THE BOOK--Playing The Percentages In Baseball

That was an excellent read tango. Just one comment about Jeter’s defense. I actually added this comment when filling out your “Fans Scouting Report”.

Jeter is a bad defensive player, relative to other shortstops. Anyone with any common sense knows it. However, the shortstop position is one of the game’s most demanding. Thus, even the worst shortstops (Jeter, Michael Young, etc.) are probably at least average when compared to MLB players as a whole. For every Adam Everett, who is vastly superior to Jeter as a defender, there is a Jason Giambi, or a Ryan Braun, or a Ryan Howard, or a Hideki Matsui, or someone else who is inferior to Jeter.

Also, Jeter’s defensive shortcoming are limited to one aspect of his game (range). His hands and his arm seem to be very good. Although range is probably the biggest factor in whether or not the defender makes the play, it is actually undervalued by your fans’ scouting report.

If I recall correctly, there were seven categories, three of which were concerned with range (reaction to the crack of the bat, acceleration, sprint speed), three on throwing (power, accuracy, and footwork), and one on hands. So, even if Jeter were judged on a non-biased manner by Yankees fans, his strong points (hands and arm) get disproportionate credit as compared to his weak point (range).

Thanks.

***

Actually, not all 7 categories are equally weighted.  And for each position, they would be differently weighted.  I could have rated his ability to grip a baseball, hold a baseball, follow through on a throw, and point to the 1B after the throw, and it would not have made a difference, since the overall throwing weight will not change, whether I have 3 or 13 categories.

***

The Fans’ view is generally that Jeter is an average to slightly-below average fielding SS, which would make him an average to slightly-above average fielder overall.

I agree that I can’t subscribe to an evaluation of Jeter as a fielder if it requires that we only consider him as a SS.  He is basically “good enough” to play SS badly.  Giambi and Thomas et al are not “good enough” to even set foot at SS.

***

However, Jeter himself insists on playing SS, or at least did at the time of The Trade. Since the Yanks are scared to death of developing any in-house SS solution, there’s really no big push to get him out of there.  If that is the case, then he deserves the negative oomph that his play warrants him, even though he shouldn’t be there.

Ah ok, thanks for the clarifications. As for the article itself, it was really surprising seeing how low he ranked using so many different methodologies. I mean, I knew he was bad, but THAT bad? Wow…

Also, I just thought it would be interesting to point out that more and more Yankees fans seem to be recognizing Jeter’s defensive problems. Most people I know, many of whom either don’t know or don’t believe in sabermetrics, have been grumbling about Jeter’s defense lately. And even the sports radio hosts here in NYC, who rarely make any intelligent insights, have been talking about Jeter’s poor defensive play.

So, maybe the public perception is finally starting to catch up to reality.

One more question I was hoping you could answer: Is there a position on the field where you think Jeter would be more appropriate? Is his hitting too weak for a corner outfield spot, or first base? Would he be better-suited at 2B, 3B, or CF? I really don’t know what options the Yankees have…

I have alot of Yanks ballots.  While I won’t reveal exactly where he stands, he’s pretty much around a “50” overall, meaning an average overall fielder.  That’s going to be around a 5 or so point drop from last year, which itself was an 8 point drop from 2006.

So, I think the Fans are getting the message on Jeter finally.

(Same thing happened with Junior last year.  It takes a while, and when it does, the Fans really get down on him.)

As for where to put Jeter.  Five years ago, I said 3B or RF.  But, right now, since he’s, what 33?  34?  That’s not an age where IF go to the OF.  Jeter still has wheels though, so he might work out ok in LF or RF.  Otherwise, he should go to 3B or 1B.

Given the Yanks particular situation, 3B is out.  And the last thing you want is yet another corner OF / 1B / DH.

Presuming no SS on the horizon, sadly, his best position may indeed be SS.

Funny thing is, about poor defensive play of certain superstars, fans get the message when their OFFENSIVE game declines.

If Jeter’s offensive production was good enough like past years, I really doubt fans would complain about his defense.

Same thing can applied to Junior’s case.

#6, I completely agree.  As well, people tend to blindly follow the media.  Also, in general, there seems to be a lag from when a player’s defense declines according to the metrics and when the fans (and baseball insiders) and the media start to realize that.

Basically, there are a lot of obstacles to the fans, the media, and baseball insiders accurately assessing defense, which is to be expected as there are no defensive “stats” staring them in the face on a daily basis, as there is for offense, other than fielding percentage, and even many casual fans realize its limitations.

#6, I was going to say this same thing. Jeter is having a bad offensive year (by his standards, he’s still pretty good for a shortstop), so people are going to pick apart his entire game.

The other reason Jeter has been overrated defensively for a long time is that his signature play is a pretty, Web Gem type play. That running, leaping throw across the diamond looks really cool. It looks difficult. It makes him look better than he is overall. People constantly see that play and go “this guy is making a heck of a play all the time!” It’s similar to a meh outfielder who makes lots of diving catches (Eric Byrnes comes to mind).

Tango, I’ll be forwarding this article to my uncle, a very smart guy who is skeptical of a lot of sabermetric research for some of the same general reasons Jeter mentions in that quote (essentially: the numbers can’t account for all the quirks, so you can’t really trust them over your eyes).

Certainly a very interesting article, and a new way to look at Jeter’s play.  My concern with this, though, is that what if his poor percentages came from multiple factors?

For instance, lets suppose that he played in a park where less balls than average were hit to shortstop, and also had pitchers who had less balls hits to the shortstop.  Then when neutralizing park effects, he would look bad because of pitchers.  When neutralizing pitchers, he would look bad because of park effects.  (For the other comparisons, he would look bad because of both).

Although UZR should avoid all of this, I would think.

Daniel, there’s no question that you have a point.  The larger worry would be park, not pitchers.  Jeter is married to Yankee Stadium, but was not married to the 100 or whatever pitchers he’s played behind.

We can handle the park issue by topping off the number of PA at any park at say 5% of all his BIP.  You can similarly do the same thing with pitchers.  In this way, you are forcing him to have at least 20 pitchers and 20 parks, of which any one of them has at most 5% of the impact.

You can’t control for everything because of sample size issues, so you just have to hope and assume that there is not going to be a strong bias in one direction for multiple things.  No metric, including WOWY is going to be perfect for this and other reasons, which is why when dealing with sample data, NO MATTER WHAT, the best you can do is estimate a player’s true talent at anything.  There is no metric that can nail true talent when dealing with sample data of less than infinite size.  Pointing out where the sample error might lie is nice, but it doesn’t get you any further than, “Yes, I know my metric is not perfect.” Take it for what it is worth and know that some percentage of the numbers and conclusions will be wrong.  We can usually estimate the magnitude of our certainty in our metrics (for any individual player), so at least you can know or at least have some idea as to the chance that the metric could be wrong to any degree, and then take that for what it is worth as well.

No multi-variate ANOVA? No significance testing? No multiple regression with control variables?

JM: are you suggesting that the article would be better had I done that, or are you suggesting that the issue can be handled?

tangotiger:
My main issue is the depth of statistical analysis. For example, you use separate tests to compare Jeter’s PA% based on pitcher and based on park. Despite the face that we have separate tests demonstrating Jeter’s PA% is lower than other SS, that does not show Jeter’s PA% would be lower if all the variables were analyzed at the same time.

Let’s assume that a certain type of pitcher (say, Brandon Webb) produces a higher percentage of balls in play a specific type of park (say, Petco Park) than in another type of park (say, Coors Field). Regardless of the direction of that relationship, the mere fact there is a relationship threatens the validity of your analysis.

If there are 6 variables influencing the dependent variable (Jeter’s PA%) and you analyze each one in a separate setting, your conclusions may be wrong since the required interaction terms amongst the 6 variables are not included.

There are a number of other issues with the analysis:

Age of a pitcher. Perhaps an older Clemens gave up more balls hit to the outfield than a younger Clemens causing Jeter to have fewer attempts than the SSs during Clemens’ younger days.

Skill of other fielders, especially the first and second baseman. Perhaps Jeter, saddled with below average first and second basemen, had fewer putouts and assists recorded than comparable SSs playing with superior fielders.

Team. Perhaps Jeter played in an organization that encouraged pitchers to go after lots of strikeouts or had other preferences with their defensive schemes and pitching strategies. Over time, this may have caused a systematic bias in the number and type of defensive chances he received.

I’m not saying any or all of these are true, but without further analysis, I would not be comfortable with the results. I also happen to believe Jeter is a below-average fielder, but I have yet to see data which documents that.

JM, I don’t disagree with anything you say.  At the very least, this is the starting point, as it enumerates the major variables that affect the player’s numbers.

The problem is summarized with your final statement: “but I have yet to see data which documents that. “

MGL, with UZR, and Pinto, with PMR, and Shane Jensen with SAFE, try to handle all the parameters in one fell swoop, and agree that Jeter is cr-p.  However, alot of people don’t believe it because they handle things in one fell swoop, allowing people to utter a statement like you just did, that they have not seen the data.

Essentially, I take the data (step A), and present it through a first pass analysis (step B).  MGL and Pinto and Shane do all the merging of the parameters, and present it as Step Z.

There’s an incredible burden being placed here that we are required to present the results of the analysis at Step C, Step D, Step E… Step Y.  I’ve given you Step B, and they are giving you Step Z.  I doubt presenting C through Y will change anyone’s mind.

***

In my new WOWY, I split the pitchers into over/under 30.  (Clemens under 30 is considered a different pitcher than Clemens over 30.) This handles the aging issue better.  I’ll present the results based on that at some point. 

I can’t apply a basic aging curve, since someone will rightfully contend that RJ did not necessarily age the same way as Frank Tanana when it comes to ball in play.

***

As for skills of the other fielders, I also do that in my new WOWY.  For example, how does Mike Schmidt play with and without Larry Bowa, how do other 3B, etc.

Again, I *need* to keep them as one-off this way, otherwise if I start merging things, the entire simplicity of the method goes away, and it simply becomes another black box that no one will believe.

http://www.nypost.com/seven/11032008/sports/yankees/jeter_gets_no_glove_from_survey_136649.htm

“Maybe it was a computer glitch,” Jeter told The Post. “Every [shortstop] doesn’t stay in the same spot, everyone doesn’t have the same pitching. Everyone doesn’t have the same hitters running, it’s impossible to do that.”

Can someone be nice enough to send Jeter a copy of the THT08 Annual?

Wow, after a nice post, you conclude it with:

I also happen to believe Jeter is a below-average fielder, but I have yet to see data which documents that.

Do you mean that “you” personally have never read about or seen any of the good defensive metrics, some of which Tango mentioned? Because that is the only way that that comment makes any sense.  Are we supposed to care about “you” happen to read or not read, as if that means that it does not exist?  Now, if you say something like that and you add (or it is implied because we “know who you are"), “and I have read pretty much everything on the subject,” that would be a much different story.

Or if by “data which documents (supports)that” you mean the equivalent of the Dead Sea Scrolls from G-d, that has a section on Jeter, well, that would be a different story as well.  (I guess that wouldn’t be “data,” but if you can’t believe G-d, who can you believe?)

Again, while JM’s points are true and well-taken, I tire of the, “You cannot be 100% certain with your methodology or your samle size, therefore I don’t believe you.”

Yada, yada, yada.

News flash for those of you who don’t read my posts too often, or don’t have a photographic memory, or are just plain stubborn:

EVERY conclusion reached based on sample data (e.g. ANY record of Jeter’s fielding or any record of anyone’s performance) is speculative - IOW, it is not 100% certain.

What determines the certainty is the amount of the data and the accuracy of the methodology with respect to what you are trying to measure.

Of suggestions, criticisms, comments on any of the methodologies, all are welcome.  If you would like to tell everyone what everyone likely knows already when someone reaches a speculative conclusion (as they ALL are) which is particularly speculative due to a small sample of data, well…

However, if you like to say something like, “Well, I don’t believe you because your methodology is not perfect or the sample size of your data is less than infinitely large,” well, that’s a double “well...”