Mike Silva's New York Baseball Digest » Blog Archive » Former Gold Glove 1B on UZR

Former Gold Glove 1B on UZR



By Mike Silva ~ August 15th, 2009. Filed under: Mike Silva.

Readers of the site continue to demand that I educate myself on modern statistical analysis. In order to do so,  I wanted to see what those in the game think of various statistics. Yesterday it was former GM Jim Duquette and today its LA Dodgers 1B Doug Mientkiewicz over at his twitter. Why not ask a gold glove 1B what his thoughts are on UZR and fielding metrics.

Another waste of time u can NOT equate someone’s range with an equation we are forgetting too many things in today’s game. Instincts r something u can’t teach or equate. You either have them or u don’t. To me it’s very important more so than any stat.

Why did I ask Mientkiewicz this question? Because I looked up Mark Teixeira’s UZR rating. Apparently Tex has a negative UZR rating this season. Anyone who has watched him play knows that is complete nonsense. He very well could be the best defensive 1B in baseball. As Tyler Kepner of the NY Times said at his twitter last night,  “UZR says Teixeira is below average at 1B, which completely negates that stat for me”. I concur Tyler. According to UZR, as of tonight, Ryan Howard has had a better defensive year than Mark Teixeira. Now Howard has improved, no doubt, but can you take this seriously? Better yet, is Daniel Murphy a better 1B than Teixeira? Well according to UZR he is. In case you are wondering, Teixeira had an 11 UZR last season.

A former GM and gold glove 1B have enlightened us. I will continue to search for advice and improve my learning’s of statistical decision making. I still say, numbers are fine, but your eyes never deceive you. By the way, for those that do this type of numbers crunching for a living, how do you incorporate the “instincts” that Mientkiewicz was talking about into a defensive formula? 

Post to Twitter Post to Yahoo Buzz Post to Delicious Post to Digg Post to Facebook

The following two tabs change content below.
Mike Silva has hosted sports shows on 107.1 FM Champions ESPN Radio Long Island ,1240 AM WGBB , Blog Talk Radio and live from Mickey Mantle’s Restaurant. He’s also built and maintained two popular social media hubs: New York Baseball Digest and Sports Media Watchdog. Mike has broken national and local stories, as well as been mentioned on the YES Network, SNY.tv, WFAN, Sports Illustrated, ESPN, NY Daily News, New York Magazine, Journal News and the NY Post. Contact Mike professionally at mikesilvamedia.com

Latest posts by Mike Silva (see all)

13 Responses to Former Gold Glove 1B on UZR

  1. Pat Hobby

    I don’t care how many letters his name has you should learn how to spell Mientkiewicz before you write about him.

  2. MGL

    I am the “inventor” of UZR. To answer your question, the only thing that matters in terms of the value or impact that a fielder has on his team is how many balls he gets to and turns into an out, as compared to an average fielder at that same position, given the exact same distribution of batted balls. Whether that comes from “instincts,” or from speed, or from quick reactions, or from positioning, or from the alignment of the stars, is irrelevant, other than from the standpoint of intellectual curiosity (some people like to know WHY certain fielders are better than others, and of course that might be important in terms of player development and the like).

    And that is what all of these advanced metrics measure – what percentage of balls a fielder turns into outs as compared to what an average fielder would do given those same types (direction, speed, etc.) of balls, and given the types of batters at the plate, and even the base runners and outs. For example, let’s say that there were 500 ground balls near Doug M. in a given season. And let’s say that we know the approximate location and speed of those ground balls. And let’s say that we know how many base runners were on base (and which base or bases) and how many outs there were for each of those ground balls. We can look at the last 5 years or so and we can see how many of those same ground balls (given the outs, batters, and base runners) were turned into outs by ALL first baseman. That basically gives us a baseline of the AVERAGE first baseman. Now we can compare that to how many of those same ground balls that Doug M. fielded. If Doug fielded (turned into outs) MORE than the average first baseman (again, given the parameters of those ground balls and the game situation), then he likely is an above average first baseman. If he fielded fewer, then he is likely a below average first baseman.

    However, because we are only getting a sample of a fielder’s talent (one season, less than a season, or even 2 or 3 seasons) AND because we have some measurement error and other biases (such as a ball we thought was right down the line was actually 5 feet from the line, or a ball we thought was hard hit was only hit at medium speed, or a ball took a bad hop or took a charity hop, which we don’t measure), in the short run, the numbers can tell a “lie” just like offensive numbers can tell a “lie” in the short run.

    Even in the long run (say 3 or 5 seasons) defensive or offensive numbers can tell a lie, for the above reasons. However, the larger the sample of data we have, the less chance there is of the numbers telling a lie. But even the best metrics will tell lots of lies especially with small samples. So the fact that Texeira has a negative UZR this season by no means suggests that UZR is a bad metric. It is guaranteed that even the best metrics will “lie” for a small percentage of players. Again, the smaller the sample, the higher the percentage of lies there will be.

    That is why if we want to evaluate someone’s fielding or offense, we generally want to use as much data as possible, to reduce the chances that the data will “lie.” Of course, we like to put more weight on more recent data, because player talent can change due to age, injury, learning, and other reasons which we may or may not be aware of. But still, we generally want to use as much data as possible. Looking at a player’s numbers, be it defensive or offensive, for less than one season in order to make a statement about them, can lead to a lot of “lies.” If we have more seasons of data on that player, we definitely want to look at that data in aggregate. Looking at it on a season to season basis is dangerous because of the likelihood of small sample size error.

    For example, to look at Teixera’s UZR for 2009, which is almost zero by the way (-.4 runs), and then proclaim that UZR is not a good metric because we think that Teixera is an excellent fielder, is disingenuous to say the least. His combined UZR since 2003 is almost +15 runs, which is pretty good for a first baseman. First baseman don’t get that many chances per game or per season, so we don’t expect there to be a tremendous difference in terms of theoretical runs saved or cost among first baseman, even very good and very bad ones. You can only be as good or bad as the number of opportunities you have to field a ball.

    There are other things of course that fielders do to help their team save runs (or not). For first baseman, being able to catch errant throws, like throws in the dirt, is one of those things. UZR does not measure that, but there are other metrics that do. In fact, if you read my article on “first baseman scoops” you will see how I can do that as well by comparing the number of errors that infielders make with one first baseman and all other first baseman. As it turns out, it appears that the difference between the best and worst first basemen at catching errant throws is only 3-4 runs per year at the most.

    As far as your comment on “the eyes don’t lie” or some such thing: If “the eyes didn’t lie” we would not need ANY metrics (BA, ERA, RBI, WARP, UZR, OPS, etc.) in order to evaluate players. The truth is the exact opposite. The eyes do lie, for various reasons. That is not to say that the “eyes” (visual scouting) should not be used to complement statistical analysis. They should and they do. That is because the metrics lie as well, as I explained above. The less data we have, the more we should rely on visual scouting, and vice versa (the more data we have, the less we should rely in it). That is because the metrics lie MORE with less data and LESS with more data.

    The other thing about visual scouting is that even if we concede that it is very accurate (for example, we can easily identify the best and worst fielders), what it cannot do is quantify that talent or value. For example, how many times have you heard a scout or manager or GM say something like, “So-and-so is great on defense – he saves us at least a run a game?” Well, from statistical analysis, we can determine exactly how many runs a good or bad defensive player saves or costs his team. That is relatively easy. And what we find is that the best SS and CF can save their teams around 20 runs per season (and the worst cost their team the same) and the best 1B and 3B, around 10 runs. Without these advanced defensive statistics, we would not know that. There is NO way that the eyes can quantify that.

    Anyway, I could write a lot more about this, but I’ll leave it at that.

  3. Mike Silva

    MGL

    Thank you for contributing to the discourse. By the way, do you have a website?

    I agree that we need both to make a decision. Stats should illuminate a position and I am not sure UZR does that. I think fielding is so much harder to judge in metrics versus offense. I am interested in your “scoops” analysis. Could you point this to me? I would like to read more into it. I believe 1B are undervalued defensively, especially in modern metrics.

  4. MGL

    http://www.insidethebook.com

    http://www.fangraphs.com/blogs/index.php/first-basemen-scoops

    How do you know or why do you believe that first basemen are “undervalued?”

    What “decision” do I need to make?

    How do you know whether UZR illuminates a position or not? Do you know exactly how it works or what it measures?

    If you think that fielding can only be evaluated by your “eye” than the discussion is over, right? NO metric, whether you understand it or not, is going to overcome that position.

    I’m not particularly interested in debating this issue with you. I wanted to shed some light on it for your readers. This may sound arrogant, but you are not a sabermetrician or a statistical analyst I assume. You do not have the credentials to debate this particular issue (the merits of an advanced defensive metric) any more than I have the credentials to debate cosmology or quantum physics with Stephen Hawking. (I am not comparing myself to him – he is WAY smarter than I am.) With all due respect.

  5. Mike Silva

    No, not looking to debate. Want to learn more about this stuff. The decisions I am referring to is with respect to teams and how much credence they give certain metrics in decision making. Not in a debate scenario.

  6. MGL

    I worked for the Cardinals for 2 years. They did and do use lots of these metrics to evaluate players for contracts, personnel moves, drafting, and other decisions.

    As far as I know, many other teams (TBA, SEA, PIT, OAK, BOS, SDN, NYN, et al.) do the same.

    I read the thread on BBTF and the comments are spot on. Other than the, “It would be like bringing in MGL for defense…” In my younger days, I was a pretty good defender!

    No team eschews scouting and “visual evaluation.” Some teams eschew a certain level of statistical evaluation. Those teams are at a disadvantage, although it does not necessarily show up in the standings. The reason it doesn’t necessarily show up in the standings is that money (spent on salaries and player development) and random chance play a significant role in a team’s w/l percentage. For example, let’s say that good statistical analysis added 3 wins to a team’s w/l record per season, which would be marvelous – the equivalent of replacing an average player with a superstar. If we add 3 wins to the Pirates, what do we get? A last place team with 3 extra wins (so, maybe a 71-win team rather than a 68-win team). Similarly, if we take a team that spends enough money to “buy” 93 wins (say, a Yankees or Mets) and they have virtually no statistical analysis (remember we are operating on the assumption that perfect analysis adds 3 wins – for this exercise), what do we get? A 90-win team, which most people would consider outstanding. So we have a 71-win team with great analysis and a 90-win team with no analysis. How are we going to use w/l records a gauge for whether statistical analysis “works?” We can’t.

    Similarly, let’s say that a team that uses very little analysis, like the Twins, has a great scouting and player development program and staff and is very successful in the w/l and post-season column. Does that mean that statistical analysis is not necessary or that traditional scouting and player development is “better?” Of course not. Maybe they would be 3 wins better if they added cutting edge statistical analysis.

    Keep in mind one very important thing! Sabermetrics is the science of uncovering the TRUTH in baseball, whatever that truth may be. By definition, it HAS to “work.” And, you have some of the smartest minds on the planet who are sabermetricians. That is a pretty powerful combination (a science which whose goal is to uncover the truth and people who are brilliant who work at that science).

    That is not to say that everything you hear and read that is packaged as “sabermetrics” is beyond reproach or that sabermetricians don’t make mistakes. That would be unrealistic in any curriculum.

  7. Rally

    “I read the thread on BBTF and the comments are spot on. Other than the, “It would be like bringing in MGL for defense…” In my younger days, I was a pretty good defender!”

    Sorry about that then. What was your UZR per 162? (regressed to the mean, of course:-)

  8. MGL

    Well, if we apply defensive MLE’s to the level I was playing at, probably around -75!

  9. Mike Silva

    MGL

    Did you read this SI article.

    http://vault.sportsillustrated.cnn.com/vault/article/magazine/MAG1153911/1/index.htm

    I know your quoted in there, but It seemed that Stl and LaRussa didn’t embrace new statistics as much as one would have liked.

    Would you ever be open to talking about some of this stuff on a future show?

  10. Metsies

    “Would you ever be open to talking about some of this stuff on a future show?”

    Ah yes, and the truth finally comes out. Silva is trolling here in attempts to gain some more contacts or whatever (Mientkiewicz, Kepner, MGL) and maybe get a few more blog hits and radio show guests. Why anyone is still paying attention to this (I’m looking at you BBTF) is odd.

  11. Mike Silva

    James – always a pleasure having you here at NYBD!

    Btw- since I own and pay for this site, I guess I can pretty much do and say whatever I want, no? :)

  12. Metsies

    It’s true that it is your site, but the “It’s my site I can say what I want!” argument is a total last resort Mike. You were pretty handily picked apart in the various BBTF threads and the comments here and have you shown a total unwillingness to change your conclusions about sabermetrics.

    It’s really not worth discussing anymore, but I’m guessing you’ll try to milk this fictional “debate/divide” for a few more posts and page hits. Enjoy.

  13. Nick

    Mike – Tony LaRussa is an idiot. He may be very good at getting his players to play to the best of there abilities, as evidenced by the fact that the Cardinals generally overperform there projected records; however, he has absolutely no idea what he is talking about when it comes to in game strategy and player valuation.

    Neither did Walt Jocketty, who was the man who imployed MGL in the first place. Since MGL left, Walt has left, and the Cardinals front office has gotten much more sabermetrically involved.

    The head in player developement, Jeff Luhnow, uses a lot of sabermetric work and making his decisions:

    http://www.beyondtheboxscore.com/2009/4/7/824829/q-and-a-with-jeff-luhnow

    You’ll note that he praises the work of MGL and his colleague, Tom Tango.

    And he has turned rebuilt the Cardinals system to be in the top half in baseball, after it was one of the worst during the Walt Jocketty days.

    The Red Sox have Bill James on payroll, the Mariners have Tom Tango, the Indians have Keith Woolner. So please, tell me that Sabermetrics aren’t valued by management.

Leave a Reply

You must be logged in to post a comment.