until you have your three years of data. With Hosmer for example, you would lightly pencil in his true talent at defense at -3.5 and wait for more data.
Go Royals!
Comparing (batting) runs to (defensive) runs is comparing apples to apples
Runs are runs.
Regressing one-year of UZR data by half was discussed above.
I just use the Eye Test
From watching the games I can tell that he is a good defender with great hand-eye coordination, I don’t need a metric to tell me otherwise. So I would definitely agree with the out of position hypothesis on this one, seems like the most reasonable explanation. Maybe just taking an extra step towards second and trusting the pitcher on balls to his right should be in order for this offseason.
Comments For This Post Are Closed
Secondary Sidebar
User Tools
Related FanPosts on Royals Review
FanPosts
Community blog posts and discussion.
Recommended FanPosts
Recent FanPosts
In Association With
Big League Stew - MLB - Yahoo! Sports
- College player makes impossible over the fence catch to save tournament game (Video)
- Security!: Camden Yards guard collides with Royals’ Mike Moustakas
- The Juice: Nelson Cruz matches career high eight RBIs, Pujols stays hot in Seattle
- Benches clear at Fenway as Red Sox-Rays rivalry heats up
- Keeping good company: Albert Pujols becomes fourth-youngest player to hit 450 home runs
