Fangraphs: Twins may have weakened a weakness

[Richie the Rally Goat]

http://www.fangraphs.com/blogs/the-twins-may-have-weakened-a-weakness/

 

I found this article very well written, and from a guy who doesn't appear to care about the Twins, but has done some good analysis on the impact of the Twins off season improvements. Not exactly encouraging, and he doesn't even touch the CF situation.

[nytwinsfan]

Mostly compelling analysis, but I think most of us (except Terry Ryan) already knew this based on what Nick and Parker have been saying for a while now. It is unlikely the Twins were really going to compete in 2015 anyway, so one more year of bad outfield defense probably isn't likely to make the difference between a playoff team and an 75-80 win team (although it may further demoralize Twins' pitchers and fans). What really worries me is that it suggests Terry Ryan will continue to evaluate outfielders' defense almost entirely on the eye test and his pitchers based on runs that should in all fairness not be attributed to their pitching.  All I can say is thank goodness for Byron Buxton.

[Seth Stohs]

I'd be curious how many anD readers read and trust advanced defensive metrics? I have little confidence in most of them. I read some, find them interesting but definitely still prefer and weigh the eye test more.

[Richie the Rally Goat]

Mostly compelling analysis, but I think most of us (except Terry Ryan) already knew this based on what Nick and Parker have been saying for a while now. It is unlikely the Twins were really going to compete in 2015 anyway, so one more year of bad outfield defense probably isn't likely to make the difference between a playoff team and an 75-80 win team (although it may further demoralize Twins' pitchers and fans). What really worries me is that it suggests Terry Ryan will continue to evaluate outfielders' defense almost entirely on the eye test and his pitchers based on runs that should in all fairness not be attributed to their pitching.  All I can say is thank goodness for Byron Buxton.

agreed that we all knew this based on Nick and Parker's analysis, but to me the point is the observation coming from a slightly more objective perspective. We all care about this team, but it speaks a bit louder when someone who doesn't care about this teams says "wow! How did they manage to get worse?"

 

What worries me is we've pinned all of our outfield improvement to one prospect. Last I saw there were three outfield positions, all of which manned by poor to terrible fielders. 2013 looked like all the rotation improvement hopes were pinned to Meyer and May, but since then, free agent acquisitions made the arrival of those two more incremental than ALL of the improvement. I was hoping to see a similar type of improvement in CF or a corner OF.

[TheLeviathan]

I don't trust them, but it's hard not to see how they were accurate about the Twins last year.  At worst, the criticism of the team's philosophies are valid.  We have an awful defensive outfield and a fly-ball pitching staff.....that's just a really bad idea no matter what stats you trust to bear that out.

[Richie the Rally Goat]

I'd be curious how many anD readers read and trust advanced defensive metrics? I have little confidence in most of them. I read some, find them interesting but definitely still prefer and weigh the eye test more.

i wouldn't use the word "trust". Gather all of the evidence, if it all of the metrics point to the same conclusion, I feel that you can reasonably conclude at least directionally the same. Eye test is important too, metrics can't tell the whole story.

[DAM DC Twins Fans]

I'd be curious how many anD readers read and trust advanced defensive metrics? I have little confidence in most of them. I read some, find them interesting but definitely still prefer and weigh the eye test more.

I definitely look at UZR etc. but take them with very few grains...more use the eye test. In any case, Twins defense is poor--maybe Plouffe maybe Dozier and definitely Escobar are above average...that's it. Torii at age 39 wont help on defense--I hope he helps Santana, Hicks and Buxton (and maybe AB Walker and Rosario) develop--I firmly believe that is why he was signed.

[Thrylos]

I'd be curious how many anD readers read and trust advanced defensive metrics? I have little confidence in most of them. I read some, find them interesting but definitely still prefer and weigh the eye test more.

 

Does Hunter pass the eye test here?

[USAFChief]

Does Hunter pass the eye test here?

Does Mike Trout going from +16 in 2012 to -10 in 2014 according to UZR pass the smell test?

[jorgenswest]

Does Mike Trout going from +16 in 2012 to -10 in 2014 according to UZR pass the smell test?

Would the same statement be made when a player's batting average goes from .260 to .159 or a pitcher's ERA goes from 3.52 to 5.38? As far as year to year reliability, DRS and UZR have been shown to be more reliable than ERA and batting average.

[Thrylos]

Does Mike Trout going from +16 in 2012 to -10 in 2014 according to UZR pass the smell test?

 

Have no idea.   Have not really been following the Angels that close...

[markos]

Does Mike Trout going from +16 in 2012 to -10 in 2014 according to UZR pass the smell test?

Is there compelling evidence that he is exactly the same player as he was in 2012? Maybe he isn't quite as quick as he used to be. His SBs have dropped from 49 int 2012 to 16 in 2014. 

[PseudoSABR]

Would the same statement be made when a player's batting average goes from .260 to .159 or a pitcher's ERA goes from 3.52 to 5.38? As far as year to year reliability, DRS and UZR have been shown to be more reliable than ERA and batting average.

I'd like to see where this has actually been substantiated, and by whom.   I tend to disregard stats that reaffirm themselves by using statistics that are supposedly irrelevant (e.g. using the difference between FIP and ERA to substantiate DRS (as the article seems to do)) (or using Runs to substantiate the importance of walks).

[Richie the Rally Goat]

Does Mike Trout going from +16 in 2012 to -10 in 2014 according to UZR pass the smell test?

wasnt Hunter supposed to be the great Mike Trout mentor?

[jorgenswest]

I'd like to see where this has actually been substantiated, and by whom.   I tend to disregard stats that reaffirm themselves by using statistics that are supposedly irrelevant (e.g. using the difference between FIP and ERA to substantiate DRS (as the article seems to do)) (or using Runs to substantiate the importance of walks).

Simple even/odd correlation tests can be used to measure variability year to year of any stat. It was argued that we shouldn't trust a metric because a single player had a significant yearly change. A correlation test will take a group of players and look for yearly change rather than a single case. As a group, DRS is more stable year to year than batting average or a ERA. Don't trust UZR or DRS? That's only reasonable if many other measures commonly cited are equally trusted?

 

There are several that have been done.

 

http://www.billjamesonline.com/how_well_do_advanced_defensive_statistics_correlate/

[Kwak]

I think people are looking too closely at the trees and ignoring the forest that they represent.  The article isn't about statistics it is about the Twins overall problems (poor team defense and "hit bats" philosophy) and what the Twins did (so far) to address said problems.  The thrust of the article is that a lot of money was committed but that we shouldn't expect much (if any) improvement in team defense, and by extension runs allowed, given the expenditure. 

[whydidnt]

The article provide additional evidence regarding what some of us have been saying over the last couple of years...Terry Ryan seems to have fallen behind the times when it comes to building teams. There was time, before all of the additional data was available, where OF defense was more of a 2nd thought. Those times have changed, as many organizations now know better. It just seems the Twins haven't really figured that out, especially when you have, and have had, a pitching staff that is by far the worst in the league at striking out hitters. 

[USAFChief]

Would the same statement be made when a player's batting average goes from .260 to .159 or a pitcher's ERA goes from 3.52 to 5.38? As far as year to year reliability, DRS and UZR have been shown to be more reliable than ERA and batting average.

I don't believe a player's true defensive ability can change that much over the course of two seasons. Particularly a player Trout's age. I guess it's possible but I don't believe it.

 

Therefore, unlike ERA or BA, which are measures of outcomes that I can understand and trust, and which I believe can vary widely over a short time period, I am skeptical of defensive metrics. Having a player vary that much, when IMO defensive ability is something that can only change gradually, over time, is just more reason for me to mistrust them.

 

I might be in the minority with that opinion, and perhaps I'll be proven wrong. Wouldn't be the first time, nor would it be a first if I'm proven correct.

[iTwins]

Belief or non-belief in defensive metrics aside, I think most Twins fans could tell you the defense is a concern.

 

Strictly by the eye test, Torii's defense has been declining each year for the past 3 or 4 years. Meanwhile, were plopping a still...defensively developing (I'll be generous) Arcia into LF. The corners are anchored by slow, plodding power bats. CF has...well, maybe a converted SS? Even if Hicks DOES start in CF, his range cannot make up for all of the lost range courtesy of Arcia and Hunter.

 

I'd argue the infield is a bit better than the numbers say. Dozier has been great at 2B and Plouffe showed marked improvement at 3B. Mauer has the athletic ability to play great defensive 1B - Danny is the only question - but he's has the potential to be solid there as well.

 

That MI would be okay if this was a particularly heavy groundball staff...alas, that's not the case.

 

The point in the article is one that's been made countless times around here - the pitching staff allows a lot of balls in play (especially fly balls), but the defense isn't exactly the greatest at catching those balls. That's a concerning way to build a team.

[TheLeviathan]

Therefore, unlike ERA or BA, which are measures of outcomes that I can understand and trust, and which I believe can vary widely over a short time period, I am skeptical of defensive metrics. Having a player vary that much, when IMO defensive ability is something that can only change gradually, over time, is just more reason for me to mistrust them.

I like the way you said this. A study that shows BA or ERA vary the same as a defensive metric isn't a good thing. Defense doesn't involve nearly as much luck or other variables that change the outcomes. A good fielder should be a good fielder barring anything but aging, injury, or Chuck Knoblauch disease. BA can and will vary due to a great many more factors.

 

It's nice that the metrics are getting better, but I always bristle at comparing a subjective-based stat with one derived purely from objective outcomes. ERA and UZR are really apples and oranges.

[Brandon]

I think the defense is below where we would like it to be.  with an adequate CF the OF defense will be fine.  We have Schaeffer who can sub for late inning defense.  Torii was a plus defender until he moved to Detroit.  I don't think he forgot routes, or how to throw, and what not.  I think he has lost a step but that can be compensated for to some degree.  He will be better defensively than the posters here are making his defense seem.  Yes a speedster who can catch everything will make the defense better but can they hit like Hunter or will they kill more rally's then they stop with their defense.  But I think Detroit's stadium is a bigger cause of his defensive decline then his actual decline.  I bet he grades out even defensively or league average which is fine with our offense.  If Suzuki hits .280 and Santana hits .300 with close to 30 SBs and Mauer gets close to .320 and Hunter hits like himself We may have a top 3 offense in the league.  Our pitching should be league average.  That should net us a positive run differential.  which = a winning season. 

[whydidnt]

  I bet he grades out even defensively or league average which is fine with our offense.  If Suzuki hits .280 and Santana hits .300 with close to 30 SBs and Mauer gets close to .320 and Hunter hits like himself We may have a top 3 offense in the league.  Our pitching should be league average.  That should net us a positive run differential.  which = a winning season. 

I would take that bet. He hasn't been league average for a few years, and unfortunately he's getting older, which typically means he isn't getting better. I actually am pretty concerned about Hunter's offense too, at this age. He kept his average up last year by swinging (and often hitting) everything.

 

That is a lot of if's just to get to a winning season. I hope things go that way, but don't really think adding Santana and Stauffer turns the worst staff in the league into league average, especially when the OF defense is still in need of major improvement.  Even if we take the defensive numbers with some skepticism, there has to be some fire around that smoke, which was really the point of the original post. The Twins have had a few weaknesses over the last few years, and seemingly don't care to improve them, with OF defense being the top of the list.  

[Richie the Rally Goat]

I think the defense is below where we would like it to be.  with an adequate CF the OF defense will be fine.  We have Schaeffer who can sub for late inning defense.  Torii was a plus defender until he moved to Detroit.  I don't think he forgot routes, or how to throw, and what not.  I think he has lost a step but that can be compensated for to some degree.  He will be better defensively than the posters here are making his defense seem.  Yes a speedster who can catch everything will make the defense better but can they hit like Hunter or will they kill more rally's then they stop with their defense.  But I think Detroit's stadium is a bigger cause of his defensive decline then his actual decline.  I bet he grades out even defensively or league average which is fine with our offense.  If Suzuki hits .280 and Santana hits .300 with close to 30 SBs and Mauer gets close to .320 and Hunter hits like himself We may have a top 3 offense in the league.  Our pitching should be league average.  That should net us a positive run differential.  which = a winning season.

 

i disagree that a good CF fixes the outfield situation. One player cannot cover from foul pole to foul pole. A good fielding outfield can have Hunter or Arcia not both. That's not to say LF and CF require having no-bat speedsters. LF just needs to have enough range to be a good fielder so CF can shade to the right a bit. TF is smaller than Detroit, but so many outfielders have trouble with the RF wall. I'm not convinced TF will make Torii look like a better fielder.

[dxpavelka]

The article provide additional evidence regarding what some of us have been saying over the last couple of years...Terry Ryan seems to have fallen behind the times when it comes to building teams. There was time, before all of the additional data was available, where OF defense was more of a 2nd thought. Those times have changed, as many organizations now know better. It just seems the Twins haven't really figured that out, especially when you have, and have had, a pitching staff that is by far the worst in the league at striking out hitters. 

All this advanced metrics stuff makes some of us wonder how guys managed to build teams for a hundred years before it became all the rage. 

[kab21]

I'm not going to blast the Twins for signing Hunter to a one year contract and ignoring defense.  They needed a one year OF'er that was decent at something.  The alternative is signing players that deserve one year (and mostly suck) or signing marginal players for multiple years. 

 

On the bright side the Twins have stocked the farm system with a number of athletic types that could be average to very good fielders.  Buxton, Santana, Walker, Rosario, Polanco, Gordon and probably a few others.  The team composition should significantly change in the next couple of years and that could greatly change the team defense.

 

The flip side of that is the stockpile of DH's on the MLB team and upper minors.  Mauer, Vargas, Arcia, Sano, Pinto and possibly another prospect or two.  You can only hide so many of these guys (Mauer is fine at 1B for example) without dragging the team defense down. 

 

Additionally anyone campaigning for Sano to stay at 3B should reconsider that stance if they are critical about team defense. 

[TheLeviathan]

All this advanced metrics stuff makes some of us wonder how guys managed to build teams for a hundred years before it became all the rage. 

 

People also managed to live without indoor plumbing but I'm not using your logic in that scenario either.

[PseudoSABR]

Simple even/odd correlation tests can be used to measure variability year to year of any stat. It was argued that we shouldn't trust a metric because a single player had a significant yearly change. A correlation test will take a group of players and look for yearly change rather than a single case. As a group, DRS is more stable year to year than batting average or a ERA. Don't trust UZR or DRS? That's only reasonable if many other measures commonly cited are equally trusted?

 

There are several that have been done.

 

http://www.billjamesonline.com/how_well_do_advanced_defensive_statistics_correlate/

Correlation doesn't substantiate the viability of the metric unfortunately (that precision v. accuracy distinction).   Correlation, I guess, means something.  For me, quantifying defense sounds a whole lot like the eye-test embedded in statistical puffery.    Of course, whatever errors exist in the metric would continue to consistently show up, hence the correlation of an individual metric with itself. 

[markos]

I don't believe a player's true defensive ability can change that much over the course of two seasons. Particularly a player Trout's age. I guess it's possible but I don't believe it.

 

Therefore, unlike ERA or BA, which are measures of outcomes that I can understand and trust, and which I believe can vary widely over a short time period, I am skeptical of defensive metrics. Having a player vary that much, when IMO defensive ability is something that can only change gradually, over time, is just more reason for me to mistrust them.

 

I might be in the minority with that opinion, and perhaps I'll be proven wrong. Wouldn't be the first time, nor would it be a first if I'm proven correct.

I am in the other camp, as I trust the defensive metrics quite a bit. I also feel like I'm in the minority with that opinion on this website. We can't both be minorities, right?

 

I have a couple comments to make:

 

1) UZR and DRS aren't trying to measure defensive ability. They are trying to determine contribution to wins by calculating runs allowed or runs saved for (almost) every single defensive play that is made or missed. This can serve as a proxy for defensive ability, but they aren't necessarily the same thing. Consider how offensive stats (like BA, OPS, wRAA, etc) don't directly measure hitting ability, as they don't care if a single was a solid line drive or a little nubber that can't be fielded. The defensive metrics have the same deficiency. They have tracked thousands and thousands of batted-balls and determined which ones, historically, are easy to field and which ones are hard. But that doesn't necessarily mean that any specific 'hard' play was actually hard (maybe it was easy because of good positioning) or any specific 'easy' play was actually easy (maybe it was hard because of sun or wind issues). So there is going to be some disconnect between value calculated and the underlying ability. However, the hope is that after hundreds and hundreds of fielding plays, the value and ability will line up. And I think the year-to-year correlation (though not quite as strong as OPS) does point to a decent relation between the value accounted for a player during a specific season and the player's underlying ability. 

 

In the instance of Torii Hunter last year, it could be the case that his season was the UZR equivalent of a low BABIP, and his actual defensive ability is much higher than the UZR number. Again, that doesn't mean that the UZR value is invalid, just like how a good hitter can have a bad season.

 

2) Similar to how offensive stats used to calculate runs (like wRC and wRAA) are adjusted based on the run environment of a given season, UZR (and maybe DRS?) are adjusted based on the quality of defenders at each position. So a player's UZR can go up and down from season to season even if that player's defensive ability is unchanged. For example, Trout may have dropped from average to below-average UZR from 2013 to 2014 because there were better defenders at centerfield in 2014. This adjustment doesn't much make sense if UZR was explicitly measuring defensive ability, but it is not - it is trying to determine contribution to wins. And if a player isn't one of the top 15 defenders at a position, he is a net negative for his team.

[USAFChief]

I am in the other camp, as I trust the defensive metrics quite a bit. I also feel like I'm in the minority with that opinion on this website. We can't both be minorities, right?

 

I have a couple comments to make:

 

1) UZR and DRS aren't trying to measure defensive ability. They are trying to determine contribution to wins by calculating runs allowed or runs saved for (almost) every single defensive play that is made or missed. This can serve as a proxy for defensive ability, but they aren't necessarily the same thing. Consider how offensive stats (like BA, OPS, wRAA, etc) don't directly measure hitting ability, as they don't care if a single was a solid line drive or a little nubber that can't be fielded. The defensive metrics have the same deficiency. They have tracked thousands and thousands of batted-balls and determined which ones, historically, are easy to field and which ones are hard. But that doesn't necessarily mean that any specific 'hard' play was actually hard (maybe it was easy because of good positioning) or any specific 'easy' play was actually easy (maybe it was hard because of sun or wind issues). So there is going to be some disconnect between value calculated and the underlying ability. However, the hope is that after hundreds and hundreds of fielding plays, the value and ability will line up. And I think the year-to-year correlation (though not quite as strong as OPS) does point to a decent relation between the value accounted for a player during a specific season and the player's underlying ability.

 

In the instance of Torii Hunter last year, it could be the case that his season was the UZR equivalent of a low BABIP, and his actual defensive ability is much higher than the UZR number. Again, that doesn't mean that the UZR value is invalid, just like how a good hitter can have a bad season.

 

2) Similar to how offensive stats used to calculate runs (like wRC and wRAA) are adjusted based on the run environment of a given season, UZR (and maybe DRS?) are adjusted based on the quality of defenders at each position. So a player's UZR can go up and down from season to season even if that player's defensive ability is unchanged. For example, Trout may have dropped from average to below-average UZR from 2013 to 2014 because there were better defenders at centerfield in 2014. This adjustment doesn't much make sense if UZR was explicitly measuring defensive ability, but it is not - it is trying to determine contribution to wins. And if a player isn't one of the top 15 defenders at a position, he is a net negative for his team.

Great post.

 

I disagree with the conclusions, but great post. I doubt, for example, that the overall quality of major league center fielders changed enough in two years for Trout to go from above average to below average. Once again, I suppose that could be true, but it doesn't pass the smell test to me. I doubt the overall quality of CF play has changed that much in two decades, much less two years.

[jorgenswest]

Batting average and a ERA aren't very good stats and do a poor job of projecting the future taken in one year lumps. They are not very useful in describing a player's skill in the span of one year. They do describe past performance. The same can be said about UZR and DRS. Those same measures in three years have more meaning in terms of skill.

 

Trout is on the extreme end of change. The following factors may have come in play. His stolen bases dropped significantly. His strike outs increased significantly. Was there something underlying effecting his speed or reactions? Players often outgrow shortstop physically. Is it also true in CF? Bourjos and his 450 innings in CF all landed on Trout who had previously played LF with Bourjos in center. Could that have been factor in the variation?

 

Too often other commonly used measures that also vary greatly are used in describing a player's skill. Many of them taken in one year blocks have no significant meaning in describing a skill. Taken in three years, it looks like a Trout is a solid CF. Is he trending down? The Angels ought to be paying attention, but they should not be jumping to a conclusion.

 

On the Twins front, Plouffe had his best year in terms of defensive measures. Has he improved? Maybe. He had little experience at 3B and his reactions and footwork must have improved with experience. One year of improved numbers shouldn't be used in an argument that his skill has changed. It isn't enough sample. It wouldn't be for ERA or batting average either.