Impact of Defense

[TheLeviathan]

 

Also, I am pretty sure that very very few people involved in sabermetrics think they are precisely or perfectly measuring anything.  It's widely acknowledged that the only precise, perfect measurement is the one that shows up on the scoreboard; sabermetrics is all about estimating the myriad factors that go into that simple number.

 

Please, drop the strawman.  The problem is not what they believe, but in how they communicate it so that others don't believe that they are.  

 

I never insinuated anyone believes that, not even close.  In fact, I made the exact opposite point about their intentions, my point was about their delivery confusing their intentions to some that might read them.  I think that issue is one of the main problems in why sabermetrics believers versus old school people continue to talk past each other.

[Otto von Ballpark]

 

 

What IS UZR trying to do?

UZR estimates a player's ability to convert opportunities at his position into outs, and compares it to his peers.

 

Criticizing it for not spitting out virtually the same value every year for a player is misunderstanding what it does.  It's not trying to show "true talent" levels, which is basically impossible to do in anything more complex than perhaps a 40 yard dash.

 

Criticizing it for not correlating great with runs on the scoreboard also misunderstands what it does.  It's not trying to be Win Probability Added for defense (a highlight reel catch with the bases loaded does not demonstrate more ability in the defender than the same catch with the bases empty).

 

And pointing out that its highest and lowest estimates over a given sample might represent outliers doesn't invalidate the metric.

[Otto von Ballpark]

 

 

Straight from Fangraphs:

 

In general, UZR isn’t perfect because it doesn’t factor in shifts, positioning, and can’t perfectly measure everything it needs to, but it’s still among the best options out there.

 

So, please, take your issue up with them and their phrasing.

IT LOOKS AT THE RESULT OF THE PLAY, WHICH INHERENTLY INCLUDES POSITIONING AND SHIFTS.

 

It doesn't parse out the credit/blame to shifts or positioning.

 

How is this so hard to understand?

 

If they actually had a way to remove the effects of shifts and positioning from the metric, that would actually be a very valuable thing!!!  That's what is hoped to do with FieldFX type data.  The end total value (player speed + reaction time + positioning/shift + whatever), in context neutral run form, will probably be somewhat close to the UZR estimates, though.

[TheLeviathan]

 

IT LOOKS AT THE RESULT OF THE PLAY, WHICH INHERENTLY INCLUDES POSITIONING AND SHIFTS.

 

I merely used the same exact phrase that the experts at Fangraphs used to describe the stat.  You should take your issue up with them. 

[Otto von Ballpark]

 

 

Please, drop the strawman.  The problem is not what they believe, but in how they communicate it so that others don't believe that they are.  

 

I never insinuated anyone believes that, not even close.  In fact, I made the exact opposite point about their intentions, my point was about their delivery confusing their intentions to some that might read them.  I think that issue is one of the main problems in why sabermetrics believers versus old school people continue to talk past each other.

So you want every internet post about the metric to include some kind of disclaimer?  That seems pretty unnecessary, especially since you admit being aware that no one actually believes these metrics are precise/perfect.

[TheLeviathan]

 

So you want every internet post about the metric to include some kind of disclaimer?  That seems pretty unnecessary, especially since you admit being aware that no one actually believes these metrics are precise/perfect.

 

They are aware of it and over time looking at the stats I am aware of it, but to those that are being introduced to it or are new to interacting with it, it is not so clear to them.  I would imagine the proponents of this work want to expand the appreciation of it, so less confusing presentation may be helpful.  Rather than continually telling people they don't "get it" or they are "reading it wrong" - perhaps attack that problem on the front end instead?

 

I feel like sometimes these stats are talked about in the same way that one might talk about OPS or batting average, when that isn't fair.  (To either statistic's strengths or weaknesses quite frankly)  Some effort to make sure these approximations are read as approximations would go a long way to helping these stats be accepted and properly interpreted IMO.

[Otto von Ballpark]

 

 

I merely used the same exact phrase that the experts at Fangraphs used to describe the stat.  You should take your issue up with them. 

Levi, last time:  they are saying "factor in positioning" as in "filter/remove/isolate the factor of positioning".  I think that's a valid use of the phrase, if one is honestly attempting to consider what they are saying..

 

Please think about your interpretation.  If the UZR estimate actually did not include the effect of shifts or positioning, it would actually be a BETTER metric in many ways than the actual UZR.  They are admitting that it is a weakness of the metric that they can't consider that factor for removal.

 

I am not going to log a complaint with Fangraphs due to your stubborness to consider what they are saying.

[TheLeviathan]

 

Levi, last time:  they are saying "factor in positioning" as in "filter/remove/isolate the factor of positioning".  I think that's a valid use of the phrase, if one is honestly attempting to consider what they are saying..

 

Please think about your interpretation.  If the UZR estimate actually did not include the effect of shifts or positioning, it would actually be a BETTER metric in many ways than the actual UZR.  They are admitting that it is a weakness of the metric that they can't consider that factor for removal.

 

I am not going to log a complaint with Fangraphs due to your stubborness to consider what they are saying.

 

I'm not being stubborn, it's part of my point.  You continually deride anyone with a different viewpoint for a lack of understanding of the statistic, but the very site employing it is less than clear about the distinction you are drawing.  In fact, I used the exact same words to describe a problem I saw as Fangraphs and you blew a gasket over misunderstanding.

 

Sort of speaks to my point of a better presentation of strengths and weaknesses being an area of improvement for sabermetrics.  I'm sure you'll disagree, but you're on the side of it that doesn't see the need for change because you THINK it's clear when it perhaps is not as much as you believe it to be.

[Otto von Ballpark]

 

 

They are aware of it and over time looking at the stats I am aware of it, but to those that are being introduced to it or are new to interacting with it, it is not so clear to them.  I would imagine the proponents of this work want to expand the appreciation of it, so less confusing presentation may be helpful.  Rather than continually telling people they don't "get it" or they are "reading it wrong" - perhaps attack that problem on the front end instead?

 

I feel like sometimes these stats are talked about in the same way that one might talk about OPS or batting average, when that isn't fair.  (To either statistic's strengths or weaknesses quite frankly)  Some effort to make sure these approximations are read as approximations would go a long way to helping these stats be accepted and properly interpreted IMO.

If anyone reads an article about stats such as "wins above average/replacement" or "runs about average/replacement", which are by name and definition theoretical constructs and quite obviously not anything that exists on a baseball scorecard, and assumes they are perfect or precise or immediately corroborated by the simple records on said scorecard, is demonstrating some level of preconceived hostility to the endeavor, I think.  Perhaps borne out of honest skepticism, but still, that's just as much an issue with the reader as the author.

 

If you are looking for an admission that defensive estimates are harder to calculate and predict than offensive estimates, I've seen plenty of such statements at Fangraphs and B-Ref.  Discussion about how to improve defensive metrics is rampant at baseball sites, even in these very threads.  That such admissions don't appear every time a defensive metric is invoked does not suggest to me a serious issue with the SABR community.

 

And I say that as someone who can't stand certain baseball writers.  But generally assumed precision/perfection is not the reason.

[jay]

What IS UZR trying to do?

 

In the most simplest sense, UZR is trying to capture what happened.

 

Just like with offensive stats, what happened may not line up with what you'd expect based on a player's true talent level. For example, look at some of the short-term HR binges in more recent years by a Valencia or a Plouffe.

 

A defensive measure like UZR can tell us that player made a series of crazy plays that had a lot of value. That doesn't mean he'll continue to do that or, to spy's point, even have the opportunity to do so.

 

That summarizes why UZR doesn't capture "true talent level" without significant data samples. Although, to be fair, most commonly accepted offensive stats need good sized samples as well in order to become predictive.

[Otto von Ballpark]

 

 

I'm not being stubborn, it's part of my point.  You continually deride anyone with a different viewpoint for a lack of understanding of the statistic, but the very site employing it is less than clear about the distinction you are drawing.  In fact, I used the exact same words to describe a problem I saw as Fangraphs and you blew a gasket over misunderstanding.

 

Sort of speaks to my point of a better presentation of strengths and weaknesses being an area of improvement for sabermetrics.  I'm sure you'll disagree, but you're on the side of it that doesn't see the need for change because you THINK it's clear when it perhaps is not as much as you believe it to be.

Where don't I see a need for change?  I welcome our FieldFX overlords.  I hope they tweak UZR as they can, and provide more detail about its calculation.  Those would be awesome things.

 

I'm not deriding anyone for lack of understanding.  Admittedly I probably am deriding folks for not so thinly veiled hostility to the idea of present-day run estimators...

 

If you see an outlier, it's OK to say it's an outlier.  I get a little testy when instead an outlier is called "absurd" and suggested (or flatly stated, Chief ) that the metric is thus untrustworthy.  And I really roll my eyes when it leads to "that's the problem with these SABR types."

 

Let's honestly ask: how big a part of the game is defense?  And discuss.  Without preconceived notions of how bad current metrics are at helping answer the question.

[jay]

A defensive measure like UZR can tell us that player made a series of crazy plays that had a lot of value.

To expand on this part a little further, UZR is based on averages from that player's position. Blatant shifts are excluded from the calculation, but if a LF is generally better positioned based on the batter than other LFers... he's going to have an advantage compared to the "average" LFer that is used for the calculations around what you'd expect the guy at that position to catch. (i.e., could explain a 2014 Alex Gordon)

 

Just because he was better positioned in 2014 and had the chance to make some plays that the "average LFer" would've had no chance at, doesn't necessarily mean that's his true talent level. He could be way better than the other guys at his position, could be in a better position to make the play (excluding categorized shifts) or some combination.

 

Simply, it's what happened compared to average. If he saved 20 runs compared to average, we shouldn't penalize him based on "how" it happened just as we wouldn't penalize the utility guy who miraculously got pitches in his wheelhouse and hit 20 homers.

[jimmer]

 

Simply, it's what happened compared to average. If he saved 20 runs compared to average, we shouldn't penalize him based on "how" it happened just as we wouldn't penalize the utility guy who miraculously got pitches in his wheelhouse and hit 20 homers.

and when you expand it out to team rankings, the team should get credit for the shift as well.

[jay]

and when you expand it out to team rankings, the team should get credit for the shift as well.

While I'd agree with that statement, to be clear, an individual player's UZR and DRS scores specifically exclude plays where the defensive scorer recognizes a shift. That's doesn't mean he (or she) captures a step or three to the right direction as a shift.

[jimmer]

 

While I'd agree with that statement, to be clear, an individual player's UZR and DRS scores specifically exclude plays where the defensive scorer recognizes a shift. That's doesn't mean he (or she) captures a step or three to the right direction as a shift.

it's why I said should :-)

 

sometimes a shift is subtle and all one needs is an extra step or three :-)

 

'Positioning, reaction, and range all factor into a player’s defensive ability. Whether an outfielder made a routine catch because he was positioned well or he had to sprint twice as hard because he got a poor read on the ball off the bat, the out still counts the same. The Plus/Minus System doesn’t know why the player made the play; it just knows that he did (or that he didn’t). As a result, the best plays according to the Plus/Minus System aren’t always flashy Web Gem nominees. As with anything else, the best defensive players can make the hard plays look easy.'

 

'Just like with offensive stats, players can have “fluke” months or seasons which don’t reflect his actual ability. It’s important to remember that all statistics—whether it’s batting average, Runs Saved, or an election poll—aren’t always indicative of true value: they’re just estimates within some amount of measurement error. Additionally, players can have uncharacteristically good or bad days, weeks, or months which have little effect on what we think of them going forward. When Mark Teixeira is hitting under .200 at the end of April, you write it off as a bad month and hope that he recovers soon (if you’re a Yankees fan, that is). Players can have seemingly good and bad defensive months and seasons too.'

[USAFChief]

If you see an outlier, it's OK to say it's an outlier.  I get a little testy when instead an outlier is called "absurd" and suggested (or flatly stated, Chief ) that the metric is thus untrustworthy.  And I really roll my eyes when it leads to "that's the problem with these SABR types."

 

Let's honestly ask: how big a part of the game is defense?  And discuss.  Without preconceived notions of how bad current metrics are at helping answer the question.

I think it might be fair to say all sides of this discussion have some preconceived notions about defensive metrics.

[USAFChief]

 

'Just like with offensive stats, players can have “fluke” months or seasons which don’t reflect his actual ability. '

I would, in fact, cite this as a "preconceived notion."

 

The statement is not a "fact." It is the opinion of the author. How does he know this to be factual? He only knows that the measurement system he designed, or advocates, sometimes shows wild fluctuations.

 

It seems to me the default answer for those fluctuations shouldn't be "it can't be the way we're measuring, so defensive performance must vary like offensive performance."

[jay]

How does he know this to be factual? He only knows that the measurement system he designed, or advocates, sometimes shows wild fluctuations.

But isn't this true of even the most well established offensive statistics?

 

Why would we hold defensive metrics to a higher (perhaps, unreasonable) standard?

 

Why would we accept HRs as a stat when Chris Davis can go from 53 in 2013 to a flop of 2014?

[Riverbrian]

In my opinion... I applaud the efforts but most defensive metrics are half cooked in my opinion... Some day... They will be cooked and I wait for that day.

 

On the subject of defense... I believe it's huge but really hard to quantify. It's all about the extra out or stolen outs and isolated moments in games and the crooked number that kills teams.

 

One ball not caught can be the difference between hanging a zero for an inning and allowing five in an inning.

 

When a pitcher is struggling and our pitchers have struggled a lot. Lorenzo Cain running one down that Shin Soo Choo Couldnt run down or Nolen Arenado making a sick play can provide that one out the struggling pitcher needed to escape any inning with average stuff.

 

On the reverse side.... Delmon Young spinning around like a confused tourist can give the opposing team one more chance to find the seats or lace a double down the line or both. It can and has led to the crooked number and ball games can end on a crooked number in the 4th inning.

 

I've seen it too many times... Most crooked numbers have someone screwing up defensively somewhere and it doesn't always show up with a big red E in the box score. And if Gibson gives up 4 in an inning because Plouffe didn't move his feet to the left to get him out of the inning.. It's going to take a few innings of scoreless ball to stabilize that ERA.

 

Make the play and you can sit down unscathed.

 

Defensively... I like Dozier... I think Mauer was very solid 2nd half of the season. And that's about it. The rest of the team is average to below average. As a whole...mthe team doesn't have the defensive skill to help our pitchers who need the help.

 

Braves pitchers love Simmons and they all know that they have lower ERA's because of Simmons.

[Riverbrian]

But isn't this true of even the most well established offensive statistics?

Why would we hold defensive metrics to a higher (perhaps, unreasonable) standard?

Why would we accept HRs as a stat when Chris Davis can go from 53 in 2013 to a flop of 2014?

The question is why is UZR fluctuating wildly. It isn't about a fielder getting hot or cold.

 

UZR has to sift through routine play after routine play.

 

Ball hit into Zone 8D -- Caught

Ball hit into Zone 8D -- Caught

Ball hit into Zone 8D -- Caught

Repeat this 35 more times and pretty soon Zone 8D has an out percentage of 98%.

 

And then

Ball into Zone 8D -- Not Caught

 

UZR will then factor in that balls in zone 8D are caught 98% of the time and weight it accordingly.

 

That one ball not caught because a player lost the ball in the sun or was shifted over into Zone 7A and he falls into a huge UZR hole that is impossible to climb back out of.

 

Why is it hard to climb out of it?

 

Because he only get a couple of chances a game and those chances are... Routine flyball into Zone 8 -- Caught... Just like everyone else is doing.

 

UZR is a bad stat because it penalizes one mistake or one bad shift to harshly.

 

Defense is the exact opposite of hitting stats. A hitter gets a hit 3 times out of ten... A fielder makes the play 9 times out 10.

 

If the fielder makes a play 9 times out of ten... It's going to take a lot of chances to correct that one time when it didnt happen.

 

And that's why UZR fluctuates like a ping pong ball on steroids.

[USAFChief]

But isn't this true of even the most well established offensive statistics?

Why would we hold defensive metrics to a higher (perhaps, unreasonable) standard?

Why would we accept HRs as a stat when Chris Davis can go from 53 in 2013 to a flop of 2014?

Because we can be absolutely positive it's not the measurement system?

[USAFChief]

Honest question, because I don't know the answer:

 

Lets say Arcia gets 250 lazy fly balls hit to him in left this year, and every one is hit right to him. He hardly has to move all year, and catches all 250.

 

What would his UZR be?

[Riverbrian]

Honest question, because I don't know the answer:

Lets say Arcia gets 250 lazy fly balls hit to him in left this year, and every one is hit right to him. He hardly has to move all year, and catches all 250.

What would his UZR be?

If all 250 lazy fly balls where in the same high percentage zone. He'd be above zero but not by a lot. If he was standing in a low percentage zone for each lazy flyball.. His UZR would be record setting.

 

Each zone is weighted using data of balls caught or not caught in that zone collected over the years. All players... all games... all plays.

 

There are zones where the ball is caught so often that a caught ball in that zone will barely move the UZR needle upward. In those Zones... The only big needle moves are down and it takes one ball not caught.

 

There are also zones where not catching a ball will barely move the needle down but if you catch the ball in that zone... The needle moves big time up on one catch.

[jimmer]

This article from the fielding bible really breaks it down in easy terms to understand their process.

 

http://fieldingbible.com/Fielding-Bible-FAQ.asp

 

'Every MLB play is entered into the computer where we record the direction, distance, speed, and type of every batted ball. Direction and distance are recorded on a computer screen by simply clicking the location of the ball on a replica of the field shown on the screen. Speed is scored as either soft, medium, or hard, while different batted ball types include bunt, groundball, liner, fly, and “fliner”. Fliners, introduced in 2006, are balls that are somewhere between flies and liners.

 

Beginning in 2009, BIS video scouts began additionally tracking batted ball timer information. This data has allowed us to make objective assessments of the velocity and trajectory of each grounder or flyball.'

[TheLeviathan]

 

If anyone reads an article about stats such as "wins above average/replacement" or "runs about average/replacement", which are by name and definition theoretical constructs and quite obviously not anything that exists on a baseball scorecard, and assumes they are perfect or precise or immediately corroborated by the simple records on said scorecard, is demonstrating some level of preconceived hostility to the endeavor, I think.  

 

Maybe and maybe that is part of the problem.  At the same time, statistics in sports have been precise for most of their existence.  If you want people to understand and accept imperfect approximations (which, AGAIN, everyone agrees is their known intent.  Trying to save you a ramble here), than perhaps the onus is on the writer.

 

Sabermetrics are new and if they are new both in how they measure and what they measure, I'd think wider acceptance and understanding isn't the job of the reader.  It's the writer.

 

Where don't I see a need for change?

 

 

You sort of made a rambling assumption after this, so I'll just respond here:  A change in how advanced metrics discuss and utlize their metrics so as not to cause so much confusion.  Wasn't talking about changing which metric is used (please, get off that, no one is saying that to you), just a change in how the metrics are discussed.

[Richie the Rally Goat]

It's definitely not zero, and poor fielding can cause home runs

[TheLeviathan]

I thought this was a helpful and amusing little article on the subject that I came across.

[Otto von Ballpark]

Honest question, because I don't know the answer:

 

Lets say Arcia gets 250 lazy fly balls hit to him in left this year, and every one is hit right to him. He hardly has to move all year, and catches all 250.

 

What would his UZR be?

Is this assuming that zero balls are hit anywhere else to LF or left center? Such plays at least occasionally converted into outs by the average LF but not converted into outs by Arcia would negatively affect his UZR, probably offsetting the cans of corn and then some.

 

But yes, if every ball hit to left or left center is coverted into an out by Arcia, he will have a positive UZR, as it should be.

[Otto von Ballpark]

On the subject of defense... I believe it's huge but really hard to quantify. It's all about the extra out or stolen outs and isolated moments in games and the crooked number that kills teams.

 

One ball not caught can be the difference between hanging a zero for an inning and allowing five in an inning.

 

 

That is an argument for defensive WPA Win Probability Added. A diving catch with the bases loaded and two outs is more valuable than an identical catch with no one on. Interesting, but not really a good way to estimate fielding performance or ability.

[Otto von Ballpark]

That one ball not caught because a player lost the ball in the sun or was shifted over into Zone 7A and he falls into a huge UZR hole that is impossible to climb back out of.

 

.

It is not that big or impossible. I don't have the exact number handy, but it is less than a run. Maybe 0.7 runs for a hit to the outfield? And the fielder only gets debited that full amount from his UZR if the average player at his position coverted it into an out at virtually 100% rate.

 

That's not why Gutierrez went from +30 to 0 UZR. He would have needed to miss 40 such balls in 2010 over 2009 for that to be true.