The Shortstop Position: Weighing the value of defense.
- Thread starter teddywingman
- Start date
Agree with Rev...great post. On this point, however, perhaps to be more precise...it's a more a measure of what SHOULD have happened, or what was EXPECTED to happen under the circumstances.kieckeredinthehead said:
I wouldn't worry about it. No one's picking in you. We all throw out these generalities from time to time not really thinking that it's much of a big thing. That's what piqued my interest in the number in the first place. "Just what does that mean?" Or, "what does that look like from an historical POV?" Turns out, apparently, it's not so simple. :blink:teddywingman said:
I'm interested in who has this job of watching each play and judging it; and how many people are employed to do this? Anyone here?
Thanks for the post. This answers my question of why UZR/DRS are less valid than more concrete measurements.kieckeredinthehead said:
I'm confused why UZR would be scaled to an average player for a single season, given that such data may itself be unreliable. Why not scale across a six-year (or 10 yesr) span?
To be sure, FanGraphs offers these cautions regarding UZR:EricFeczko said:
- Beware of sample sizes! If a player only spent 50 innings at a position last season, it’d be a good idea not to draw too many conclusions from their UZR score over that time. Like with any defensive statistic, you should always use three years of UZR data before trying to draw any conclusions on the true talent level of a fielder.
- UZR uses Baseball Info Solutions (BIS) data in calculating its results. It’s important to note that this data is compiled by human scorers, which means that it likely includes some human error. Until FIELDF/x data gets released to the public, we are never going to have wholly accurate defensive data; human error is impossible to avoid when recording fielding locations by hand, no matter how meticulous the scorers. That said, BIS data is still the best, most accurate defensive data available at this time, so just be careful not to overstate claims of a player’s defensive prowess based solely on defensive stats.
- Since UZR is a counting statistic like RBIs or HRs, the more playing time a player accrues, the higher (or lower) their UZR will be. In order to compare players with different amounts of playing time, UZR can be scaled on a 150 game basis (UZR/150). If you want to compare a player with 90 games played to someone with 140, UZR/150 would be the way to do so.
- UZR is park-adjusted, meaning it adjusts for the fact that fielders have to deal with odd quirks in certain ballparks.
Sure. My point before (in a separate pre-merged thread) was that X-man has not played like a well-below average or non ML caliber shortstop so far (in fact he's been below-average to average, depending on what measure one uses). These cautionary notes only regard reliability, and whether the measure is predictive of ability (i.e. true-talent). As kieckeredinthehead points out, UZR/DRS has variance along individual player measures:SoxJox said:
http://www.fangraphs.com/blogs/the-fangraphs-uzr-primer/#15
Therefore, its possible for X-man to have played terrible defense so far (independent of ability), but not have it show up in any of the advanced metrics.
Reverend said:
Thanks. I don't know what the error bars are like, the UZR primer says that after a year, it's probably okay to regress halfway to zero and have a conservative estimate. However, because SS get more chances, the sample size is larger in a shorter amount of time.
Might be illustrative to look at the available data from Fangraphs. In 2013 at SS, Xander had 17 balls hit into his "zone", and made 16 plays on them. He also made plays on 5 balls out of zone. That gave him a 53.2 UZR/150 (compare to Andrelton Simmons' 24.9 during his time in the league). Meanwhile, he had 12 balls in his zone as a third baseman, made 6 of them, and made 1 play out of zone, for a UZR/150 of -47.3. Both of those numbers are absurd, and what's more they feel, on the face of it, to represent the opposite of what we observed: he was a reliable, if unspectacular, third baseman, and a rookie shortstop.
This year, Xander's had 84 balls in zone, made 65 plays (.774), and made 8 out of zone (0.09 / inning), good for a UZR/150 of -3.6 (that's split evenly between missing double plays, bad range, and errors). That's in 302.2 innings this year. Compare to Drew last year, who played 1093.1 innings at shortstop. He had 287 balls in zone, made 226 of them (.787), and got to 55 balls out of zone (0.05 / inning). Drew had a 6.7 UZR/150 last year (compared to 10.1 and 8.4 in his final two healthy seasons in Arizona). My gut is that after half a season, we can probably start to classify X into a few bins: "bad, below average, above average, or good."
As others have said, Drew is hopefully a good comp because he was really bad his first few years (-12.6, -13.9, -17.7), and then became an asset defensively. I don't know that I totally agree with the assessment that Xander's main problem has been his range. It's that he doesn't know how fast an average major leaguer gets to first, and so rushes plays that he has more time on. And he doesn't know physically what to do with himself when a play is a little off-kilter. He looks bad because he butchers a bunch of balls that he gets to, which Drew would have made comfortably.
I said this elsewhere, but Butterfield et al. mentioned that Xander is learning not just how to be a shortstop in general, but how to play on very different surfaces. I don't think it was a coincidence that there was a bunch of talk about his fielding when they were in Toronto (super fast artificial surface), basically quieted down when they were at Fenway, and is back now that they're on the road again.
EricFeczko said:
That seems like just a decision that Fangraphs made so that UZR would be comparable to league average. I agree that on the face of it, it probably makes the statistic less reliable.
For reference, Michael Lichtman ( creator of UZR) on why in small samples it doesn't measure what's happened on the field and why it's inherently different from something like batting average in that manner: http://www.fangraphs.com/blogs/the-fangraphs-uzr-primer/#15
teddywingman said:
UZR is based on data collected by Baseball Info Solutions (http://www2.baseballinfosolutions.com/)
As was sort of suggested at the end of the last X defense thread....Experience is likely to make him a better fielder. And he is likely to be a well above average hitter for the position. Given his current level of experience, isn't analysis beyond that kind of pointless?
SoxJox said:
I know you're not necessarily suggesting this, but basing an analysis on whether a particular SS made the WS seems pretty pointless. Was Ozzie a lesser SS in any given year because his team wasn't good enough overall to make it to the WS?
Since UZR can't tell us what actually happened on the field, and you need 3 years of data to draw any conclusions on a fielder's true talent level, then it seems like it's not a very useful stat, and one that can really muddy the waters if used the way most people here want to use it.
Players can change a ton over three seasons. So using info from 3 seasons prior can easily lead to things like Mike Cameron with the Red Sox. UZR said that his data from the 3 years before we signed him was relevant, but the player at that point was in the process of falling off a cliff as his body broke down, so in reality his data from 3 years before was not only irrelevant, it was misleading. Instead of "UZR says Cameron was +9 above average the past 3 years total, so he is a good defender" it was more like "UZR says Cameron was a good defender about a year or more ago, but we have no idea if he is anymore or not." Which is nice, but not helpful when deciding whether it's a good idea to offer him a two-year contract.
UZR has its uses, but there are so many limitations on it, it needs to be taken with like a hundred grains of salt. Just because something is the best science we have right now doesn't mean that it is necessarily good science.
Players can change a ton over three seasons. So using info from 3 seasons prior can easily lead to things like Mike Cameron with the Red Sox. UZR said that his data from the 3 years before we signed him was relevant, but the player at that point was in the process of falling off a cliff as his body broke down, so in reality his data from 3 years before was not only irrelevant, it was misleading. Instead of "UZR says Cameron was +9 above average the past 3 years total, so he is a good defender" it was more like "UZR says Cameron was a good defender about a year or more ago, but we have no idea if he is anymore or not." Which is nice, but not helpful when deciding whether it's a good idea to offer him a two-year contract.
UZR has its uses, but there are so many limitations on it, it needs to be taken with like a hundred grains of salt. Just because something is the best science we have right now doesn't mean that it is necessarily good science.
Agree completely Grey Eagle. That's why I think the threads we occasionally have around here that log the specific plays ourselves are very valuable.
Link to the previous thread on X's defense:
http://sonsofsamhorn.net/topic/83042-how-concened-about-xander-bogaerts-defense/
I actually haven't been watching much lately due to other priorities, so I can't say whether I've noticed him getting better from the observation perspective.
Link to the previous thread on X's defense:
http://sonsofsamhorn.net/topic/83042-how-concened-about-xander-bogaerts-defense/
I actually haven't been watching much lately due to other priorities, so I can't say whether I've noticed him getting better from the observation perspective.
Why is Pokey Reese listed there for 2004, as opposed to Nomar/Cabrera? IIRC Pokey filled in more at second.SoxJox said:
Thanks for the link, such a thread is quite valuable.Plympton91 said:
Keep in mind that unless those threads log every specific play (e.g. sharp grounder 45 feet to the right of third base, fielded by Bogaerts and thrown to first), there will be systematic bias towards outliers (i.e. really bad plays and really good ones). Such outliers may not be representative of what Bogaerts has done either.
Crowdsourcing via a poll may be more representative of what Bogaerts has done (or his present ability, depending on the phrasing used) than anything else.
chrisfont9 said:
Pokey actually led the 2004 team in innings played at SS. He played 507.2 innings while Cabrera logged 491 innings. He only logged 152 innings at 2B that season. Bellhorn led the team there with over 1000 innings.
Fangraphs also "logs" the difficulty of Bogaerts' plays this year, although I don't think they have a game-by-game breakdown.
Looking at the Inside Edge fielding stats on Fangraphs I think really drives home just how small the sample sizes are that go into UZR and fielding stats. When most of the plays are 'routine', only a handful of plays separate the best and worst SS at this point in the season. Supposedly the Inside Edge plays are judged by scouts who then assign the play a difficulty score, so you get a sort of 'scouting score', which is interesting. Just another flavor of information to add to the defensive assessment stew.
Looking at the Inside Edge fielding stats on Fangraphs I think really drives home just how small the sample sizes are that go into UZR and fielding stats. When most of the plays are 'routine', only a handful of plays separate the best and worst SS at this point in the season. Supposedly the Inside Edge plays are judged by scouts who then assign the play a difficulty score, so you get a sort of 'scouting score', which is interesting. Just another flavor of information to add to the defensive assessment stew.
Reverend said:
Even still, teddy is right.
nm - already covered, but here it is in a spoiler if you care about my drivel.
The problem is that UZR and their ilk don't record something tangible like a hit or out in the same way that AVG or OBP would. Yes, you can say "they are measuring exactly what happened on the field" but they are actually comparing players to averages and here the small data set is important, because of course it presumes that for those 30 games, XB has seen a bunch of balls that correspond to the six year rate of balls that have been hit in that direction on all fields to all players. In such a small data set, he could actually be fielding only the most difficult of those chances, or he could actually be fielding only the easiest, we don't necessarily know. As well, no matter how sophisticated they want to be, at the end of the day, all of the defensive systems rely on subjective interpretation of speed and direction and that data is lumpy (3 levels of speed, for instance). Perhaps when we can finally integrate something like hit fx and translate that to field fx, then we can really move somewhere. After that, we will have to make choices between the same lumpiness of the data and the size of each data set.
Until such time, the influence of single plays and the subjective interpretation of them subjects the result to a lot of uncertainty.
Until such time, the influence of single plays and the subjective interpretation of them subjects the result to a lot of uncertainty.
You're right. I'm not suggesting that. If you read my original post, you'll understand my motivation. I wanted to take a look at what "recording one more out per game" would take. I wanted to do that without having to analyze the entire history of every SS. So I selected a subjective sample: SSs who have gone to the WS.glennhoffmania said:
Not very SABResque, to be sure. Just a poor boy's trek down rumination lane to, you know, have a little fun looking at the numbers..
OK, I stand corrected then, thx. I guess Nomar missed a fair amount of time before the trade...Red(s)HawksFan said:
chrisfont9 said:
He had an achilles injury to start the season, perhaps from BP, perhaps from playing (gasp!) soccer. It was a hot topic around here 10 years ago.
EricFeczko said:
I agree. Every single play that Bogaerts makes, or fails to make, has an outsize chance to affect the results of the eventually sufficient sample size.
It's a bit like watching Lowrie accumulate enough chances to show that the scouts were right and the data were insufficient.
I hope it doesn't have the same outcome, because we'll have a wasted season at shortstop and a three-way logjam at 3B.
EricFeczko said:
I don't see why asking a bunch of non-experts to compare the value of Bogaerts vs. all other shortstops is really any better than using UZR or any other defensive metric. Yes, UZR is going to be awful after a month, but so are most of our opinions. Does anyone really feel like they have better predictive value in a very short sample than UZR does?
Well, collectively quite a few of us have watched him as much as anyone possibly can (this year) and probably more than the UZR guys assigned the task of scoring his opportunities. I mean to say that we might have as much of an idea, or better than a stat which is dubious to begin with--let alone the already discussed insufficient sample size.
Jnai said:
It's at the the short-term stage that observational insight should have an advantage over any data-based predictive metric. The Iggy Principle: if there's such a thing as fielding genius, it should be evident in single plays.
And besides, play-by-play dissection for footwork, anticipation and positioning is exactly what SoSH could contribute.
The evidence of fielding proficiency that I've seen of Bogaerts has only rarely been at shortstop (one bare-handed pickup coming in on a dribbler). If Manny Machado is a third baseman, Bogaerts is triply so. He's got the arm and the reflexes, but an indication of range might be his first.
The discussion about Xander’s defense, and Kiecker’s poll, really have been interesting not just with respect to Xander and the Red Sox, but in illuminating how much more resistant defense is to numerical analyses than are pitching and batting.
The ultimate goal if the defender is to prevent runs (and, secondarily, to minimize his pitcher’s pitch count, which also saves runs over the longhaul, but in a way that I don’t think is fully measured by any of the sabermetrics). But I don’t think I really understood how subjective DRS is.
This thread has looked at assists and put outs, which do seem to track pretty well with shortstops that, based on the eye-test, seem outstanding. But I’m also a bit skeptical unless we’re talking about a big sample size, or unless the question is put in some kind of ratio, like assists per chance, or something. A guy who spends 200 innings behind Randy Johnson gets fewer chances than a guy who spends 200 innings behind a sinkerballer/home run guy. Also, not all assists are the same. There’s sort of a hierarchy in assist value – throws that result in put outs of balls batted to the shortstop (either force outs on the bases or double plays on caught balls) are more notable when not made than completed. Assists for the second out on ground ball double plays seem more indicative of good defense to me. And assists that result in outs on the bases from batted balls hit to other fielders (usually as cut-off man) seem to me most noteworthy – though they can be misleading in that they can be accumulated due to great outfielders instead of great cut-off abilities, they at least say something about cut off positioning. (And, of course, a great cut-off man can actually save runs and increase his teammates' stats without ever touching the ball, which thus doesn't get measured at all.)
In any event, I thought the discussion about Xander might benefit a little by talking about the ways that a middle infielder saves runs, in rough order of my subjective guess of importance. I'm probably missing some.
1. Getting to balls in a manner that creates a reasonable chance of making an out or holding a runners to fewer bases than if the ball is not fielded; 2. Minimizing errors; 3. Speed in fielding and releasing ground balls that increases the chance of making a force out or a tag on an advancing runner, or holding a runner in scoring position; 4. Double-play turns; 5. Throwing accuracy and strength, increasing the likelihood of outs on close plays; 6. Cut-off positioning and instincts; 7. Leaping (height and timing).
The point of the exercise was that I don’t feel yet like I have much of a sense of any but number 1. My eye-test view there for Xander is -- somewhat a bit below average. Not much else to say. Kiecker and others have done a good job of unwrapping the numbers we have so far. The rest? I don’t feel like I could say, except maybe it seems like they play much more vanilla cut-off positioning this year than last, especially out of the shift. Napoli seems to take more of the cut off duties, coming across the infield on hits to left. I think the importance of number 1 probably dwarfs the others in terms of saving runs, but this exercise really made it clear to me how little I actually understand about the value of defense and measuring it.
The ultimate goal if the defender is to prevent runs (and, secondarily, to minimize his pitcher’s pitch count, which also saves runs over the longhaul, but in a way that I don’t think is fully measured by any of the sabermetrics). But I don’t think I really understood how subjective DRS is.
This thread has looked at assists and put outs, which do seem to track pretty well with shortstops that, based on the eye-test, seem outstanding. But I’m also a bit skeptical unless we’re talking about a big sample size, or unless the question is put in some kind of ratio, like assists per chance, or something. A guy who spends 200 innings behind Randy Johnson gets fewer chances than a guy who spends 200 innings behind a sinkerballer/home run guy. Also, not all assists are the same. There’s sort of a hierarchy in assist value – throws that result in put outs of balls batted to the shortstop (either force outs on the bases or double plays on caught balls) are more notable when not made than completed. Assists for the second out on ground ball double plays seem more indicative of good defense to me. And assists that result in outs on the bases from batted balls hit to other fielders (usually as cut-off man) seem to me most noteworthy – though they can be misleading in that they can be accumulated due to great outfielders instead of great cut-off abilities, they at least say something about cut off positioning. (And, of course, a great cut-off man can actually save runs and increase his teammates' stats without ever touching the ball, which thus doesn't get measured at all.)
In any event, I thought the discussion about Xander might benefit a little by talking about the ways that a middle infielder saves runs, in rough order of my subjective guess of importance. I'm probably missing some.
1. Getting to balls in a manner that creates a reasonable chance of making an out or holding a runners to fewer bases than if the ball is not fielded; 2. Minimizing errors; 3. Speed in fielding and releasing ground balls that increases the chance of making a force out or a tag on an advancing runner, or holding a runner in scoring position; 4. Double-play turns; 5. Throwing accuracy and strength, increasing the likelihood of outs on close plays; 6. Cut-off positioning and instincts; 7. Leaping (height and timing).
The point of the exercise was that I don’t feel yet like I have much of a sense of any but number 1. My eye-test view there for Xander is -- somewhat a bit below average. Not much else to say. Kiecker and others have done a good job of unwrapping the numbers we have so far. The rest? I don’t feel like I could say, except maybe it seems like they play much more vanilla cut-off positioning this year than last, especially out of the shift. Napoli seems to take more of the cut off duties, coming across the infield on hits to left. I think the importance of number 1 probably dwarfs the others in terms of saving runs, but this exercise really made it clear to me how little I actually understand about the value of defense and measuring it.
Thanks DDB--your post nudges at the idea I was thinking about when I started this thread. I must admit that really digging into this subject might be tedious.
So from my earlier exaggeration of an extra out per game, we've established that a great SS is never going to come close to anything like that, but is still involved in making the largest percentage of infield assists and general chances/touches on balls in play that may or may not result in an out.*
Texas - Bottom of 7th SCORE John Lackey pitching for Boston BOS TEX P Fielder flied out to center. 5 1 A Rios singled to center. 5 1 A Rios stole second. 5 1 M Moreland doubled to deep left, A Rios scored. 5 J Arencibia reached on infield single to shortstop. 5 L Martin struck out swinging. 5 Odor struck out swinging. 5 2
This recent scenario fortunately didn't result in any extra runs, but it sure added on to Lackey's work load. How do you measure a pitcher's workload when he's up around 100 pitches, with 1 out and runners on first and third, when there should be 2 outs and only 1 runner on base.
Now, I understand that not even Ozzie Smith makes every single play. And I probably sound like an asshole for picking on this kid Bogaerts--but believe me, I can imagine how great the pressure must be--and I'm not really picking on Xander Bogaerts. I think he's going to be a great player.
Seems like questioning the FO is unpopular these days, which is cool.
*Actually, we didn't establish that yet. I'm making the hypothesis, despite a terrible lack of data skills.
So from my earlier exaggeration of an extra out per game, we've established that a great SS is never going to come close to anything like that, but is still involved in making the largest percentage of infield assists and general chances/touches on balls in play that may or may not result in an out.*
Texas - Bottom of 7th SCORE John Lackey pitching for Boston BOS TEX P Fielder flied out to center. 5 1 A Rios singled to center. 5 1 A Rios stole second. 5 1 M Moreland doubled to deep left, A Rios scored. 5 J Arencibia reached on infield single to shortstop. 5 L Martin struck out swinging. 5 Odor struck out swinging. 5 2
This recent scenario fortunately didn't result in any extra runs, but it sure added on to Lackey's work load. How do you measure a pitcher's workload when he's up around 100 pitches, with 1 out and runners on first and third, when there should be 2 outs and only 1 runner on base.
Now, I understand that not even Ozzie Smith makes every single play. And I probably sound like an asshole for picking on this kid Bogaerts--but believe me, I can imagine how great the pressure must be--and I'm not really picking on Xander Bogaerts. I think he's going to be a great player.
Seems like questioning the FO is unpopular these days, which is cool.
*Actually, we didn't establish that yet. I'm making the hypothesis, despite a terrible lack of data skills.
Kind of OT, Is it impossible for Xander to have more value as a good 3b over a below average SS?
If he's 8 runs better with the glove at 3rd and 8 runs worse at SS, that's a 16 run swing.
How much better would he have to be at 3b over SS defensively to make up for the value lost offensively leaving SS?
If he's 8 runs better with the glove at 3rd and 8 runs worse at SS, that's a 16 run swing.
How much better would he have to be at 3b over SS defensively to make up for the value lost offensively leaving SS?
The positional difference moving from SS to 3B is -5 runs according to the formula that FG uses. You combine the positional adjustment with the UZR for the defensive component of WAR.bosox79 said: