Value of Defensive Prowess

[oldtimer]

A lot of talk has been directed at the evaluation of offensive prowess using Sabermetrics. Evaluation of defensive prowess and the evaluation of it doesn't seem to be something that can be quantified as easily. Errors are assigned judgmentally and vary depending on things like a players range, etc. We have seen though that a team like the Royals who were excellent defensively, had an advantage which helped them win the series.

 

What got me thinking about this subject is the call up of Moncada. He may be somewhat error prone at this stage in his development, but clearly has room to grow into a more consistent defender. So what method can we look at his and other players defensive capability?

 

1. Ist I would say that defensive ability is relative to what the average player at that position can do.

2. 2nd that errors made on balls the average player at that position should handle, including fielding and throws, are bad errors and deserve a (-2)

3. 3rd that errors made on balls that a player with quickness and range can get to but an average player cannot are still errors but should be less of a negative. (-1)

4. 4th an average player who doesn't have the range to make a play on a ball the top 1/4 of players would make should be given a (-1)

5 5th that plays made on balls that a player with quickness and range can get to but and average player cannot make should be given a (+1)

6. 6th Brooks Robinson type plays that have a wow factor and often times preserve a lead or such deserve a (+2)

7. There is another category which is on the mental error side. Not covering a base, throwing to the wrong base, not backing up in the outfield, etc should be a cause to be given a (-1), since doing the right thing is expected.

 

The realization is that such a system is highly subjective but could be applied to every field position except catcher where it becomes extremely subjective since catching a knuckleball vs a traditional pitcher would greatly skew the results.

 

Right now I know of no system that measures defensive prowess effectively and instead the day to day watching of players. Clearly it is unlikely that anyone will try to implement such a system as it would require watching every game and scoring every play. Right now many of us who follow our team do the mental gymnastics to identify really good defensive players vs only average ones. JBJ vs Hanley might be an example.

 

If there are good metrics for defensive prowess and how they may impact the teams overall performance please point me in the right direction to find them.

[Kimmi]

UZR and DRS do a pretty comprehensive job of what you're talking about. No, they're not perfect, but they're pretty doggone good, despite what the critics say.

[S5Dewey]

UZR and DRS do a pretty comprehensive job of what you're talking about. No, they're not perfect, but they're pretty doggone good, despite what the critics say.

 

And I'm one of those critics.I'm firmly of the belief that defensive statistics are mostly voodoo, which makes WAR mostly voodoo too. I have no problem with offensive WAR in that it's made up of hard numbers with no subjectivity involved, but when one starts mixing defensive "stats" with offensive hard numbers it devalues the total of the two.

[moonslav59]

UZR does use trained observers to calculate how many balls hit into a player's zone are fielded. They rank easy ones, hard ones and near impossible ones.

 

It sounds a lot like the one you are pining for.

 

 

 

[Thunder]

Why don't we just replace all position players with robots?

[Hugh2]

There's a big difference between subjective and voodoo. By that definition an umpires strike zone is voodoo, I think all non cynical baseball fans would agree that while umpires can have a bad day and be a little off a strike zone is subjective. It might vary but it's pretty darn close, and like Moon said they have trained observers.

 

Even though we know umpires are imperfect we trust them enough to never blame the umpire for a pitcher with bad control. No one would ever say that we can't trust Henry Owens WHIP, and BB/9 because umpires strike zones are voodoo....you just wouldn't. Defensive metrics are very similar in this sense, people just don't see it that way because it's new and what is new scares people.

[Emp9]

The average player is pretty good, lets not forget. WAR for example isn't based on the average player, its based on the typical replacement player. A replacement player is usually of varying degree not even close to an average player.

 

With that said, I wouldn't mind a stat similar to WAR in every way that uses the average player as the baseline. WAA?

[oldtimer]

UZR does use trained observers to calculate how many balls hit into a player's zone are fielded. They rank easy ones, hard ones and near impossible ones.

 

It sounds a lot like the one you are pining for.

 

 

 

 

Thanks for the input. I will look at the UZR. Does to place any value on defensive capability? It would be of value if runs saved or lost compared to others might be good. Like I say, I will look

[Slasher9]

Why don't we just replace all position players with robots?

 

umps first. then players.

[mvp 78]

Why don't we just replace all position players with robots?

 

Would the robots have stirrups?

[S5Dewey]

Would the robots have stirrups?

 

Yes. And they wouldn't be allowed to wear the legs of their pants down at their ankles.

[Slasher9]

[mvp 78]

Yes. And they wouldn't be allowed to wear the legs of their pants down at their ankles.

 

How about batting gloves or eye-black?

[sk7326]

And I'm one of those critics.I'm firmly of the belief that defensive statistics are mostly voodoo, which makes WAR mostly voodoo too. I have no problem with offensive WAR in that it's made up of hard numbers with no subjectivity involved, but when one starts mixing defensive "stats" with offensive hard numbers it devalues the total of the two.

 

The stats and the eyes complement each other. Now on TV, the eyes lie a lot - you don't see the route an outfielder takes, even if the catch and end is pretty good. It is good when the numbers fit what your eyes see - Ozzie Smith's stardom was entirely justified analytically.

 

I think what WAR (which is a normalized sum of everything a player does that is measurable on the field) has advanced the conversation in some important ways. When I was a kid, you only thought in terms of pitchers and batters as far as getting down to runs. Yeah there were errors and wild pitches and such, but team defense as a meaningful contribution to run prevention was rarely talked about. We have a much better understanding (and it is still fuzzy granted) of the split between pitcher and fielder with regards to run prevention.

[sk7326]

How about batting gloves or eye-black?

 

Should robots be allowed to wear elbow pads? Or do you spend extra to reinforce the bodies?

 

For authenticity maybe they can power them via fuel coming from burning Dip or Chaw

[Kimmi]

And I'm one of those critics.I'm firmly of the belief that defensive statistics are mostly voodoo, which makes WAR mostly voodoo too. I have no problem with offensive WAR in that it's made up of hard numbers with no subjectivity involved, but when one starts mixing defensive "stats" with offensive hard numbers it devalues the total of the two.

 

Offensive stats are subjective too. Perhaps they don't have as much subjectivity as defensive stats do, but they are subjective nonetheless. As Hugh said, every ball or strike called by an umpire is subjective. Hits versus errors are subjective, which makes BA, OBP, and OPS all subjective.

 

Calling defensive statistics "mostly voodoo" is a little out of line, IMO, though you are certainly entitled to view them that way. The key is, as always, to look at as many stats as possible in conjunction with the eye test. However, in a crunch, I'm going with UZR/150 and DRS and feeling fairly confident that I have a pretty good read on that player's defense.

[Kimmi]

There's a big difference between subjective and voodoo. By that definition an umpires strike zone is voodoo, I think all non cynical baseball fans would agree that while umpires can have a bad day and be a little off a strike zone is subjective. It might vary but it's pretty darn close, and like Moon said they have trained observers.

 

Even though we know umpires are imperfect we trust them enough to never blame the umpire for a pitcher with bad control. No one would ever say that we can't trust Henry Owens WHIP, and BB/9 because umpires strike zones are voodoo....you just wouldn't. Defensive metrics are very similar in this sense, people just don't see it that way because it's new and what is new scares people.

 

Well said.

[Bellhorn04]

There's a big difference between subjective and voodoo. By that definition an umpires strike zone is voodoo, I think all non cynical baseball fans would agree that while umpires can have a bad day and be a little off a strike zone is subjective. It might vary but it's pretty darn close, and like Moon said they have trained observers.

 

Even though we know umpires are imperfect we trust them enough to never blame the umpire for a pitcher with bad control. No one would ever say that we can't trust Henry Owens WHIP, and BB/9 because umpires strike zones are voodoo....you just wouldn't. Defensive metrics are very similar in this sense, people just don't see it that way because it's new and what is new scares people.

 

Very good comments. The only thing I would disagree on, mildly, is that I don't think the new baseball stats scare people so much as there are factors of mistrust, and a little fatigue as well, because we're constantly getting new information shoved at us. It took me a while to accept that the new defensive stats, while flawed, are well intended and pretty good stuff.

[moonslav59]

There's a big difference between subjective and voodoo. By that definition an umpires strike zone is voodoo, I think all non cynical baseball fans would agree that while umpires can have a bad day and be a little off a strike zone is subjective. It might vary but it's pretty darn close, and like Moon said they have trained observers.

 

Even though we know umpires are imperfect we trust them enough to never blame the umpire for a pitcher with bad control. No one would ever say that we can't trust Henry Owens WHIP, and BB/9 because umpires strike zones are voodoo....you just wouldn't. Defensive metrics are very similar in this sense, people just don't see it that way because it's new and what is new scares people.

 

Yes, very well said.

 

The payed and trained observers who analyze each play in person are trained and rotated, unlike hometown scorekeepers.

 

Sure, there's some subjectivity and imperfections involved but the idea is rather simple.

 

Most fans get to see their own player every game or nearly every game. They see other team's players on defense rarely, some very rarely. If we did get a chance to see every play made and not made by every player over a season, we'd have a better understanding of how our players are on defense- good or bad. (That's assuming we can stay objective.)

 

The observers with UZR/150 are probably more objective than us, and you'd think with all the observations they make, they'd be better than we are at determining what play was makeable or not.

.

I don't totally trust their numbers, and even they say we should only look at large sample sizes (sometimes larger than a full season) to make any definitive judgments on a player's defensive ability.

 

UZR/150 is not perfect, but it beats Flg% and RF/9. It beats the two combined. It takes into account how many balls are hit to a player, how many were easy, medium hard and in-between. It beats my own personal observations by themselves, because I don't know jack about how good the Colorado Rockies SS is defensively through observations and looking at Fldg% and RF/9 only. If I don't know how good he is, how can I rank Bogey comparatively?

 

[moonslav59]

So, with all this talk of UZR/150, where do our players stand?

 

2014-2016 & 2,000+ innings (to get a true sample size):

 

CF (23 qualified):

JBJ +11.9 (6th behind Kiermaier, Pillar, B Hamilton, Cain and Lagares)

 

OF (71 qualified)

10) JBJ +13.5

14) Betts +7.1

41) Ellsbury -0.3

 

SS (25 qualified)

19.8 Lindor

19.2 Simmons

14.6 JJ Hardy

13.3 Cozart

11.0 Crawford

#15 Bogey -2.3

 

2B (21 qualified)

13.5 Pedey

8.3 Kinsler

6.9 Panik

5.0 Phillips

4.1 Utley

3.1 LeMahieu

2.6 Gordon

 

 

 

[oldtimer]

Offensive stats are subjective too. Perhaps they don't have as much subjectivity as defensive stats do, but they are subjective nonetheless. As Hugh said, every ball or strike called by an umpire is subjective. Hits versus errors are subjective, which makes BA, OBP, and OPS all subjective.

 

Calling defensive statistics "mostly voodoo" is a little out of line, IMO, though you are certainly entitled to view them that way. The key is, as always, to look at as many stats as possible in conjunction with the eye test. However, in a crunch, I'm going with UZR/150 and DRS and feeling fairly confident that I have a pretty good read on that player's defense.

 

I have gone back and looked into UZR/150, something I was unaware of at the time I initiated the post, and it is very similar to what I described as being a tool that at least could measure defense relative to other players. The assignment of a + or - runs gives a method of identifying an overall player worth. I am not a great believer in stats as the only judgment of the relative ability of a player but I can see when trying to determine who to trade or who to bring aboard it definitely should be a part of any consideration.

[moonslav59]

I have gone back and looked into UZR/150, something I was unaware of at the time I initiated the post, and it is very similar to what I described as being a tool that at least could measure defense relative to other players. The assignment of a + or - runs gives a method of identifying an overall player worth. I am not a great believer in stats as the only judgment of the relative ability of a player but I can see when trying to determine who to trade or who to bring aboard it definitely should be a part of any consideration.

 

Glad to see you expanding your horizons, oldtimer.

 

It's not a perfect method, but it sure beats Fielding % which is also flawed by hometown scorers assigning errors with subjectivity.

[Hugh2]

Very good comments. The only thing I would disagree on, mildly, is that I don't think the new baseball stats scare people so much as there are factors of mistrust, and a little fatigue as well, because we're constantly getting new information shoved at us. It took me a while to accept that the new defensive stats, while flawed, are well intended and pretty good stuff.

 

Yes I used the term scared very liberally and subjectively. I don't actually think people are cowering in the corners trembling in fear when words like UZR/150 come on the screen; although the thought of people curling up in the fetal position at the sight of defensive stats is pretty amusing. What I really meant is when something is new and doesn't have a strong track record or people don't understand it they tend to find reason to dismiss it or just outright dismiss it. I suspect a lot of the other technology that is coming out that tracks a fielders route towards baseballs and the angle and speed in which it travels could further improve defensive metrics one day to the point that they won't really need trained observers.

[moonslav59]

Yes I used the term scared very liberally and subjectively. I don't actually think people are cowering in the corners trembling in fear when words like UZR/150 come on the screen; although the thought of people curling up in the fetal position at the sight of defensive stats is pretty amusing. What I really meant is when something is new and doesn't have a strong track record or people don't understand it they tend to find reason to dismiss it or just outright dismiss it. I suspect a lot of the other technology that is coming out that tracks a fielders route towards baseballs and the angle and speed in which it travels could further improve defensive metrics one day to the point that they won't really need trained observers.

 

Yes, and the computer will be able to tell how long the ball was in the air, and how fast the player was able to get to the ball or not, and then compare that data to the norm.

 

The one area that may be contentious is player positioning. Do you blame the player for being farther away from a hit ball to start with?

[S5Dewey]

My biggest problem with defensive stats is that we like to use them to make a case, but we ignore them in favor of offensive stats when it suits us. IMO that's mostly because deep down we know that defensive stats aren't as accurate as the offensive ones.

 

When most people talk about a player they talk about his offense. His slash line. And defense barely gets mentioned. Even in the case of JBJ, when he was struggling to stay above the Mendoza line, very few of US were talking about how much of his stellar defense offset his offensive shortcomings.

 

I see from Moon's post that JBJ is 6th in UZR. The fact aside that I want to see five guys in MLB who have better range than JBJ, that's ONLY UZR that's being talked about and JBJ leads the league in OF assists. Doesn't that move him up some on the chart?

 

When we can put both offense and defense on a scale of, say, 1-100 and say that JBJ has an offensive rating of 85 and a defensive rating of 91 - and start talking about players whose defense is better than their offense like it matters (because it does!) I'll be a little more willing to consider defensive stats.

 

And please don't talk to me about WAR. Have you even seen the "formula" for WAR? It reminds me of a 6'7" pitcher with control problems. He has a big motion and there's a lot that can go wrong in there, and often times when there are a lot of different things that can go wrong something often does. I believe in KISS. Keep it simple, stupid. When two sites can determine a player's WAR and even THEY don't agree, how are we supposed to have any faith in it?

[moonslav59]

My biggest problem with defensive stats is that we like to use them to make a case, but we ignore them in favor of offensive stats when it suits us. IMO that's mostly because deep down we know that defensive stats aren't as accurate as the offensive ones.

 

When most people talk about a player they talk about his offense. His slash line. And defense barely gets mentioned. Even in the case of JBJ, when he was struggling to stay above the Mendoza line, very few of US were talking about how much of his stellar defense offset his offensive shortcomings.

 

I see from Moon's post that JBJ is 6th in UZR. The fact aside that I want to see five guys in MLB who have better range than JBJ, that's ONLY UZR that's being talked about and JBJ leads the league in OF assists. Doesn't that move him up some on the chart?

 

When we can put both offense and defense on a scale of, say, 1-100 and say that JBJ has an offensive rating of 85 and a defensive rating of 91 - and start talking about players whose defense is better than their offense like it matters (because it does!) I'll be a little more willing to consider defensive stats.

 

And please don't talk to me about WAR. Have you even seen the "formula" for WAR? It reminds me of a 6'7" pitcher with control problems. He has a big motion and there's a lot that can go wrong in there, and often times when there are a lot of different things that can go wrong something often does. I believe in KISS. Keep it simple, stupid. When two sites can determine a player's WAR and even THEY don't agree, how are we supposed to have any faith in it?

 

JBJ is not the fastest CF'er in MLB. He gets exceptional breaks on balls, and takes the straightest route, as far as I know. Lorenzo Cain is a phenom. The others, I don't know much about.

 

Without seeing other CF'ers as much as JBJ, I can't say nwho is best, but by seeing the aggregate CF'er, I think JBJ is about top 2-3. 6th is not that far off from 2-3.

 

WAR is seriously flawed, but so is standing around the water cooler and trying to compare defensive, base running and batting value mixed together and at the same time.

 

WAR attempts to put a single number to all values. It is flawed, and I can see why you and others don't like it, but you probably don't agree with "that other guy" at the water cooler either.

[Hugh2]

Yes, and the computer will be able to tell how long the ball was in the air, and how fast the player was able to get to the ball or not, and then compare that data to the norm.

 

The one area that may be contentious is player positioning. Do you blame the player for being farther away from a hit ball to start with?

 

I don't think it will matter, if a player has to run 20 yards to get to a ball he has to run 20 yards.

[Hugh2]

My biggest problem with defensive stats is that we like to use them to make a case, but we ignore them in favor of offensive stats when it suits us. IMO that's mostly because deep down we know that defensive stats aren't as accurate as the offensive ones.

 

When most people talk about a player they talk about his offense. His slash line. And defense barely gets mentioned. Even in the case of JBJ, when he was struggling to stay above the Mendoza line, very few of US were talking about how much of his stellar defense offset his offensive shortcomings.

 

I see from Moon's post that JBJ is 6th in UZR. The fact aside that I want to see five guys in MLB who have better range than JBJ, that's ONLY UZR that's being talked about and JBJ leads the league in OF assists. Doesn't that move him up some on the chart?

 

When we can put both offense and defense on a scale of, say, 1-100 and say that JBJ has an offensive rating of 85 and a defensive rating of 91 - and start talking about players whose defense is better than their offense like it matters (because it does!) I'll be a little more willing to consider defensive stats.

 

And please don't talk to me about WAR. Have you even seen the "formula" for WAR? It reminds me of a 6'7" pitcher with control problems. He has a big motion and there's a lot that can go wrong in there, and often times when there are a lot of different things that can go wrong something often does. I believe in KISS. Keep it simple, stupid. When two sites can determine a player's WAR and even THEY don't agree, how are we supposed to have any faith in it?

 

Give me an example where WAR grossly over rates or under rates a guy.

[sk7326]

My biggest problem with defensive stats is that we like to use them to make a case, but we ignore them in favor of offensive stats when it suits us. IMO that's mostly because deep down we know that defensive stats aren't as accurate as the offensive ones.

 

When most people talk about a player they talk about his offense. His slash line. And defense barely gets mentioned. Even in the case of JBJ, when he was struggling to stay above the Mendoza line, very few of US were talking about how much of his stellar defense offset his offensive shortcomings.

 

I see from Moon's post that JBJ is 6th in UZR. The fact aside that I want to see five guys in MLB who have better range than JBJ, that's ONLY UZR that's being talked about and JBJ leads the league in OF assists. Doesn't that move him up some on the chart?

 

When we can put both offense and defense on a scale of, say, 1-100 and say that JBJ has an offensive rating of 85 and a defensive rating of 91 - and start talking about players whose defense is better than their offense like it matters (because it does!) I'll be a little more willing to consider defensive stats.

 

And please don't talk to me about WAR. Have you even seen the "formula" for WAR? It reminds me of a 6'7" pitcher with control problems. He has a big motion and there's a lot that can go wrong in there, and often times when there are a lot of different things that can go wrong something often does. I believe in KISS. Keep it simple, stupid. When two sites can determine a player's WAR and even THEY don't agree, how are we supposed to have any faith in it?

 

The formula for WAR is complicated - because there is a lot of normalization (to get numbers measured on different scales to add up) - but the fact that bWAR and fWAR are calculated differently should INCREASE your confidence.

 

The two sites differ in WAR, because they have a couple of fundamentally different assumptions about player performance. That is great - better to have a couple of perspectives on this question than a stat like fielding percentage which tells you literally nothing helpful. One of the major differences between the two was the notion of what "replacement level" is - but they have come to some sort of consensus on that. This is big, as it at least puts the two on the same scale.

 

But you look at position players - WAR is essentially identical between the two. The differences are in defensive measurement, and neither of those differences are that huge. Pitcher WAR differs, but it differs because the two sites have different ways of trying to separate a pitcher's contribution to run prevention. Fangraphs fundamentally gives pitchers less control over batted balls in play Baseball Reference does. This explains the huge gap in David Price's season for instance. The key thing is - I don't think that the question is settled (how separable is pitcher and fielding performance in terms of run prevention), and it is worthwhile to see how tweaking the assumptions changes the picture.

 

The two versions of WAR ask fans to basically engage the numbers and why they are the way they are - the fundamental assumptions about player value, and why good stuff happens. Remember, RBIs are also a made up stat - arbitrarily assigning credit for a run scoring.

 

You are right to a certain degree about offense vs defense. Chicks dig the longball - so people will always talk about homers and hitting and such - there is an inherent bias there. At the same time, defensive measurement IS much flimsier than offensive measurement. So yes, if I see two players with close WAR and one builds it with defensive value (see Victorino in 2013), it would move them down the rankings for me a little bit. But the defensive value is there - and it is nice that we can at least bound the contribution. When Bradley stunk at the plate, we DID hear about his defensive brilliance - because we knew that the equation could work if he hit enough. But he was so dreadful offensively, that no defense could overcome it. Outfield assists are a great extra parlor trick Bradley has - uncommonly good for a CF - but outfield assists are generally fairly random and a function of opportunity. And in a sense, your best bet to get outfield assists is to have an above average arm, but not a great one (because then nobody would run on you).

[S5Dewey]

Give me an example where WAR grossly over rates or under rates a guy.

 

The issue is not whether WAR grossly underrates or overrates a player. My gripe with it is that when two sites try to calculate the same thing and get different results it brings the entire process into question. And the issue isn't on the offensive side. Anyone with a calculator and a rudimentary knowledge of baseball can calculate BA, OBP, SLG, OPS, etc. The error is entirely on the defensive side of the WAR equation, which magnifies the error.