FanGraphs Baseball

Comments

RSS feed for comments on this post.

  1. MGL – What conversion rate did you use for converting plus minus plays made to plus minus runs?

    Comment by Peter Jensen — February 9, 2009 @ 9:04 am

  2. You’ve got an error in your example: “Now let’s say that 20 throwing errors were made on those throws to Helton and only 15 were made to all the other first basemen. We can safely say (with some uncertainty of course, due to sample error) that Helton was 5 plays per 1000 throws (about a full season actually) better than the other pool of first basemen, who we are assuming are average.”

    If there were 20 errors on throws to Helton and only 15 errors on throws to other 1B, then Helton is below-average at scooping, not better than the other players.

    Comment by Brett — February 9, 2009 @ 10:19 am

  3. MGL – Also, what database did you use to determine who was the intended receiver on error throws? One of the very few details of a play that Retrosheet lacks is what base was being thrown to when a throwing error was made.

    Comment by Peter Jensen — February 9, 2009 @ 11:26 am

  4. This is helpful, MGL. The leaders and the trailers also easily pass the smell test, which should make this information more readily accepted.

    You might try renaming it “with and without you” for the puerile reference to the U2 song. Keep the masses happy with marketing…

    Comment by Mike Green — February 9, 2009 @ 12:02 pm

  5. There’s a good Beatles fan for ya, Mr. Mike Green. It’s “With Or Without You” but I’ll forgive him because I know he has Beatles on the brain.

    Comment by Ari — February 9, 2009 @ 12:19 pm

  6. A further query: catching errors seem to be omitted here, but catching-errors-on-a-throw are not included in UZR, are they?

    I realise that not all catching errors are on bad throws, but a significant number of them are: the catching/throwing error decision lies at least partially in the scorer’s whim, which would be nice to take out.

    Comment by Excalabur — February 9, 2009 @ 12:58 pm

  7. Is this a fair grading system, when two players who are being judged are primarily being judged against one another? D. Johnson and N. Swisher for instance spent most of their scooping opportunities while on the A’s sharing mostly the same infielders exclusively for them on the A’s. If one player is a above average scooper, he would negatively effect the other’s scooping ability to unfair degree.

    Comment by Nick — February 9, 2009 @ 2:34 pm

  8. It is called With Or Without You.

    Comment by TangoTiger — February 9, 2009 @ 2:46 pm

  9. Nick – MGL spends a fair amount of time discussing this shortcoming of the WOWY method in the article.

    Comment by Peter Jensen — February 9, 2009 @ 3:06 pm

  10. Probably could also run into the mind set of the scorer being biased because the first basemen has a good reputation as a fielder.

    Comment by Scappy — February 9, 2009 @ 3:08 pm

  11. For V1.0 this is a hell of a lot better than what i could come up with. Would it be possible to included the all star games and the WBC to spice up the data any?

    Comment by Scappy — February 9, 2009 @ 3:09 pm

  12. I simply MUST know how JT Snow measures on this…

    Comment by zenbitz — February 9, 2009 @ 3:21 pm

  13. Seconded. It was like the #1 excuse for why the Giants should have kept Snow for as long as they did. “He scoops throws so well!”

    Comment by Chris — February 9, 2009 @ 3:37 pm

  14. Click on my name, and look at post #2 for career leaders.

    Comment by TangoTiger — February 9, 2009 @ 4:17 pm

  15. Oops… looks like Rally’s site is caput at this moment.

    Comment by TangoTiger — February 9, 2009 @ 4:18 pm

  16. Do you mean “Within You Without You” :)

    Mike Green was making a U2 reference, as he said!

    Comment by John — February 9, 2009 @ 4:20 pm

  17. Yes, UZR includes all errors, including receiving throws. Errors in UZR are treated separately from “range” and then everything is combined at the end. A lot of the PBP or semi-PBP systems treat errors the same as a missed ball. They are not! Scorer judgment and bias aside, an error is much worse than a missed ball. A missed ball is a ball that may or may not have been playable, even though the UZR program thinks it should have been caught some percentage of the time. An error is a ball which, according to the scorer at least, should have been caught 98% (or whatever the average fielding % at that poition is) of the time. If we treat an error as a missed ball, it might be a ball that, according to ther various parameters and buckets, UZR thinks is fielded 20% or 50% or 70% of the time (or whatever).

    Anyway, the sata set I used was STATS which includes who receives every ball.

    And yes, Nick, as Peter indicated, I explained that you are right in that each player is actually compared to whoever happened to receive a lot of throws from the same fielders, which tends to be their teammates. I suspect that even when I adjust for that, the numbers won’t change that much. If it is not too much trouble, I’ll do that soon.

    This was a quickie synopsis of some work I was doing the other night. Hope that the results were helpful. I think that the illuminating part was the magnitude of the difference between the best and worst, which this kind of methodology should pick up (and keep in mind that sample results always, or at least usually, pick up a larger spread than than the “true” spread, just as if you looked at the difference between the best and worst BA in any one or even two-year period).

    Some time ago, people were guessing that “scooping” was worth a few runs a year, and that seems to be about right. Maybe a run or two more since some scoops save hits rather than errors.

    Thanks for the comments. I always like to “vett” these methodologies to get ideas for making them better.

    I can post the entire results on Google Docs if anyone would like.

    Comment by MGL — February 9, 2009 @ 9:11 pm

  18. Anyway, the sata set I used was STATS which includes who receives every ball.

    Thanks for this info. It is very cool that STATS does this. I wish that that information could become a part of Retrosheet.

    It still would be interesting to me to see how you go from plus minus plays made to run values. An example would be great. Thanks.

    Comment by Peter Jensen — February 10, 2009 @ 8:25 am

  19. If I have understood the process correctly, there may be a underestimation of the value of scooping. On a ground ball in the hole or down the third base line followed by a throw in the dirt, but on line and barely in time, the most common scoring results are “out” if the first baseman makes the scoop and “infield single” if he does not. I understand that these plays are effectively not counted because of the absence of “error”, as a likely scoring possibility.

    I would guess also that the height advantage might be lessened if these plays were taken into account.

    Comment by Mike Green — February 10, 2009 @ 10:53 am

  20. A possibility already specifically acknowledged in MGL’s reply above.

    Comment by Peter Jensen — February 10, 2009 @ 11:33 am

  21. Yup, as Peter said, I think I mentioned that one can assume that the scoop values are a little low due to some bad throws that are scored as hits and not errors. I don’t think it will make much of a difference though. Maybe another run or so for the best or worst.

    Peter, an error is worth around .5 runs and an out around .27. I used .8 runs per “difference between an out and an error,” the same as you would use for a regular defensive system that tracked missed balls versus outs. I suppose that a throw to first suggests that no other runners on on base unless there are 2 outs, so maybe the value of an error on a bad throw to first is closer to .4 runs. So maybe .7 runs is more accurate. No big deal though.

    And whoever mentioned that my Helton example was backwards, you are correct of course. Thanks for pointing it out.

    Comment by MGL — February 10, 2009 @ 8:54 pm

  22. I am new to all this so I hope I am not asking too basic a question, but doesn’t STATS keep a record of actual scoops? Couldn’t the actual scoops be used in place of this formula?

    Comment by Julio — February 11, 2009 @ 12:06 pm

  23. I suppose that a throw to first suggests that no other runners on on base unless there are 2 outs,

    I don’t know why you would suppose that. Infielders hold runners on 2nd and/or 3d all the time and throw to first with less than 2 outs. They also throw to first when runners are in motion and there is no chance for a play elsewhere. But you are correct that the run value for an error on the throw is closer to .7. Not for the reason you gave, but because an error on the throw can either put the batter on or not and can also either advance the batter beyond first or not. The run values for these four buckets vary from .35 to .75 and average .445. That added to the average run value of an infield out ground ball of .258 gives .703. As you said, no big deal, your conclusions are still the same. However, dividing into 4 buckets might result in a slightly greater spread between the very worst and very best first basemen as the very best may be able to prevent the batter and other runners from advancing more often than the very worst first basemen.

    Comment by Peter Jensen — February 11, 2009 @ 1:31 pm

  24. First of all, why is everyone saying this measures scooping ability, when it really just measures % of outs made vs. opportunities. Richie Sexson rates well based strictly on his reach. The fact that his fielding percentage is lower than most would tell you that he’s probably worse at scooping low throws as well. These stats are interesting, but mostly it tells us two things we already knew: 1) lefties have an advantage at 1b because their “reaching” foot is naturally closer to the bag and glove hand is naturally closer to the rest of the field, while righties have to pivot 180 degrees to get in proper stretch position, and 2) tall guys make an easier target. Assuming Sexson and Snow both have arm lengths in proportion to their height, Sexson’s reach gives him a target radius somewhere between 6″ and 1′ longer than Snow’s, which would give him a huge advantage in the number of square feet he can cover while keeping his foot on the bag. I don’t know if anyone actually keeps a stat of throws in the dirt that are actually scooped by a 1B, but I would bet that the percentage of successful scoops more closely mirrors the players fielding percentages.

    Comment by Ben Marshall — February 11, 2009 @ 5:11 pm

  25. One more thing in defense of JT Snow: It was his overall defense, not just his scooping ability, that made him valuable. He was a great fielder, and a great thrower. A lot of 1Bs wouldn’t even try to start a double play, or throw home on a close play, but Snow excelled at throwing players out at the other bases. Of all the 1Bs I’ve ever seen, most are either just a little better or a little worse than average. Snow was the only one that ever really stood out to me. And I’m not a huge Snow fan. I would have traded him in a second for an average fielding 1B who’d hit 30-40 HRs a year. But he was a great fielder.

    Comment by Ben Marshall — February 11, 2009 @ 5:17 pm

  26. Peter, as I said, a throw to first with less than 2 outs SUGGESTS that there are no other runners on base. Unless I am on crack or completely lost my mind, I certainly did not say or think that a throw to first with less than 2 outs ALWAYS means that there are no other runners on base.

    If I told you that “there was a throw to a base” or I told you that “there was a throw to first base” (with less than two outs), which bucket would you think had more men on base, on the average?

    That being said, if it is closer to .7 runs per “swing” (error or out), that’s fine by me. I don’t think it will make any (other than de minimus) difference at all, keeping track of what “bucket” the error is in, to be honest.

    Ben, from the article:

    “However, in the long run, we can assume that a certain fixed percentage of bad throws from each infield position will always result in an error while another fixed percentage of those bad throws have a chance to be “saved” by a skilled (OR TALL) first baseman.”

    “What UZR does not measure for first basemen, because the requisite data is not readily available, are the theoretical runs saved or cost by virtue of a first baseman’s skill at successfully catching ERRANT THROWS OR THROWS IN THE DIRT. FOR LACK OF A BETTER WORD, I CALL THESE “SCOOPS”, EVEN THOUGH THEY NECESSARILY INCLUDE POOR THROWS THAT ARE NOT IN THE DIRT.”

    You said:

    “These stats are interesting, but mostly it tells us two things we already knew…”

    You could have saved me a few hours if you told us exactly how much this information “that we already knew” was worth. 1 run a year? 10 runs?

    You also could have saved me some time if you told me which players, independent of their height and handedness, were particularly good or bad at saving errors. There are tall lefties in the data who were not good and short righties who were good. But apparently you knew that already.

    You are right in that a few of the things that UZR or “scoops” does not capture is a first baseman’s skill at starting the DP (although I do have a metric which looks at that, which is debuting on Fangraphs soon – “turning the DP by infielders”) and throwing to other bases, like home plate or perhaps to second or third on a bunt. But you probably know that too. Can you share with us all of the players who are good or bad at that (other than J.T. Snow) and exactly how many runs that kind of skill is worth?

    ;)

    Comment by MGL — February 12, 2009 @ 1:14 am

  27. Ben Marhsall made a comment of JT Snow excelling at throwing runners out at other bases. Are there stats that show a percentages of outs/safe on Fielders choices? Are some 1B (or IF) more aggressive in trying to start a Double Play or getting the lead runner? How successful are they? How many runs saved or cost based on thier decisions?

    Comment by Jason B — May 30, 2009 @ 11:42 am

  28. Wonderful website, I recently came across it and I’m already a fan. I recently dropped thirty pounds in 30 days, and I want to share my weight loss success with as many people as possible. If I can lose weight then any one can. Whatever you do, never quit and you WILL achieve every one of your weight loss objectives!

    Comment by Whitley Shad — July 31, 2010 @ 10:45 pm

Leave a comment

Line and paragraph breaks automatic, e-mail address never displayed, HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>


Close this window.

0.250 Powered by WordPress