Article: Twins Must End Revolving Door At Shortstop

[Riverbrian]

If anything suggests that Valencia is good defensively... It is wrong!

[snepp]

I've never understood the theory that something which everyone says is inaccurate in small samples is suddenly accurate if you pile enough of those inaccurate small samples on top of each other.

 

That seems to me like saying "I know 1 + 1 doesn't equal 3, but if I add 1 + 1 = 3 enough times, I'm sure it will be correct."

 

A large sample size doesn't get accurate by adding together a bunch of inaccurate small samples.

 

Drew Butera has hit incredibly well over short stretches of time, but given a large enough sample, it inevitably regresses to his true level of putridity. It's no different for something like UZR, odd things can happen in small samples.

[Brock Beauchamp]

No, it says more about the misuse of a small sample than it does about UZR.

 

 

And he was +6, not 13, in said sample (roughly the equivalent of one month's worth of PA's).

 

I've never understood the theory that something which everyone says is inaccurate in small samples is suddenly accurate if you pile enough of those inaccurate small samples on top of each other.

 

That seems to me like saying "I know 1 + 1 doesn't equal 3, but if I add 1 + 1 = 3 enough times, I'm sure it will be correct."

 

A large sample size doesn't get accurate by adding together a bunch of inaccurate small samples.

 

http://en.wikipedia.org/wiki/Law_of_large_numbers

 

Roll a six-sided dice once and you may get a six. Terribly inaccurate if you're trying to average a dice roll.

 

Roll that same die 10,000 times and that average moves to 3.5. The roll of six was not representative of a typical dice roll but it still counts. Other times, you'll counter it with a roll of one if you repeat the process enough times. At the end of the day, you'll end up where you expect to be: 3.5. An accurate representation of an "average" dice roll.

 

For your "1+1=3" comparison to work, you have to assume that all methods of recording defensive metrics are wrong. If that's the case, so be it... I'm not going to argue that point with you because it would take days and math I don't care to calculate. But if you believe them to be *generally* accurate, then the Law of Large Numbers applies.

[Guest USAFChief]

I've never understood the theory that something which everyone says is inaccurate in small samples is suddenly accurate if you pile enough of those inaccurate small samples on top of each other.

 

That seems to me like saying "I know 1 + 1 doesn't equal 3, but if I add 1 + 1 = 3 enough times, I'm sure it will be correct."

 

A large sample size doesn't get accurate by adding together a bunch of inaccurate small samples.

 

Drew Butera has hit incredibly well over short stretches of time, but given a large enough sample, it inevitably regresses to his true level of putridity. It's no different for something like UZR, odd things can happen in small samples.

 

Except nobody says, "I know the numbers say Butera hit .350 over the last month. But that's not what happened. I know he's a bad hitter, therefore that couldn't have happened."

 

But that's exactly what the inventor of UZR, and others, say about defensive metrics.

 

Either Valencia saved 6 runs that year, or the metric is extremely unreliable at best, worthless at worst. Which is it?

 

BTW, do you agree with the idea (as implied above) that Valencia's UZR was positively influenced by playing next to Hardy? Or, as I've seen stated here that Span's UZR benefitted from Revere? If that's the case, wouldn't that be another reason to question whether UZR actually measures anything?

[Brock Beauchamp]

I've never understood the theory that something which everyone says is inaccurate in small samples is suddenly accurate if you pile enough of those inaccurate small samples on top of each other.

 

That seems to me like saying "I know 1 + 1 doesn't equal 3, but if I add 1 + 1 = 3 enough times, I'm sure it will be correct."

 

A large sample size doesn't get accurate by adding together a bunch of inaccurate small samples.

 

Drew Butera has hit incredibly well over short stretches of time, but given a large enough sample, it inevitably regresses to his true level of putridity. It's no different for something like UZR, odd things can happen in small samples.

 

Except nobody says, "I know the numbers say Butera hit .350 over the last month. But that's not what happened. I know he's a bad hitter, therefore that couldn't have happened."

 

But that's exactly what the inventor of UZR, and others, say about defensive metrics.

 

Either Valencia saved 6 runs that year, or the metric is extremely unreliable at best, worthless at worst. Which is it?

 

BTW, do you agree with the idea (as implied above) that Valencia's UZR was positively influenced by playing next to Hardy? Or, as I've seen stated here that Span's UZR benefitted from Revere? If that's the case, wouldn't that be another reason to question whether UZR actually measures anything?

 

We've been through this sooooo many times, Chief. Nobody says "that didn't happen". They say "that was a statistical variance". It happened. Just as Butera hitting a homerun is not indicative of his usual performance, Delmon Young making an out-of-zone diving catch is not indicative of his usual performance. That doesn't mean it did not happen, it states that it is outside the statistical mean.

[Mike Sixel]

It is not a question of if it measures"anything". It is a question of how well it measures things. The formulas do not yet know how to factor in defensive shifts, or the effect of other players.

[Brock Beauchamp]

It is not a question of if it measures"anything". It is a question of how well it measures things. The formulas do not yet know how to factor in defensive shifts, or the effect of other players.

 

And those are absolutely valid criticisms of the metrics. I don't think anyone would argue that there isn't room for improvement. Defensive metrics are barely out of their infancy; they have a lot of room to grow and improve. Even I try to avoid relying on them too much. WAR is a good quick-and-dirty metric but I wouldn't base my entire opinion on the metric because of its reliance on somewhat flawed defensive ratings.

 

But we're off to a good start and the metrics continue to improve as they are refined.

[DaveW]

I've never understood the theory that something which everyone says is inaccurate in small samples is suddenly accurate if you pile enough of those inaccurate small samples on top of each other.

 

That seems to me like saying "I know 1 + 1 doesn't equal 3, but if I add 1 + 1 = 3 enough times, I'm sure it will be correct."

 

A large sample size doesn't get accurate by adding together a bunch of inaccurate small samples.

 

Drew Butera has hit incredibly well over short stretches of time, but given a large enough sample, it inevitably regresses to his true level of putridity. It's no different for something like UZR, odd things can happen in small samples.

 

Except nobody says, "I know the numbers say Butera hit .350 over the last month. But that's not what happened. I know he's a bad hitter, therefore that couldn't have happened."

 

But that's exactly what the inventor of UZR, and others, say about defensive metrics.

 

Either Valencia saved 6 runs that year, or the metric is extremely unreliable at best, worthless at worst. Which is it?

 

BTW, do you agree with the idea (as implied above) that Valencia's UZR was positively influenced by playing next to Hardy? Or, as I've seen stated here that Span's UZR benefitted from Revere? If that's the case, wouldn't that be another reason to question whether UZR actually measures anything?

Though I am not a fan of your UZR bashing, I will say that over the years I have grown to trust UZR less and less but it still has plenty of value when used to show the WHOLE picture of what a player does, this includes: Sabr stats, traditional stats, scouting reports and the good old eyeball test. Anyone who only uses UZR or WAR to claim a player is great is almost as foolish as someone who only uses RBI to state the same thing.

 

You are correct with Span, his UZR doubled by having Revere in the OF along side him, which makes perfect sense.

[DaveW]

It is not a question of if it measures"anything". It is a question of how well it measures things. The formulas do not yet know how to factor in defensive shifts, or the effect of other players.

 

And those are absolutely valid criticisms of the metrics. I don't think anyone would argue that there isn't room for improvement. Defensive metrics are barely out of their infancy; they have a lot of room to grow and improve. Even I try to avoid relying on them too much. WAR is a good quick-and-dirty metric but I wouldn't base my entire opinion on the metric because of its reliance on somewhat flawed defensive ratings.

 

But we're off to a good start and the metrics continue to improve as they are refined.

 

The fact is, until we get FieldFx defensive statistics will continue to be pretty flawed.

[Guest USAFChief]

 

We've been through this sooooo many times, Chief. Nobody says "that didn't happen".

 

We have been through this, and I've explained why "the law of large numbers" isn't applicable, when you don't know the accuracy of the data you're inputing.

 

As for "that didn't happen," From Mitchel Lichtman ("MGL" on various chat boards, the man responsible for UZR):

 

 

People often say something like, “Well, he had a +10 UZR last year, which means that he actually played well, even though he might be an average or even below average defender.” For example, Jeter had a very nice UZR in 2009, a decent one in 2008, and some terrible ones for many years prior to that. So, he is a perfect example of a below-average defender who played excellent defense last year and pretty good defense the year before, right? Well, maybe and maybe not. A player’s UZR does not necessarily tell you how he actually played just as it does not necessarily tell you what his true talent is. That is a very important point. It is not like we pulled a coin from our pocket and flipped it 100 times and came up with 60 heads (which is entirely possible, even though we presumably have a fair coin). In that case, we can safely say that, yes, we did in fact get 60 heads (Jeter did in fact play well last year), even though we know that the true heads percentage of our coin is around 50% (Jeter’s true talent at SS is very likely below-average). UZR does not work that way. Why is that?That is because it is not measuring something that is categorized, like a coin flip which either comes up heads or tails, or BA, whereby a player either gets a hit or he doesn’t, or even simple Zone Rating, where a fielder either fields a ball within his zone or he doesn’t.

[kab21]

I like using multiple seasons of UZR to determine if a player is excellent, above average, average, below average or awful defensively but I have issues with its weighting in WAR equations and I don't like WAR much as a result.

 

It would be great to address the SS issue but it's just not going to happen. The Twins have limited resources to trade or to spend and that needs to focus on starting pitching since that is a far bigger issue than addressing SS longterm. I am hoping that the Twins at least get a guy like Drew, YEscobar or Kelly Johnson to at least give the Twins one competent MI'er instead of a herd of utility players.

[old nurse]

No, it says more about the misuse of a small sample than it does about UZR.

 

 

And he was +6, not 13, in said sample (roughly the equivalent of one month's worth of PA's).

 

I've never understood the theory that something which everyone says is inaccurate in small samples is suddenly accurate if you pile enough of those inaccurate small samples on top of each other.

 

That seems to me like saying "I know 1 + 1 doesn't equal 3, but if I add 1 + 1 = 3 enough times, I'm sure it will be correct."

 

A large sample size doesn't get accurate by adding together a bunch of inaccurate small samples.

 

Edited. Should have read the whole thread. Can somebody send Chief a book like "statistics made easy" ?

[Brock Beauchamp]

That is a very poorly written paragraph. Jesus. I had to read it four times just to understand where he was going with it. Do you have a link to the full transcript or is that all there is to it?

 

Anyway, there are so many variables to UZR that the Law of Large Numbers applies even more than most statistics. When you're splitting the field into 80 (or whatever) zones, factoring out homers, line drives, infield flies, etc., then also factoring in pitcher GB rate, pitch speed, and a myriad of other factors, you're going to need a really large sample size for the metric to work.

[Guest USAFChief]

Though I am not a fan of your UZR bashing, I will say that over the years I have grown to trust UZR less and less but it still has plenty of value when used to show the WHOLE picture of what a player does, this includes: Sabr stats, traditional stats, scouting reports and the good old eyeball test. Anyone who only uses UZR or WAR to claim a player is great is almost as foolish as someone who only uses RBI to state the same thing.

 

You are correct with Span, his UZR doubled by having Revere in the OF along side him, which makes perfect sense.

 

I'm "bashing" something that both you and Brock seem to admit isn't worth relying on? That both of you seem to admit can be "doubled" for a centerfielder by having a faster rightfielder? That the inventor of said metric himself says "it doesn't mean what it says it means."

 

Sheesh. Doesn't sound like bashing to me.

[Guest USAFChief]

That is a very poorly written paragraph. Jesus. I had to read it four times just to understand where he was going with it. Do you have a link to the full transcript or is that all there is to it?

 

I'm assuming you're addressing the Lichtman quote. It's from the Fangraphs "UZR Primer:"

 

http://www.fangraphs.com/blogs/index.php/the-fangraphs-uzr-primer/

[Guest USAFChief]

BTW...I'm curious...was Danny Valencia 6 runs better than the average AL third baseman over that sample size, or was he not? Or don't we know? Isn't that the crux of the issue?

 

Seems to me if you believe in UZR, you're obligated to believe, at least for that period of time, that Valencia was an above average third baseman.

 

Otherwise, we're back to "well no, but add 1+1=3 enough times...".

[old nurse]

 

We've been through this sooooo many times, Chief. Nobody says "that didn't happen".

 

We have been through this, and I've explained why "the law of large numbers" isn't applicable, when you don't know the accuracy of the data you're inputing.

 

As for "that didn't happen," From Mitchel Lichtman ("MGL" on various chat boards, the man responsible for UZR):

 

 

People often say something like, “Well, he had a +10 UZR last year, which means that he actually played well, even though he might be an average or even below average defender.” For example, Jeter had a very nice UZR in 2009, a decent one in 2008, and some terrible ones for many years prior to that. So, he is a perfect example of a below-average defender who played excellent defense last year and pretty good defense the year before, right? Well, maybe and maybe not. A player’s UZR does not necessarily tell you how he actually played just as it does not necessarily tell you what his true talent is. That is a very important point. It is not like we pulled a coin from our pocket and flipped it 100 times and came up with 60 heads (which is entirely possible, even though we presumably have a fair coin). In that case, we can safely say that, yes, we did in fact get 60 heads (Jeter did in fact play well last year), even though we know that the true heads percentage of our coin is around 50% (Jeter’s true talent at SS is very likely below-average). UZR does not work that way. Why is that?That is because it is not measuring something that is categorized, like a coin flip which either comes up heads or tails, or BA, whereby a player either gets a hit or he doesn’t, or even simple Zone Rating, where a fielder either fields a ball within his zone or he doesn’t.

 

And here is more from the same kind of post

 

UZR tries to record a player’s likely true talent and estimate his future performance based on the nuances of the batted ball and the player’s response to those nuances. It is not trying to capture exactly what happens on the field according to some arbitrary categories, like most of the offensive metrics (which make no distinction between a lucky ground ball bleeder through the “5-hole” or a clean, line drive base hit to the outfield), even the advanced ones like wOBA or linear weights.

 

Sorry it is not a black and white number for you.

[Brock Beauchamp]

I'm "bashing" something that both you and Brock seem to admit isn't worth relying on? That both of you seem to admit can be "doubled" for a centerfielder by having a faster rightfielder? That the inventor of said metric himself says "it doesn't mean what it says it means."

 

Sheesh. Doesn't sound like bashing to me.

 

I'm not admitting that Span's UZR was doubled because of Revere. Ben almost certainly had an impact but the bulk of Span's "improvement" likely came from being healthy.

[kab21]

BTW...I'm curious...was Danny Valencia 6 runs better than the average AL third baseman over that sample size, or was he not? Or don't we know? Isn't that the crux of the issue?

 

Seems to me if you believe in UZR, you're obligated to believe, at least for that period of time, that Valencia was an above average third baseman.

 

Otherwise, we're back to "well no, but add 1+1=3 enough times...".

 

Was Valencia really a .311 hitter or was he not? Or don't we know? Isn't that the crux of the issue?

[Brock Beauchamp]

That is a very poorly written paragraph. Jesus. I had to read it four times just to understand where he was going with it. Do you have a link to the full transcript or is that all there is to it?

 

I'm assuming you're addressing the Lichtman quote. It's from the Fangraphs "UZR Primer:"

 

http://www.fangraphs.com/blogs/index.php/the-fangraphs-uzr-primer/

 

Jesus, Chief. This is the very next paragraph:

 

The analogy with BA is, just because a player had a .334 BA does not mean that he hit the ball well. It is entirely possible the only reason that he hit .334 was because he got a lot of bloops and bleeders and most of his hard hit balls dropped for a hit. But, because we can verify that a player did indeed hit .334, we say that a player’s BA is a good record of what actually happened.

So he's not saying that he didn't play good defense, he's simply saying that the fielder was not challenged. It's akin to Joe Mauer getting to face AAAA pitchers for the entirety of September. Did he hit .450? Sure. Is that representative of how well he batted? Absolutely not.

[Guest USAFChief]

BTW...I'm curious...was Danny Valencia 6 runs better than the average AL third baseman over that sample size, or was he not? Or don't we know? Isn't that the crux of the issue?

 

Seems to me if you believe in UZR, you're obligated to believe, at least for that period of time, that Valencia was an above average third baseman.

 

Otherwise, we're back to "well no, but add 1+1=3 enough times...".

 

Was Valencia really a .311 hitter or was he not? Or don't we know? Isn't that the crux of the issue?

 

Yes, for that period of time, Valencia hit .311. Unequivocal fact.

 

See, that wasn't hard, now was it?

 

Now that we've dealt with your deflection, was Valencia 6 runs above the average AL thirdbaseman, or was he not?

[Brock Beauchamp]

BTW...I'm curious...was Danny Valencia 6 runs better than the average AL third baseman over that sample size, or was he not? Or don't we know? Isn't that the crux of the issue?

 

Seems to me if you believe in UZR, you're obligated to believe, at least for that period of time, that Valencia was an above average third baseman.

 

Otherwise, we're back to "well no, but add 1+1=3 enough times...".

 

Was Valencia really a .311 hitter or was he not? Or don't we know? Isn't that the crux of the issue?

 

Yes, for that period of time, Valencia hit .311. Unequivocal fact.

 

He hit .311 but that tells you virtually nothing about how he actually played. I could also look at a guy's walk rate and then declare he had a stellar season because he walked 52 times in 500 PAs.

 

But then I'd probably tell you that it was Jamey Carroll and you'd laugh me out of the room.

[E. Andrew]

The door should have stopped with Hardy. The day we picked him up nearly rivaled the stadium deal as one of my most exciting off-season days as a fan. I had spent several rediculously jealous years watching Hardy in Milwaukee. Stupid to let him go IMO. Few teams win without a solid SS, as Versalles and Gagne proved. Hopefully another will come along.

[Brock Beauchamp]

The door should have stopped with Hardy. The day we picked him up nearly rivaled the stadium deal as one of my most exciting off-season days as a fan. I had spent several rediculously jealous years watching Hardy in Milwaukee. Stupid to let him go IMO. Few teams win without a solid SS, as Versalles and Gagne proved. Hopefully another will come along.

 

The door should have stopped with Bartlett.

[kab21]

 

Yes, for that period of time, Valencia hit .311. Unequivocal fact.

 

See, that wasn't hard, now was it?

 

Now that we've dealt with your deflection, was Valencia 6 runs above the average AL thirdbaseman, or was he not?

 

To some extent for the same reason he was .311 hitter. A few diving plays and some easy grounders hit to him and he makes more plays than an average 3Bman. I think the runs saved part of the equation is a little questionable since it's more of a plays made data set though.

[Guest USAFChief]

That is a very poorly written paragraph. Jesus. I had to read it four times just to understand where he was going with it. Do you have a link to the full transcript or is that all there is to it?

 

I'm assuming you're addressing the Lichtman quote. It's from the Fangraphs "UZR Primer:"

 

http://www.fangraphs.com/blogs/index.php/the-fangraphs-uzr-primer/

 

Jesus, Chief. This is the very next paragraph:

 

The analogy with BA is, just because a player had a .334 BA does not mean that he hit the ball well. It is entirely possible the only reason that he hit .334 was because he got a lot of bloops and bleeders and most of his hard hit balls dropped for a hit. But, because we can verify that a player did indeed hit .334, we say that a player’s BA is a good record of what actually happened.

So he's not saying that he didn't play good defense, he's simply saying that the fielder was not challenged. It's akin to Joe Mauer getting to face AAAA pitchers for the entirety of September. Did he hit .450? Sure. Is that representative of how well he batted? Absolutely not.

 

1. He's saying that Jeter had good UZR numbers because he played SS in the major leagues for TWO YEARS and "wasn't challenged?" A) that's ridiculous, and how would he know that?

 

2. I would say if Joe Mauer hit .450 for a month that is EXACTLY representative of what his BA was. EXACTLY. Now you can parse the numbers, or the situation, and say he won't keep that up, because... . And you'd be right. But factually, he hit .450. Combine that with the next exactly accurate small sample of his BA, and the next, and the next, and you get an exactly accurate representation of what Joe Mauer hit, for any sample size you choose to construct.

 

So we're back the question, was, or was not, Danny Valencia better than average for that period of time as a third baseman? And if not, why should we trust the next small sample size? or the next? And why should we believe that adding together innacurate small samples leads to an accurate large sample?

[DaveW]

I'm not admitting that Span's UZR was doubled because of Revere. Ben almost certainly had an impact but the bulk of Span's "improvement" likely came from being healthy.

 

Dude what on earth are you talking about? We are talking about how he doubled his UZR in 2011/2012 vs what he was putting up yearly in 2008-2010. In 2008-2010 he was healthy almost the entire time, 2011-2012 is when he had injury issues. Unless you think his concussion issues made him magically have a doubled UZR over his career average it is easy to assume that his UZR yearly average doubled 95% because of the fact that Revere was manning the OF with him.

[Guest USAFChief]

I'm not admitting that Span's UZR was doubled because of Revere. Ben almost certainly had an impact but the bulk of Span's "improvement" likely came from being healthy.

 

Dude what on earth are you talking about? We are talking about how he doubled his UZR in 2011/2012 vs what he was putting up yearly in 2008-2010. In 2008-2010 he was healthy almost the entire time, 2011-2012 is when he had injury issues. Unless you think his concussion issues made him magically have a doubled UZR over his career average it is easy to assume that his UZR yearly average doubled 95% because of the fact that Revere was manning the OF with him.

 

Perhaps he went 2 years without being challenged.

[old nurse]

[

 

2. I would say if Joe Mauer hit .450 for a month that is EXACTLY representative of what his BA was. EXACTLY. Now you can parse the numbers, or the situation, and say he won't keep that up, because... . And you'd be right. But factually, he hit .450. Combine that with the next exactly accurate small sample of his BA, and the next, and the next, and you get an exactly accurate representation of what Joe Mauer hit, for any sample size you choose to construct.

 

So we're back the question, was, or was not, Danny Valencia better than average for that period of time as a third baseman? And if not, why should we trust the next small sample size? or the next? And why should we believe that adding together innacurate small samples leads to an accurate large sample?

 

By the numbers Valencia played well for a short stretch of time. Whatever motivated him to do so is clearly ancient history. As the sample size of his defense grew larger, his overall predictor of play became clearer. That is why you do not look at several small samples. You look at the whole pile of statistics of the playing career.

 

Is a Joe Mauer batting average against AAAA pitchers in September indicative of his talents against an ace?

 

Is what any September call up does 100% reliable for predicting future batting average? Brian Dinkleman batted .305 in 2011. Batting averages in spring training. By spring training numbers, Sean Burrows should have been a star.Small sample sizes. Look at the at bats versus the top line pitchers they faced would give a better indication of the talent. The talent is better assessed when you have more data. That is very basic, first concept, simple concept of statistics.

[Brock Beauchamp]

1. He's saying that Jeter had good UZR numbers because he played SS in the major leagues for TWO YEARS and "wasn't challenged?" A) that's ridiculous, and how would he know that?

 

2. I would say if Joe Mauer hit .450 for a month that is EXACTLY representative of what his BA was. EXACTLY. Now you can parse the numbers, or the situation, and say he won't keep that up, because... . And you'd be right. But factually, he hit .450. Combine that with the next exactly accurate small sample of his BA, and the next, and the next, and you get an exactly accurate representation of what Joe Mauer hit, for any sample size you choose to construct.

 

So we're back the question, was, or was not, Danny Valencia better than average for that period of time as a third baseman? And if not, why should we trust the next small sample size? or the next? And why should we believe that adding together innacurate small samples leads to an accurate large sample?

 

1. He's saying that Jeter's situation and the low frequency of "determining" statistics (which is going to happen when you're dealing with a large number of variables and conditions) made Jeter look better than he actually was in reality. Conditions were favorable to him looking good as a fielder and there weren't enough "outlier" opportunities to draw a meaningful conclusion. Given the low frequency of certain situations that determine a player's "range", that's going to happen over the course of a season, even to shortstops (and definitely to outfielders, who receive even fewer opportunities to show their defensive acumen).

 

2. And I can list you fielding percentages. It doesn't tell you much of anything on its own. Yes, those fielding percentages "happened". It doesn't mean they're of much use to anyone. At the end of the day, looking at batting average is going to tell you just that, batting average. Which, on its own, is of limited use. UZR is attempting to do much more than that. It's trying to factor in dozens of things and give an accurate representation of what the fielder does in the field. Unfortunately, all those variables mean that you need A LOT of opportunities to draw any sort of meaningful conclusion... Again, we're back to the Law of Large Numbers.

 

Was Danny Valencia better than average defensively over that period of time? I have no idea, just like you'd have no idea if he faced scrub pitching to attain that .311 batting average. In the end, what do you prefer? A flawed metric that tries to take into account all possibilities faced by a player or a boolean metric that tells you only a fraction of the story? Me, I'll take the flawed metric that tries to explain the entire situation.