Monday reading: MGL on outfield defense

DMZ · October 29, 2007 at 9:00 am · Filed Under General baseball 

Check it out.

There’s some interesting tangents in there, particularly in the infield positions. Also, it’s cool to see that in working out the fair territory area, it does appear that Seattle (at ~109,000 sq ft) is not that large a field at all.

Assuming that Safeco’s marked fence dimensions are accurate, of course. I haven’t surveyed the field myself.


21 Responses to “Monday reading: MGL on outfield defense”

  1. Mike Snow on October 29th, 2007 9:20 am

    Yes, that was a pretty good analysis. Interesting conclusions, and still quite a bit of room for further research. I’d especially like to see more work with an expanded dataset, to see for example whether the outlier parks still increase UZR at both extremes.

  2. Gomez on October 29th, 2007 9:44 am

    Awesome stuff to read. I would think that the Safeco dimensions are accurate, keeping in mind that the RF porch is as shallow as the LCF porch is deep.

  3. smb on October 29th, 2007 11:42 am

    “t does appear that Seattle (at ~109,000 sq ft) is not that large a field at all.”

    So THAT’S why Cirillo couldn’t find the gaps! All this time I just thought he sucked.

  4. dcmarinerfan on October 29th, 2007 11:51 am

    #3 – Yea, both him and Rich Aurilia both. Man, that could have been a heck of a left-side-of-the-infield if we’d just have kept him the C-man around for another year.

    Do you think Bavasi might end up reading this and thinking to himself, “Hmmm, well, Safeco isn’t very big, so we can do with slower than average outfielders. Let’s keep Ibanez out there.”

    Oh, hold that thought. Bavasi probably can’t read, or at least chooses not to.

  5. King Rat on October 29th, 2007 11:55 am

    Oy! I hope he gets paid for that analysis. I wouldn’t have the dedication.

  6. smb on October 29th, 2007 12:00 pm

    Oh he can read…he’s not a dumb guy, quite the opposite. Our problems come from above him, and his personnel decisions just tend to fit within the organizational philosophy, when what we really need is for some of them to buck the trends that have us middling in mediocrity.

    This article is a good example of the type of data the M’s seem to refuse to consider, like it’s too heady or something…and believe me, I would also be afraid of the FO drawing the kind of conclusion from it that you suggest. I would like to see them recognize a faster outfielder as an even bigger advantage in a smaller field, rather than using a smaller field as an excuse to perpetually run a mule out to LF to play D, as if it means he’s no longer a liability in the field.

  7. dcmarinerfan on October 29th, 2007 12:54 pm

    I don’t know how much faith I put in Bavasi’s intelligence. I grew up and went to high school with a pitcher who’s now speedily moving through the Mariner’s farm system.

    His [my friend’s] characterization of Bavasi, and to a lesser extent Armstrong are that they are both painfully unaware of most everything.

    That point aside, guys like Rich Aurilia don’t “fit within the organizational philosophy.” Unless of course your organizational philosophy is ‘getting over the hill players who don’t seem to fit by any statistical analysis into a winning system at Safeco or elsewhere.’

    Frankly, being a GM is hard. Anyone who says otherwise is a liar. They put in a lot of hours, pour through a ton of information and are constantly under pressure.

    So, I’m not saying I think anyone off the street could do a better job than Bavasi’s done, but you can’t blithely stand beside him and just accept that it’s problems from higher-ups that give us trouble.

    There are easily 100 Mariners fans, both here, and at LL, that know the organization front and back, know the quality statistical tools that are helpul at making quality analysis of talent, understand projected performance, and much more.

    Watching a guy like Epstein win the World Series with the Red Sox, after having drafted and developed players like Papelbon, Lester, Ellsbury, Pedroia, Youkilis, Manny Delcarmen and probably soon to be World Series winners Clay Bucholz, Kyle Snyder, it’s frustrating as all heck to know that we are led by a seemingly spineless, and I would argue senseless General Manager.

    But, here’s to progressing to the mean.

  8. galaxieboi on October 29th, 2007 12:58 pm

    Thanks for the link, Derek. That is a badass amount of research. Whew.

    Yeah, this may be one article I don’t want Buzz Jr. getting a hold of.

    “See! Raul’ll be fine. It’s not that he’s craptastic defensive player, he’s just sssllllllooowwww…”

  9. smb on October 29th, 2007 1:33 pm

    Well, I don’t know if your pitcher buddy is more Greg Maddux or Jose Paniagua as far as his mental acumen, so forgive me for taking his analysis of Bavasi’s intelligence with a grain of salt, not having met him myself. My defense of Bavasi begins and ends with making sure Chuck and HoLinc get their share of the blame as well. Call it my assumption that gritty veteran-ness and promotions dept-friendly player personalities perhaps aren’t solely Bavasi trademarks of earned value found through personnel analysis. Don’t get me wrong, I’m no fan…I still get sick thinking of Jones at AAA and Soriano for HoRam, but to me those are symptoms of the disease, not the whole disease itself.

    If they were, I doubt he’d have the balls (or sense) to show up to USSM feeds and try to cultivate any kind of positive relationship with the fans here.

    Some good Chuck quotes have been posted here in the past week or so to illustrate how fucked we are philosophically, top to bottom. Blaming Bill for all the current ills is kinda like blaming Rice for the Iraq War. Solely responsible? No. Complicit? Most certainly yes.

    Now, what I REALLY want to know is, given the dimensions of the field and the relative short porch in left, have they been seeing Raul as the answer to the need for a left handed power bat all along? This article re-raises all sorts of questions for me about the stadium we built to our own specifications, and how we may (or may not) be trying to field a team that turns our home park into a competitive advantage in terms of the dimensions (not portals).

    I would say, for example, that the data suggests the slower outfielder has not so much an advantage in a smaller outfield, as a reduction of his inherent disadvantage. To me, that’s not the same thing, because the faster player doesn’t become less useful in a smaller outfield. At least, I don’t see how you can penalize the value of a faster guy in a smaller outfield just because he can cover every square inch and then some. He still potentially creates the maximum amount of defensive outs from his position, while the slower guy, while performing better than he would in a bigger outfield, most likely does not. So it’s no validation of “hiding” Raul in left…even if his lack of range is mitigated by a smaller outfield, Jones is the better option defensively there in ANY size outfield.

  10. Mat on October 29th, 2007 1:36 pm

    I kind of wonder if classifying parks into “large,” “medium,” and “small” is just too simplistic. Safeco is a good example of this. A large LF and relatively small RF lead to an overall medium park. Considering that it’s a pretty small data set in the first place, it seems like this could be clouding the results significantly. It’s still a thought-provoking look at the data, though.

  11. Chris Miller on October 29th, 2007 2:02 pm

    #10, I was thinking that when I read the article.

  12. Mike Snow on October 29th, 2007 2:35 pm

    I agree that the classifications are simplistic, but given the way the metric was generated, I’m not sure there was a better way. Trying to use finer-grained classifications might well suggest too many conclusions that aren’t really related to the difference in size.

    With more information incorporated, like fence heights, maybe it would be possible to develop a metric one might call “adjusted outfield size.” Anyway, I look forward to any further work along these lines.

  13. dcmarinerfan on October 29th, 2007 3:07 pm

    After re-reading the study, now for a third time, I am struck by a couple of things:

    1. The Mariners play 108 games (roughly) each season between at Texas, Oakland and Anaheim, and in Seattle.
    2. Each of these stadiums (Tex 112k, Oak 111k, ANA 111k) is average to above average in total size (with the exception of Safeco, but the difference is skewed towards a small right field).
    3. The “advantage” that slow outfielders have over fast outfielders appears when you focus on playing in smaller than average to average sized ballparks
    4. The Mariners play a vast majority of their games in average to larger than average ballparks (if you call Safeco a larger than average ballpark – which, for the point of any Ibanez in Left argument, you have to do).
    5. The “disadvantage” to having a slow outfielder or outfield in general is VERY apparent when a team plays the majority of its games in average to larger than average ballparks.
    6. Raul Ibanez cannot be regarded as anything better than a slow outfielder.
    7. Rual Ibanez playing left field more than 10 times a season puts the Mariners defense at a significant disadvantage.

    Anyone beg to differ? I’m looking at you John McLaren.

  14. dcmarinerfan on October 29th, 2007 3:39 pm

    One caveat to the conclusions made in this study:

    The author seeminly arbitarily picks 106k and 116k as the breaking points for smaller than average and larger than average fields.

    Yet, the mean size of ballparks in baseball, as reported by this study, is 109,900 sq. ft.

    The mean size of ballparks in the American League (where the Mariners play a vast majority of their games) is 109,357 sq. ft.

    The median size of all ballparks in baseball, is 109,500 sq. ft.

    The median size of ballparks in the AL is 108,500 sq. ft.

    Thus, having “average sized” ballparks be 106-116k sq. ft. seems a bit skewed to me.

    Nine ballparks in this study are deemed below average size. Four of them in the AL, five in the NL. Six ballparks are deemed above average size, two of them in the AL and four in the NL.

    I wonder if the conclusion, or at least the values popped out by the metric would be significantly altered if the true “mean” or at least a different one (one maybe closer to the actual geometric mean) were used as the base standard for judging which ballparks were above average in size, or below average in size.

    Just a thought.

  15. Broadcast James on October 29th, 2007 3:50 pm

    I felt like the second part of the article can only be classified as “interesting.” There’s too many variables to make any concrete conclusions. I wanted to note that perhaps the increased UZR in bigger parks may be do to teams with bigger parks employing better fielders – and also I believe balls that would be HR’s in smaller parks might be caught in a larger park adding a positive mark for the fielder.

    Defensive metrics work better on a per-team basis. I wonder how the study would look if it was done on a team-by-team basis.

    Great work, better link, thanks guys.

  16. Eastside Crank on October 29th, 2007 10:25 pm

    My take home message is that speed is a good thing for each position except for third base. I do not think that the speed metric takes into account the routes outfielders take to get to balls and their ability to get a jump on the ball. I am not sure how UZR accounts for the arm strength of outfielders and their ability to throw runners out. For infielders, the length of the infield grass can be a great equalizer as long as the players have strong arms (yes I have a certain book). A player with poor range can simply play deeper and rely on his arm strength to throw batters out. Overall it is an interesting analysis that gives us another way to evaluate Bavasi’s moves this winter.

  17. Evan on October 30th, 2007 9:21 am

    Assuming that Safeco’s marked fence dimensions are accurate, of course.

    Those would be easy to confirm. Google Earth.

  18. feingarden on October 30th, 2007 11:03 am

    The author is also assuming that all the stadium diagrams he used were the same scale. If that’s not the case the whole ball of research goes out the window.

  19. Evan on October 30th, 2007 3:08 pm

    Though it would be easy to tell if they were the same scale. The distance from base to base should be fixed.

  20. joser on November 2nd, 2007 11:43 am

    I kind of wonder if classifying parks into “large,” “medium,” and “small” is just too simplistic. Safeco is a good example of this. A large LF and relatively small RF lead to an overall medium park. Considering that it’s a pretty small data set in the first place, it seems like this could be clouding the results significantly. It’s still a thought-provoking look at the data, though.

    That’s a valid point. When evaluating fields as “hitter’s” or “pitcher’s” parks we’ve progressed to the point where we talk about a park being a “right-handed hitters’ park” or “plays as a pitcher’s park against left-handed batters” so that’s the obvious next step in research like this. The configuration may be even a little more complicated than that, because the area an outfielder can cover is basically a rough disk with the radius being determined by their jump off the ball and their leg speed (obviously the disk is going to be elongated coming in vs going out, but a disk is a reasonable approximation). But a disk is going to map onto some left and right fields better than others (and then you have places like Fenway where being able to field the caroms is at least as important). But that gets really complicated, really fast and it’s unclear if we need that level of detail anyway. (Someday we’ll have the position and velocity data to track all balls in play as they pass through cubic meter volumes of space, and will be able to evaluate not only how well every fielder handles balls of every sort of trajectory, but how well they position themselves prior to the ball being put in play). As you say, I do think we need the detail to divide it up into left, right, and center.

  21. joser on November 2nd, 2007 11:48 am

    The author is also assuming that all the stadium diagrams he used were the same scale. If that’s not the case the whole ball of research goes out the window.

    From the site:

    Generally speaking, the diagrams are accurate to within a few feet, i.e., one or two pixels. As a scholar and as a member of the Society for American Baseball Research, I aspire to uncover historical facts and share my findings with others who have a like interest….Estimates for the position of the fences and walls lining the playing field are more accurate than for the exterior walls. The scale is identical in all of the diagrams: 3 pixels = 5 feet.

    Obviously you’re looking at compounded errors when you’re talking about plus or minus a couple of feet along every dimension, but for the purposes of this study it’s probably good enough.

Leave a Reply

You must be logged in to post a comment.