Minnesota Twins News & Rumors Forum
Page 2 of 4 FirstFirst 1234 LastLast
Results 21 to 40 of 65

Thread: Doumit ranks as worst pitch framer for 2nd straight week

  1. #21
    Senior Member All-Star
    Posts
    1,469
    Like
    66
    Liked 145 Times in 94 Posts
    Blog Entries
    30
    Quote Originally Posted by Reginald Maudling's Shin View Post
    But, they are giving updates by the week! That's crazy. I thought UZR was stupid because you need 3 years of data to formulate an opinion. Then people still extrapolate ridiculously small sample sizes out if it. This is orders of magnitude worse. I heard a good line from a statistician that seems appropriate here: if you want, you can torture the data until it tells you what you want it to say.
    i think we get updates by the at bat on a lot of other data. It isn't OK to get weekly updates on this data?

    In any case, this isn't UZR. It appears much more stable than many other commonly used measures and stats. The top and the bottom ranked catchers entering 2012, were the top and bottom ranked catcher in 2012 in spite of changing teams. How many other stats or measures have similar reliability? It does have the benefit of a very large sample compared to batting stats or UZR. While an outfielder may only see a handful of balls in their zone during a game, a catcher is going to receive many pitches.

  2. #22
    Owner MVP Brock Beauchamp's Avatar
    Posts
    8,017
    Twitter
    @rocketpig76
    Like
    47
    Liked 1,512 Times in 791 Posts
    Blog Entries
    6
    Quote Originally Posted by jorgenswest View Post
    While an outfielder may only see a handful of balls in their zone during a game, a catcher is going to receive many pitches.
    But only a few of them will be borderline strike/ball calls to the point where a catcher can try to influence the umpire's call.

    On top of that, in the case of Doumit, we're often talking about a single game of data. That means he catches exclusively for one umpire, whose zone could be anything from the inside half of the plate to the entire batter's box.

    To pile on another variable, a single game of data means that anywhere from 1/2 to 2/3rds of Doumit's data comes from a single pitcher, one whose control may be off (or spot on) that day and influences the umpire's zone heavily by it.

    I'm intrigued by pitch framing. I think the concept has merit. But to use one week's data for a part time player is bad statistical analysis, the same way using two weeks of UZR/150 is bad statistical analysis.

  3. #23
    Senior Member All-Star
    Posts
    1,469
    Like
    66
    Liked 145 Times in 94 Posts
    Blog Entries
    30
    What one week set of data would be useful in baseball?

    In any case, I think the data cmathewson referred to is a running total for the year (not the numbers for the week) just like any of the other number we see attributed to a player for 2013.

    As for Doumit, if it were an aberrant one week there would be nothing to write about. His data is consistent from 2007.

  4. #24
    Owner MVP Brock Beauchamp's Avatar
    Posts
    8,017
    Twitter
    @rocketpig76
    Like
    47
    Liked 1,512 Times in 791 Posts
    Blog Entries
    6
    Quote Originally Posted by jorgenswest View Post
    As for Doumit, if it were an aberrant one week there would be nothing to write about. His data is consistent from 2007.
    It doesn't matter. Partial data is still partial data.

    It reminds me of a paper I was reading a few weeks ago about the average lifespan of a man in Victorian England. The point the author was trying to make was that given the "perfect storm" of Victorian living (some medicine, no tobacco, clean air), that our lifespans had only marginally increased once you compensated for infant mortality. The author spent several pages trying to inject bad statistical analysis over good data (excepting infant mortality in Victorian times but not today, using a death chart based on age 65 expectancy for Victorian people and a death chart from birth in modern times, etc) and by the end, I stopped reading the paper because it was so obvious that the author was doing everything he could to prove his point instead of letting the data fall as it may, which still proved his point. He wasn't wrong and he had a convincing argument. But in his attempt to twist the data to better suit his argument, he only deterred me from reading further into the paper and left me feeling disgusted at his blatant attempts to "rig the game" when he already had it won.

    The point? There are five years of data showing that Doumit is bad at pitch framing. You're only diluting the point by hammering away at it on a weekly basis and giving people a reason to ignore the wealth of real data on the subject.

  5. #25
    Senior Member All-Star
    Posts
    1,469
    Like
    66
    Liked 145 Times in 94 Posts
    Blog Entries
    30
    Point taken.

  6. #26
    Senior Member Double-A
    Posts
    182
    Like
    0
    Liked 2 Times in 2 Posts
    Blog Entries
    2
    Quote Originally Posted by jorgenswest View Post
    What one week set of data would be useful in baseball?

    In any case, I think the data cmathewson referred to is a running total for the year (not the numbers for the week) just like any of the other number we see attributed to a player for 2013.

    As for Doumit, if it were an aberrant one week there would be nothing to write about. His data is consistent from 2007.
    If a guy is knocking the cover off the ball over a week I think that's a relevant stat if you're an opposing pitcher.

    But again, does it really take 6 years of data to tell us Doumit isn't a good defensive catcher? I think we already know that. Also if you are trying to validate the data, having the guy who's dead last remain dead last isn't necessarily proving a statistical trend.

    I don't know what one would do with pitch framing data, other than make your own player aware of the problem. But I don't think ranking players is valuable. I have the same problem with defensive metrics in general, if you need 3 years of data what good is the stat? Why can't you just have your scouts who've watched thousands of games tell you if the guy is a good defensive player or not? Are you going to use UZR on high school age players who play maybe 60 games a year?

  7. #27
    Senior Member Triple-A
    Posts
    425
    Like
    0
    Liked 42 Times in 31 Posts
    Blog Entries
    3
    I will be the first to admit that I think pitch framing is a minor catcher skill. At best "a skilled pitch framer" is very likely to steal a strike no more than once or twice a game, in my opinion. So I am not going to waste my studying data on this. But, if it is true that the same people are at the top and bottom of the list over a 6 year period regardless of the team they play for and changing pitching staffs, I would consider that a huge red flag concerning the validity of this data.

    There is no way that pitchers and umpires don't have huge effects on whether borderline pitches are called balls or strikes. If this data consistently gives the same catchers the same positions in the ratings, there is a good chance there is something wrong with how this data is collected. Especially if it is happening from week to week as suggested by the starter of this thread.

    I get pretty nervous when effort is made to separate out small parts of a team game and credit or discredit one player. In this case you have a pitcher, a catcher, a batter and ultimately an umpire who makes the call. Trying to figure exactly why an umpire missed a call, assuming he actually missed the call and it wasn't just poor camera angles or some other reason for poorly collected data, is going to be damn hard. Since generally, there aren't really as many "umpire misses" as seem to be suggested here, I doubt that pitch framing has much effect on the game. Still if Doumit is always on the bottom, and the same catcher is always on top, every week the data is collected, you better examine how that data is collected. There is really no way that can happen with that small of a sample.

  8. #28
    Speediest Moderator All-Star snepp's Avatar
    Posts
    4,039
    Like
    1,763
    Liked 1,238 Times in 492 Posts
    The data is collected through pitchf/x, not some schlub watching a game with a lousy camera angle.


    I don't see how certain players routinely grading poorly, while others routinely grading well, can be a "red flag." OPS doesn't lose its validity when the same hitters top the leaderboard year after year, with the same hacks bring up the rear. What am I missing here?
    Last edited by snepp; 05-05-2013 at 05:35 PM.
    "Maybe you could go grab a bat and ball… and learn something. Maybe you will get it."
    - Strib commenter educating the elitists on the value of RBI's

  9. #29
    Owner MVP Brock Beauchamp's Avatar
    Posts
    8,017
    Twitter
    @rocketpig76
    Like
    47
    Liked 1,512 Times in 791 Posts
    Blog Entries
    6
    Quote Originally Posted by snepp View Post
    The data is collected through pitchf/x, not some schlub watching a game with a lousy camera angle.


    I don't see how certain players routinely grading poorly, while others routinely grading well, can be a "red flag." OPS doesn't lose its validity when the same hitters top the leaderboard year after year, with the same hacks bring up the rear. What am I missing here?
    Yeah... While I'm very skeptical of framing's overall influence on a game, it's not a "red flag" that Molina, widely regarded to be a top defensive catcher by everyone with a pulse, scores very well in the metric and Doumit, widely regarded to be a pretty poor defensive catcher by those very same people with pulses, scores low.

    I'd say that's a pretty convincing argument that pitch framing exists. How much it influences runs per game can still be debated but there is SOMETHING there worth investigating.

    When a metric's raw numbers so closely align with reports from scouts, players, and managers, that's a good thing and lends weight to the validity of the numbers, not the other way around.

  10. #30
    Speediest Moderator All-Star snepp's Avatar
    Posts
    4,039
    Like
    1,763
    Liked 1,238 Times in 492 Posts
    Quote Originally Posted by Brock Beauchamp View Post
    How much it influences runs per game can still be debated but there is SOMETHING there worth investigating.
    Yeah, how much of that something is there, and if there's enough of it there to be significant, is certainly wide open for scrutiny.
    "Maybe you could go grab a bat and ball… and learn something. Maybe you will get it."
    - Strib commenter educating the elitists on the value of RBI's

  11. #31
    Senior Member Double-A
    Posts
    182
    Like
    0
    Liked 2 Times in 2 Posts
    Blog Entries
    2
    Quote Originally Posted by snepp View Post
    Yeah, how much of that something is there, and if there's enough of it there to be significant, is certainly wide open for scrutiny.
    The other thing I don't quite understand is what you do with the data. Is pitch framing something you can practice and get better at? I would think a guy like Molina just knows where to set up to get the call, knows the strike zone, and knows where to put his glove and not move it. It's seems more of an instinctive thing. Its like asking players to somehow learn to take pitches as close as say, Mauer does. That's just natural ability to some extent.

  12. #32
    Senior Member All-Star Willihammer's Avatar
    Posts
    2,755
    Like
    640
    Liked 380 Times in 212 Posts
    Blog Entries
    9
    Quote Originally Posted by Brock Beauchamp View Post
    I'm intrigued by pitch framing. I think the concept has merit. But to use one week's data for a part time player is bad statistical analysis, the same way using two weeks of UZR/150 is bad statistical analysis.
    In even 1 week of data the leaderboards are basically the same as what we have for larger studies that look at thousands of pitches. I don't know what the cutoff is where a sample of pitch framing data becomes reliable, it might be that in part time duty Doumit didn't reach it. But judging from the consistency of these weekly reports (so far anyway), the cutoff might be a lot less than something like UZR.

  13. #33
    Owner MVP Brock Beauchamp's Avatar
    Posts
    8,017
    Twitter
    @rocketpig76
    Like
    47
    Liked 1,512 Times in 791 Posts
    Blog Entries
    6
    Quote Originally Posted by Willihammer View Post
    In even 1 week of data the leaderboards are basically the same as what we have for larger studies that look at thousands of pitches. I don't know what the cutoff is where a sample of pitch framing data becomes reliable, it might be that in part time duty Doumit didn't reach it. But judging from the consistency of these weekly reports (so far anyway), the cutoff might be a lot less than something like UZR.
    It can't be accurate from week to week. It just can't. When a player is often playing only once a week under one umpire, that umpire's zone is going to distort results much of the time.

    Like defensive metrics, there are simply too many variables in pitch framing to use small sample sizes, even if those samples occasionally "look good" over the short term. The umpire has his zone (which we all know, varies wildly from umpire to umpire and from day to day) and the pitcher influences how many strikes/balls is called as well. Both of those factors are entirely ignored in pitch framing metrics.

    Hell, I don't even trust pitch framing metrics over one season yet. A good pitching staff (which a player is generally going to catch for an entire season) can distort results pretty badly, one would think. In a perfect world, you'd want multiple seasons from multiple teams to draw a meaningful conclusion about the catcher.

  14. #34
    Senior Member All-Star Willihammer's Avatar
    Posts
    2,755
    Like
    640
    Liked 380 Times in 212 Posts
    Blog Entries
    9
    Quote Originally Posted by Brock Beauchamp View Post
    IThe umpire has his zone (which we all know, varies wildly from umpire to umpire and from day to day) and the pitcher influences how many strikes/balls is called as well. Both of those factors are entirely ignored in pitch framing metrics.
    This one ignores them but these things have been measured in other pitch framing studies. There's about a 10 strike range between the stingiest umpires and the more free-strike calling ones (over 150 pitches). The effects of pitchers on strike calling are even smaller. The fact is, catcher pitch framing is as big or bigger than any other variable.

  15. #35
    Owner MVP Brock Beauchamp's Avatar
    Posts
    8,017
    Twitter
    @rocketpig76
    Like
    47
    Liked 1,512 Times in 791 Posts
    Blog Entries
    6
    Quote Originally Posted by Willihammer View Post
    This one ignores them but these things have been measured in other pitch framing studies. There's about a 10 strike range between the stingiest umpires and the more free-strike calling ones (over 150 pitches). The effects of pitchers on strike calling are even smaller. The fact is, catcher pitch framing is as big or bigger than any other variable.
    But they're still variables, unaccounted variables at that.

    And in small sample sizes, those are generally disastrous when trying to parse data.

  16. #36
    Senior Member All-Star Willihammer's Avatar
    Posts
    2,755
    Like
    640
    Liked 380 Times in 212 Posts
    Blog Entries
    9
    Quote Originally Posted by Brock Beauchamp View Post
    But they're still variables, unaccounted variables at that.

    And in small sample sizes, those are generally disastrous when trying to parse data.
    The sample sizes aren't that small when you think about it. 2 games of catching is about 300 pitches. If you want to compare that to UZR, that's better than a 1/2 season of chances for a everyday shortstop. A 1/2 season of defensive data is still variable but not totally meaningless. I suspect the same is true of catching data.

  17. #37
    Owner MVP Brock Beauchamp's Avatar
    Posts
    8,017
    Twitter
    @rocketpig76
    Like
    47
    Liked 1,512 Times in 791 Posts
    Blog Entries
    6
    Quote Originally Posted by Willihammer View Post
    The sample sizes aren't that small when you think about it. 2 games of catching is about 300 pitches. If you want to compare that to UZR, that's better than a 1/2 season of chances for a everyday shortstop. A 1/2 season of defensive data is still variable but not totally meaningless. I suspect the same is true of catching data.
    That's still a very small sample size. Just because it's not as unwieldy as defensive metrics doesn't mean it's reliable week-to-week, especially for part-time players.

  18. #38
    Twins Moderator MVP ashburyjohn's Avatar
    Posts
    8,573
    Like
    2,585
    Liked 3,199 Times in 1,699 Posts
    Blog Entries
    28
    Quote Originally Posted by Reginald Maudling's Shin View Post
    I would think a guy like Molina just knows where to set up to get the call, knows the strike zone, and knows where to put his glove and not move it.
    At least one trait you didn't mention is head movement. That seems like something that can be practiced (to not do).

    If a pitcher is tipping his pitches, you can work with him to achieve muscle-memory or whatever so that he does the same thing the same way every time. You could drill a catcher (hm, maybe not the best choice of phrase ) on the things relevant to his job in a similar fashion.

  19. #39
    Senior Member Double-A
    Posts
    182
    Like
    0
    Liked 2 Times in 2 Posts
    Blog Entries
    2
    Quote Originally Posted by Brock Beauchamp View Post
    But they're still variables, unaccounted variables at that.

    And in small sample sizes, those are generally disastrous when trying to parse data.
    Of those 300 pitches, how many qualify as borderline where catcher A gets a strike and B doesn't get the call? Then the sample size becomes much, much smaller.

  20. #40
    Senior Member All-Star
    Posts
    1,395
    Like
    53
    Liked 36 Times in 24 Posts
    Quote Originally Posted by Reginald Maudling's Shin View Post
    Of those 300 pitches, how many qualify as borderline where catcher A gets a strike and B doesn't get the call? Then the sample size becomes much, much smaller.
    As of 5/1 (so through roughly 20 games for a catcher) you're looking at 125-200 calls that were influenced by starting catchers. So it is somewhere between 5-10 pitches per game so far this year.

Page 2 of 4 FirstFirst 1234 LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  
©2014 TwinsCentric, LLC. All Rights Reserved.
Interested in advertising with Twins Daily? Click here.