FanGraphs Baseball


RSS feed for comments on this post.

  1. I’m floored, Dave. Thank you, thank you, thank you.

    Comment by Double06 — April 11, 2010 @ 3:58 pm

  2. Blengino got to you, huh?

    Comment by Jack Moore — April 11, 2010 @ 3:58 pm

  3. This is completely outstanding. Awesome. Thanks Dave, this site is the best!

    Comment by Larry Smith Jr. — April 11, 2010 @ 4:04 pm

  4. Great!

    My question: Is there any reason to use RZR when UZR is available?

    Comment by supermets — April 11, 2010 @ 4:04 pm

  5. I would say, not really, but OOZ can tell you something new.

    Comment by David Appelman — April 11, 2010 @ 4:06 pm

  6. I don’t know if it’s ever been addressed, but when Dewan released his Fielding Bible Volume 2, his outfield arm ratings were messed up.

    Due to what the stats are trying to convey, if you add up all the defensive stats for a position in a league, you get zero (or extremely close to it).

    This is not the case with the arm ratings. You get something like +150 at RF, +100 at CF, +50 at LF (I don’t remember the exact numbers, I had added them up when the book came out a year ago).

    Meaning that total plus/minus for OFers were incorrect because the arm runs weren’t runs above average. I mean, comparing two players was fine, i.e. if OFer A is five runs better than OFer B, it’s going to show that. But the +10 OFer wasn’t 10 runs above average, he was less than that.

    Comment by Nny — April 11, 2010 @ 4:07 pm

  7. I do not know how he calculates the +/- ARM ratings exactly, but you’re correct that they still do not add up to zero.

    I would imagine this is something which we could normalize to 0 on our end, but right now we’re really just publishing the numbers.

    Comment by David Appelman — April 11, 2010 @ 4:09 pm

  8. Just yesterday I was thinking of renewing my Bill James Online account.

    Comment by philkid3 — April 11, 2010 @ 4:22 pm

  9. What’s RZR scaled to? +/-, like UZR, is intuitive: positive good, negative bad. Not sure how we’re supposed to evaluate RZR.

    Comment by Kevin S. — April 11, 2010 @ 4:28 pm

  10. Dave, thank you so much for bringing these new stats to the site!

    One question, though- is it possible to also present scoop opportunities for first basemen as well? Otherwise, we’ve got no real context.

    Comment by J.T. — April 11, 2010 @ 4:31 pm

  11. Kevin,

    RZR is simply the rate of balls in a player’s zone converted into outs. You can change this into plays and runs above/below average by comparing the player’s RZR to the league average. UZR is a rate, but it is presented as runs saved or cost compared to the average player at the position.

    Let’s say, for example, we have a league average RZR at SS of .800. The player we’re looking at has 440 balls hit into his zone, and he converts .810 of them into an out. This means he’s made 356 plays. A league average SS, however, is expected to make .800*440 = 352 plays, meaning that our player has made 356 – 352 = 4 plays above an average shortstop. Converted to runs, that’s about +3 runs saved.

    This isn’t taking into account a player’s Out Of Zone (OOZ) plays, which can be done by looking at the player’s OOZ rate compared to balls hit in his zone.

    I hope that helps some.

    Comment by J.T. — April 11, 2010 @ 4:37 pm

  12. It’s awesome to see sabermetrics continue to grow like this

    Comment by firnie14 — April 11, 2010 @ 4:47 pm

  13. Thanks, this is great!

    I do want to point out though that there seems to be a problem on the leaderboards. If I look at 2009 outfielders I see that Nyjer Morgan appears near the top at +22. If I look at 2009 for all players then Morgan is nowhere to be found when he should be near the top. Might this glitch be related to the fact that he played for two teams last year?

    Comment by mickeyg13 — April 11, 2010 @ 4:48 pm

  14. That’s correct. Morgan isn’t “qualified” at either CF or LF in 2009, but for OF overall, he is. The “overall” leaderboards break out each player by individual position and doesn’t include “OF” overall as a position.

    Comment by David Appelman — April 11, 2010 @ 4:56 pm

  15. OK I’m confused…what is the difference between rPM and +/-? I thought that +/- was scaled to plays, not runs?

    Comment by mickeyg13 — April 11, 2010 @ 4:59 pm

  16. rPM is +/- runs above average. +/- is just the sum of all the stuff that comes before it (rSB + rBU + rGDP + rARM + rHR + rPM). Maybe it needs to be relabeled if it’s causing confusing.

    Comment by David Appelman — April 11, 2010 @ 5:02 pm

  17. Thanks for the quick clarification. You guys rock!

    If I may ask, how frequently will these new stats be updated. If I’m not mistaken UZR is done weekly, but it looks like you already have some 2010 data for +/-.

    Comment by mickeyg13 — April 11, 2010 @ 5:33 pm

  18. Do park factors come into play at all with rARM? ( I guess you could look at home/road splits but I’m guessing the sample sizes are probably already pretty small?)

    For example LF in Boston seems like a huge advantage for preventing an extra base (either 2nd to home 1B to 3B or even 1B to home on on a double). And on the opposite end a large park may hurt (where it may be easier to score from 1st on a double)? Or say Coors where you may be playing deep and it might be easier to take an extra base on a single.

    Comment by Joe — April 11, 2010 @ 5:45 pm

  19. Any chance of adding a (IZ_plays + OOZ_plays)/(IZ_chances) metric? I think that’s the best use of the data.

    Comment by Sky Kalkman — April 11, 2010 @ 5:52 pm

  20. It’s unfortunate that it’s still like that, I had tried to contact him about it when I first got the book but to no avail. As you said, it’s something that likely could be normalized on Fangraphs end, and if you guys were to actually do that I think that would be amazing. While UZR is likely the superior stat, it’s still better to look at as much information as possible.

    Comment by Nny — April 11, 2010 @ 6:04 pm

  21. How long before +/- sample become reliable? 2 years or 3?

    Comment by SF 55 for life — April 11, 2010 @ 6:17 pm

  22. Are UZR and +/- on the same scale?

    Comment by Gina — April 11, 2010 @ 6:31 pm

  23. As with all the stats, there is not a single point where they switch from “unreliable” to “reliable.” Larger sample sizes mean they will be more reliable, and the larger the better.

    There is, however, a point where there is more signal than noise in the data. In other words, once we have n datapoints we think a player’s true value is closer to his observed value than to the mean value. Is that what you are after? If so I would guess it’s pretty much the same as UZR…

    Comment by mickeyg13 — April 11, 2010 @ 6:43 pm

  24. Yes; they’re both presented as runs above or below the average fielder at the position.

    Comment by J.T. — April 11, 2010 @ 6:49 pm

  25. 2-3 years of defensive data is roughly equal to one year of offensive data. It depends on the position, though.

    Comment by J.T. — April 11, 2010 @ 6:50 pm

  26. This is great.

    Comment by Evan Kirkwood — April 11, 2010 @ 7:11 pm

  27. All hail David Appelman.

    Comment by RKO36 — April 11, 2010 @ 7:12 pm

  28. Thinking about using this stuff for a THT article Tuesday…how is this data gathered? By BIS? And are certain ballparks not covered, or covered differently? Only 20 players have Scoops to their name from last season.

    Comment by Josh Fisher — April 11, 2010 @ 7:32 pm

  29. YAY!!!!!

    Comment by TJ — April 11, 2010 @ 7:41 pm

  30. Scoops should be first base only.

    Both DRS and UZR should be based off the exact same play-by-play and batted ball data, collected by BIS.

    Comment by Colin Wyers — April 11, 2010 @ 8:06 pm

  31. Switch from qualified only to all players, there are more than 20.

    Comment by Judy — April 11, 2010 @ 8:36 pm

  32. So just to be clear, UZR should be compared to DRS, not rPM, right?

    Comment by Jon — April 11, 2010 @ 8:43 pm

  33. This is great.

    So WAR is still based on UZR Defensive runs saved?. There is quite a difference between +/- DRS and UZR Defensive runs saved for some players (example +4 for JD in +/- and + 10.5 by UZR). Any chance of averaging the 2 to smooth things out and improving overall accuracy (in meterology several models are averaged to improve accuracy in forecasts).

    Comment by pft — April 11, 2010 @ 9:03 pm

  34. Essentially. The equivalent for rPM should be ErrR+RangeR.

    Comment by Colin Wyers — April 11, 2010 @ 9:04 pm

  35. For people with questions on the different components, here’s an explanation from the website:

    Defensive Runs Saved (Runs Saved, for short) is the innovative metric introduced by John Dewan in The Fielding Bible—Volume II. The Runs Saved value indicates how many runs a player saved or hurt his team in the field compared to the average player at his position. A player near zero Runs Saved is about average; a positive number of runs saved indicates above-average defense, below-average fielders post negative Runs Saved totals. There are eight components of Runs Saved:

    • Plus Minus Runs Saved evaluates the fielder’s range and ability to convert a batted ball to an out.
    • Earned Runs Saved measures a catcher’s influence on his pitching staff.
    • Stolen Base Runs Saved gives the catcher credit for throwing out runners and preventing them from attempting steals in the first place.
    • Stolen Base Runs Saved measures the pitcher’s contributions to controlling the running game. • Bunt Runs Saved evaluates a fielder’s handling of bunted balls in play.
    • Double Play Runs Saved credits infielders for turning double plays as opposed to getting one out on the play.
    • Outfield Arm Runs Saved evaluates an outfielder’s throwing arm based on how often runner advance on base hits and are thrown out trying to take extra bases.
    • Home Run Saving Catch Runs credits the outfielder 1.6 runs per robbed home run.

    Comment by philkid3 — April 11, 2010 @ 9:11 pm

  36. Well put, Colin!

    Comment by philkid3 — April 11, 2010 @ 9:48 pm

  37. Fantastic stuff. Keep up the great work!

    Comment by 3FingersBrown — April 11, 2010 @ 11:55 pm

  38. This is awesome and I’m one who actually likes the RZR/OOZ component that Hardball Times used to show. Plus/Minus and UZR may be the industry standard now, but I still like to see if RZR/OOZ provide support.

    Thanks for adding this at FanGraphs. This really is awesome.

    Comment by razor — April 12, 2010 @ 12:29 am

  39. Thank you! You take a great site and continually make it better, it is much appreciated – as you can see by the long line of comments, and I’m happy to add my kudos.

    Comment by obsessivegiantscompulsive — April 12, 2010 @ 4:51 am

  40. RZR has a defined, tangible output. it tells you the rate of plays made to plays in the zone. you combine that with plays made out of zone (OOZ), and fielding %, and you’ve got a hell of a lot more information that UZR could convey.

    Comment by steagles — April 12, 2010 @ 9:30 am

  41. Awesome! Thank you.

    Comment by nick — April 12, 2010 @ 9:39 am

  42. It does, and I actually misstated my question – it wasn’t the scaling I was having issues with, it was knowing the baseline used. Thanks.

    Comment by Kevin S. — April 12, 2010 @ 9:41 am

  43. I agree with the sentiment above…this is excellent. I also like to look at OOZ, because I think it tells you a different piece of information about players. I think Hardball Times originally separated RZR and OOZ because out of zone plays should have more value, and HT decided to let the user assign their own weighting to OOZ. I think it’s interesting to see the different profiles for players who have high RZR ranking but low OOZ, and vice versa, for example. I also like comparing Fielding Bible and UZR results to get a higher comfort level in the defensive ratings.

    I subscribe to Bill James on-line part because I like the defensive data; and I will continue subscribing in the future. I don’t want to punish that web site when they do a good thing like sharing defensive data with Fangraphs. (Besides, Bill James on-line is cheap.)

    Comment by CJ — April 12, 2010 @ 10:23 am

  44. I am new to FanGraphs. I have seen the Fielding Bible book in stores and perused it. I am trying to get into this but feel I need a beginner’s course in sabermetrics.

    Comment by Jim Don Seculius Dudek — April 12, 2010 @ 10:53 pm

  45. Glad to see the RZR numbers posted on fangraphs, Thank You!
    Question- Is the “runs saved” statistic an actual number, or a projection based on other stats?
    2. As a classic example, Placido Polanco led the majors in UZR among second basemen in 2009, yet he was way down the list even among AL second basemen in RZR. Which is a better measure of Polly’s fielding performance vs his peers?
    And here’s the modest position from whence I come: I like RZR. Number of balls hit into a player’s “zone” that he turns into outs. Okay, I can see that it has to be taken together with OOZ plays. But UZR confuses me. The explanation of those numbers make my head spin, and that’s extremely rare for me. Any clarification using PP as an example would be appreciated.

    Comment by Tigerdog — April 13, 2010 @ 12:21 am

  46. Right. RZR isn’t really a great measure of value, but it is more descriptive and intuitive than UZR. It’s a great gateway fielding stat.

    Comment by vivaelpujols — April 13, 2010 @ 1:56 am

  47. I feel stupid asking this….but is the “+/-” the same as DPS?

    Because if you compare DPS to +/- on THT website, they’re off. In 2008, Robinson Cano posted a DPS of -13 but a +/- of -16. Every player I look at is the same. Chase Utley has a DPS of 33, but a +/- of 47.

    What’s the cause of this difference?

    Comment by Rob — April 18, 2010 @ 4:10 am

  48. +/- is plays above/below average. DRS is in runs above average. Do you know where the +/- data is on THT?

    Comment by David Appelman — April 18, 2010 @ 11:33 am

  49. I do not. I haven’t figured out how to find stuff on THT. Maybe it’s just me.

    I can find the +/- data for the Fielding Bible Awards…but that’s only a limited amount of information.

    Comment by Rob — April 19, 2010 @ 9:31 am

  50. Huh? You just said that you compared DRS to +/- on the THT website?

    Comment by David Appelman — April 19, 2010 @ 11:03 am

  51. Oops. That was my mistake. I meant the Fielding Bible site.

    Since the +/- is different from DPS, is there comprehensive +/- data available other than the Fielding Bible site? I see a lot of new fielding info here on Fangraphs, but I’m not seeing the +/-.

    I mean, is this something really simple that I’m overlooking, or am I just really confused (or possibly both)?

    Comment by Rob — April 19, 2010 @ 11:36 am

  52. Rob,

    Plus/Minus is the same system as DRS. The difference between the two is that Plus/Minus is presented in plays above or below average; DRS is presented as *runs* above or below average.

    Comment by J.T. — April 23, 2010 @ 12:25 pm

  53. Looking at DRS at the team level just now, I notice that teams collectively are + 232 runs above average. How is this possible, should it not come out to 0?. If it should then this inflation comes out to about 7 runs per team and must also be inflated at the player level.

    Comment by pft — May 15, 2010 @ 2:18 am

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.

0.206 Powered by WordPress