Izturis a Gold Glove snub???

[Lucky Jim]

So if a systematic approach to defensive evaluation comes up with a different evaluation of Nick and Adam for 2009 than the prior consensus, your only conclusion is that the system is bad?

Do you say the same thing about batting average when Darin Erstad hit .350 one year and .250 the next? Or about OPS (and almost all other offensive metrics) when Nick has nearly a .900 in 2008 but barely an .800 in 2009? "When players like Markakis and Huff come up with below average OPS numbers then it really points out that the system is flawed and has a way to go before it can be relied on as a true measure of on offensive ability."

It's astounding that people actually think there's no such thing as a down year defensively. If a well-regarded metric (backed up by observation) says your favorite player had an off year, it's the metric and the observations that are wrong. I guess Izturis basically leading the league in UZR was wrong, too? Or was it just coincidentally right?

It's nice to see that I'm not the only person frustrated by this line of argument.

1. We realize that measuring defense by watching it is made difficult by the fact that value accrues in small differences over a multitude of transactions.

2. You invent a metric that tries to "see" what you can't see.

3. You continue to improve and refine that metric.

4. When that metric conflicts with what you think you see - i.e., shows you what you've admitted you can't "see" - you doubt the metric because it conflicts with what you've admitted you can't see.

Brilliant.

[wildcard]

We know what Batting Average, OBP and Slugging percent shows. For better or worse we have used them for a period of time and understand their strength and weaknesses.

UZR/150 is an evolving stat. It is unclear how good it is. We don't really know what the impact of Jones expanding his range in center has on Markakis's URZ/150 in right. There is no way to measure that as far as I can tell. There is OOZ, another new stat but the interact between players is not really measured. Just the chances, either in or out of zone.

Stat guys value UZR/150. It should be very apparent to everyone that baseball professionals do not for the most part. They believe what they see. They believe their eyes and the opinions of people they trust.

Stat guys may think that their system is better but until it is established with some kind of explanation for how it coordinates with what knowledgable baseball professional see, I don't look it for it to trust in many baseball circles.

[BaltimoreTerp]

Stat guys value UZR/150. It should be very apparent to everyone that baseball professionals do not for the most part. They believe what they see. They believe their eyes and the opinions of people they trust.

If you believe that they only believe their eyes and certain opinions, I know 30 teams that would love to have one of their opponents hire you.

[wildcard]

If you believe that they only believe their eyes and certain opinions, I know 30 teams that would love to have one of their opponents hire you.

Do you have any evidence that baseball professionals have widely accepted UZR/150?

Certainly, it does not appear to influence the baseball managers and coaches that voted for the Gold Glove award.

[TGO]

Do you have any evidence that baseball professionals have widely accepted UZR/150?

Certainly, it does not appear to influence the baseball managers and coaches that voted for the Gold Glove award.

Managers and coaches don't make personnel decisions. Generally speaking.

[RVAbird]

Missed too many games to even have a shot at it, and in all honesty, I can't hold that against the voters, if that was a reason why they didn't vote for him, as it's something I'd like to see more of. Even if he had been healthy and played 130-140 games I doubt he'd have won it anyway going up against Jeter, Andrus, and others.

Yeah, they probably dismissed Izturis for playing in only 114 games.

It was Adam Jones' gritty 119 game performance that earned him the Gold Glove in their eyes.

[wildcard]

Managers and coaches don't make personnel decisions. Generally speaking.

Right, they influence the decisions.

[amateurfan]

If you believe that they only believe their eyes and certain opinions, I know 30 teams that would love to have one of their opponents hire you.

I don't understand this post either. It's apparent that the baseball professionals who voted for the GGs did not use advanced defensive stats analysis. Therefore, I don't see how WC is wrong. It seems your usual snarkiness is out of place here.

[DrungoHazewood]

Do you have any evidence that baseball professionals have widely accepted UZR/150?

Certainly, it does not appear to influence the baseball managers and coaches that voted for the Gold Glove award.

If you don't think that many or most GMs and other front office personnel use things like UZR every single day you haven't been paying attention. Don't you remember that UZR was proprietary for a few years because the inventor got hired by a major league team to work in their front office, so he couldn't release the data anymore?

Of course the managers don't use it as much. They're not paid to analyze transactions and acquisitions.

[RShack]

Stat guys value UZR/150. It should be very apparent to everyone that baseball professionals do not for the most part. They believe what they see. They believe their eyes and the opinions of people they trust.

Stat guys may think that their system is better but until it is established with some kind of explanation for how it coordinates with what knowledgable baseball professional see, I don't look it for it to trust in many baseball circles.

This is right. The problem with iffy evolving stats, which is what we have with D-stats so far, is that some stat guys take the position that they should be trusted unless you can prove otherwise. That's the main problem: they take some iffy and evolving thing and act like it's the preferred way to look at things. They act like the onus is on others to prove an iffy thing is wrong. The proper attitude is to treat it like a hypothesis, as something under investigation, something based on incomplete information that can see some things but not everything. But, as should be clear from this thread, that's not what happens. Instead, they take the view that it should be trusted because, well, because it's a stat that they believe has fewer problems than other stats. In other words, it's the least-objectionable stat. That is not a good reason to trust it. Even worse, some (not all, but some) act like people who don't accept that, who question whether we should put high levels of trust in the least-objectionable stat, are somehow Neanderthal idiots for not trusting it. That is not a good way to treat people who question unproven and evolving metrics. It's the same kind of thinking that had some doctors doing lobotomies on people who they thought were crazy, had them using massive electric shocks to scramble the brains of people who were depressed (way different than current techniques). The attitude was "it's state of the art, so we should trust it". Meanwhile, lots of other doctors were yelling "Wait a minute! You don't know what you're doing!", and it's the latter group who were right.

Doing D-stats right likely requires data acquisition methods beyond what are currently in place. The current stat-of-the-art boils down to best guesses based on only partial information, but some people treat them as if they are actual facts about D-goodness. There is no reason to conclude that they are. When they become adequate, one way we'll know is that most everybody will agree with them.

[Lucky Jim]

This is right. The problem with iffy evolving stats, which is what we have with D-stats so far, is that some stat guys take the position that they should be trusted unless you can prove otherwise. That's the main problem: they take some iffy and evolving thing and act like it's the preferred way to look at things. They act like the onus is on others to prove an iffy thing is wrong. The proper attitude is to treat it like a hypothesis, as something under investigation, something based on incomplete information that can see some things but not everything. But, as should be clear from this thread, that's not what happens. Instead, they take the view that it should be trusted because, well, because it's a stat that they believe has fewer problems than other stats. In other words, it's the least-objectionable stat. That is not a good reason to trust it. Even worse, some (not all, but some) act like people who don't accept that, who question whether we should put high levels of trust in the least-objectionable stat, are somehow Neanderthal idiots for not trusting it. That is not a good way to treat people who question unproven and evolving metrics. It's the same kind of thinking that had some doctors doing lobotomies on people who they thought were crazy, had them using massive electric shocks to scramble the brains of people who were depressed (way different than current techniques). The attitude was "it's state of the art, so we should trust it". Meanwhile, lots of other doctors were yelling "Wait a minute! You don't know what you're doing!", and it's the latter group who were right.

Doing D-stats right likely requires data acquisition methods beyond what are currently in place. The current stat-of-the-art boils down to best guesses based on only partial information, but some people treat them as if they are actual facts about D-goodness. There is no reason to conclude that they are. When they become adequate, one way we'll know is that most everybody will agree with them.

Anyone who treats UZR or +/- as gospel is looking at it and using it incorrectly. Anyone who rebuts it with this...

However, when players like Markakis and Jones come up with below average UZR/150 numbers then it really points out that the system is flawed and has a ways to go before it can be relied on as a true measure on defensive ability.

...is even more incorrect.

[TommyD4207]

We know what Batting Average, OBP and Slugging percent shows. For better or worse we have used them for a period of time and understand their strength and weaknesses.

UZR/150 is an evolving stat. It is unclear how good it is. We don't really know what the impact of Jones expanding his range in center has on Markakis's URZ/150 in right. There is no way to measure that as far as I can tell. There is OOZ, another new stat but the interact between players is not really measured. Just the chances, either in or out of zone.

Stat guys value UZR/150. It should be very apparent to everyone that baseball professionals do not for the most part. They believe what they see. They believe their eyes and the opinions of people they trust.

Stat guys may think that their system is better but until it is established with some kind of explanation for how it coordinates with what knowledgable baseball professional see, I don't look it for it to trust in many baseball circles.

Good point.

Couple years back, Rate on Baseball Prospectus was the stat that everyone used as much as people today use UZR and UZR/150. Everyone used it to shoot down observations and it was regarded as truthful as UZR is today.

Well, for BP Rate, it has Adam Jones at 105 (meaning that he had an above average season).

So there's that.

[RShack]

Anyone who treats UZR or +/- as gospel is looking at it and using it incorrectly. Anyone who rebuts it with this...

...is even more incorrect.

Do you believe that both Nick and AJ performed like sub-standard OF'ers this season? Do you believe that both of them abruptly became worse than average?

[DrungoHazewood]

This is right. The problem with iffy evolving stats, which is what we have with D-stats so far, is that some stat guys take the position that they should be trusted unless you can prove otherwise. That's the main problem: they take some iffy and evolving thing and act like it's the preferred way to look at things. They act like the onus is on others to prove an iffy thing is wrong. The proper attitude is to treat it like a hypothesis, as something under investigation, something based on incomplete information that can see some things but not everything. But, as should be clear from this thread, that's not what happens. Instead, they take the view that it should be trusted because, well, because it's a stat that they believe has fewer problems than other stats. In other words, it's the least-objectionable stat. That is not a good reason to trust it. Even worse, some (not all, but some) act like people who don't accept that, who question whether we should put high levels of trust in the least-objectionable stat, are somehow Neanderthal idiots for not trusting it. That is not a good way to treat people who question unproven and evolving metrics. It's the same kind of thinking that had some doctors doing lobotomies on people who they thought were crazy, had them using massive electric shocks to scramble the brains of people who were depressed (way different than current techniques). The attitude was "it's state of the art, so we should trust it". Meanwhile, lots of other doctors were yelling "Wait a minute! You don't know what you're doing!", and it's the latter group who were right.

Doing D-stats right likely requires data acquisition methods beyond what are currently in place. The current stat-of-the-art boils down to best guesses based on only partial information, but some people treat them as if they are actual facts about D-goodness. There is no reason to conclude that they are. When they become adequate, one way we'll know is that most everybody will agree with them.

Even if you're right, and the newer metrics are just best guess based on partial information, I'll take that 100% of the time over the much worse guesses that pass as traditional defensive rankings. If UZR is a guess, the Gold Glove selection methods are pulling names out of a hat in a dark room where you're not even sure there are hats.

Maybe UZR is only as good as, say, RBI. But RBI is a lot better metric than "swing prettiness", which is the offensive equivalent how most defenders are graded by writers, managers, and coaches.

[Lucky Jim]

Do you believe that both Nick and AJ performed like sub-standard OF'ers this season? Do you believe that both of them abruptly became worse than average?

I do believe that both of them performed poorly in the field this year, a performance that improved some over the course of the year. Thus, both UZR and +/- confirmed my eyewitness account.

I'm not sure what my response would be if both metrics had said that they played well. I'd probably be inclined to mitigate my harsh assessment.

Here's my assessment in late May of the most disappointing aspect of the season so far: http://forum.orioleshangout.com/forums/showpost.php?p=1760733&postcount=37http://forum.orioleshangout.com/forums/showpost.php?p=1760733&postcount=37

That had nothing to do with UZR, because the sample was far too small. See: http://forum.orioleshangout.com/forums/showpost.php?p=1728268&postcount=10