FanGraphs Baseball


RSS feed for comments on this post.

  1. Nice piece. He has been pretty special this year, for sure.

    Considering his peculiar track record though, who know’s what the Bosox will get next year.

    Comment by The Duder — September 22, 2010 @ 5:31 pm

  2. Semi-serious question: Should we consider the fact that Beltre sent two left fielders to the DL? If it had just happened once, I’d call it a fluke, but the exact same incident twice? Maybe we should be subtracting Ellsbury’s projected contribution from Beltre’s…

    Comment by JD — September 22, 2010 @ 5:31 pm

  3. Once again his numbers rely heavily on an inflated BA. Yes, he’s had a great year. And yes, he will not be as good next year. He’s a career .275 hitter hitting .325. When the BA goes, so does the OBP, his walk rate is not great.

    Comment by John — September 22, 2010 @ 6:13 pm

  4. I’ve always loved Adrian Beltre and was ecstatic when he signed with the Sox this past offseason. Truly one of the most fun players to watch in the league. Part of me wants him to resign just for what he’s done and how much fun he is to watch, but I know it’d be an overpay in years and money. Still, he’s a special player and deserves any credit he’s given. He’s an easy 2nd pick for MVP.

    Comment by Evan Kirkwood — September 22, 2010 @ 6:17 pm

  5. RC doesn’t factor into the MVP race? Really? Despite what fWAR and bWAR and their average (as Tango suggests) says?

    Not that WAR is end all be all, but it’s a good start.

    Comment by Disco — September 22, 2010 @ 6:31 pm

  6. Keep him in Fenway! Get it done, Theo.

    Comment by Obligatory Sox Fan — September 22, 2010 @ 6:35 pm

  7. robinson cano plays elite defense (albeit at a less important defensive position) and he is hitting very well for the second year in a row while posting career highs in homers and rbis. I think he should be in the discussion if beltre is.

    Comment by phoenix — September 22, 2010 @ 7:16 pm

  8. Can we stop with the Rangers are just resting Hamilton garbage? The guy is injured, has 2 cracked ribs, has had a couple of shots and is not simply resting because the Rangers have a big lead.

    He probably should win the MVP but this attempt by a few Fangraph writers to waive away the loss of playing time as some sort of choice is getting ridiculous and undercuts the other (and more objective and analytical) arguments.

    Comment by tom — September 22, 2010 @ 7:19 pm

  9. Elite defense?

    Comment by Not David — September 22, 2010 @ 7:19 pm

  10. I honestly don’t see #1. How Robinson cano is mentioned as comparable to shin-soo choo and Luke Scott when neither have anywhere near the batting average, rbis, runs, or homeruns (i.e. Every major category) and #2 how konerko is listed as a more elite hitter than cano. Like I get that he has a lot of hrs and RBI, but I kinda don’t understand how cano is not at least listed as as good of an overall hitter as konerko. Plus konerko has hardly been an MVP, considering he’s not on a contender. The difference between canos chance for MVP vs. Choo and Scott are the fact that cano is by far a better hitter, plays better defense, and legit has been the most valuable player on a world series contending team.

    Comment by Louis — September 22, 2010 @ 7:38 pm

  11. Sorry, but I do not think Luke Scott is an excellent hitter.

    Comment by Jon — September 22, 2010 @ 7:38 pm

  12. Why do you think 2nd base is less important than 3rd. Isn’t the positional adjustment the same? Is there some other reason 3rd is more important? more chances?

    Comment by brendan — September 22, 2010 @ 7:52 pm

  13. Beltre’s great offensive season and elite defense are key but you failed to mention my favorite story of the season. The sportscenter footage of Beltre throwing wild punches in the dugout everytime V-Mart touches his head is priceless.

    Comment by Paul — September 22, 2010 @ 7:59 pm

  14. wOBA
    Konerko – .418
    Cano – .394
    Choo – .382
    Scott – .385

    Comment by Not David — September 22, 2010 @ 8:01 pm

  15. I’d be more than willing for the Tigers to overpay him in money and years to come to Detroit in 2011. As long as “overpay in years” isn’t defined as a number greater than four.

    Comment by Larry Smith Jr. — September 22, 2010 @ 8:23 pm

  16. Paul Konerko WAR: 4.5
    Brett Gardner WAR: 5.0

    Isn’t WAR fun?

    Comment by tom — September 22, 2010 @ 8:36 pm

  17. I’m not disputing Beltre’s two-sided dominance this season. I just think that this number is extremely peculiar.

    Beltre’s WPA is -0.38. A guy with a .396 wOBA. That’s a head-scratcher. His clutch factor is -1.80, although I’d be lying if I said I knew how exactly to calculate that.

    I’m not implying that his situational hitting has made his excellence meaningless. I’m just pointing out something odd, that’s all.

    And unfortunately for Beltre, voters tend to shy from players on teams who won’t make the playoffs.

    Comment by TribeFanV — September 22, 2010 @ 10:21 pm

  18. Cano’s UZR is 1.5. Above average at a prime position? Of course. But Beltre sits at 11.9.

    Comment by TribeFanV — September 22, 2010 @ 10:24 pm

  19. Cano is a better hitter and a more valuable player than Choo and Scott this year. But I feel there are misconceptions afoot.

    Choo’s UZR is 7.7 compared to Cano’s 1.5. Choo’s a surprisingly good outfielder, particularly because of his strong throwing arm. I’ll grant you that 2B is the more demanding position and merits more positional weight both offensively and defensively, but Choo is farther above the league-average RF than Cano is above the league-average 2B

    Luke Scott, however, sits at -2.3 when it comes to UZR.

    Comment by TribeFanV — September 22, 2010 @ 10:30 pm

  20. Umm… how is Cano “not really” in the discussion for MVP? He has the third highest fWAR (higher than all the players mentioned except Hamilton and Beltre) and plays a premium defensive position well. Then lumping Cano (6.6 WAR) with Choo (5.4 WAR) and Luke Scott (2.8 WAR)?

    Comment by Francisco — September 23, 2010 @ 12:53 am

  21. Of course Choo was -2.7 in RF last year (similar # of chances)… beware the ultrareliance on 1 year UZR estimates.

    The 2 years before that in RF (both limited playing time) were -3.1 and 4.9. So coming into this year his rightfield cumulative UZR was -1.

    And this is why WAR’s based on 1 year sample sizes are iffy… I can’t believe Fangraphs does not do a regression on this component in WAR.

    Comment by tom — September 23, 2010 @ 1:30 am

  22. @Tom – not really, it’s actually pretty stupid when you consider the holiness Dave Cameron and his disciples attach to it

    Comment by anonymous — September 23, 2010 @ 1:41 am

  23. Cano has been easily the best player on the Yankees during a season in which the old stalwarts of production have fallen. If people told you that Jeter, Tex and A-rod would contribute what they have so far, tell me, would you have said the team would be in first place right now? Not likely, now Gardener did need to have a monster season, but Cano is the number one reason why the Yankee team is in first.

    Comment by Cliff Lee's Changeup — September 23, 2010 @ 1:57 am

  24. In fairness he was lumping Cano in with Scott from just a hitting perspective.

    Of course even that is absurd as Scott is about where Billy Butler is on the hitting component to WAR and is substantially behind the likes of Nick Swisher (who is having a fine year). Also the guy really seems to like the home cooking

    Home: 1.096 OPS, .459 wOBA
    Away: .691 OPS, .301wOBA

    Are we really putting Luke Scott in with these other hitters? He’s had a good year, but are does he belong in the 2nd tier below the Hamilton/Cabrera tier?

    Comment by tom — September 23, 2010 @ 2:30 am

  25. Once again, he’s played most of his career in two parks that strongly suppress BA. His true talent is >.275. He squares up everything with great plate coverage.

    Comment by alskor — September 23, 2010 @ 2:48 am

  26. tom – weird… its almost like WAR is accounting for the fact you would expect a bad glove 1B to hit much more than a plus defensive OFer.

    WAR – fun AND instructive.

    Comment by alskor — September 23, 2010 @ 2:51 am

  27. Well, if we’re looking at more than 1 year with UZR (as we should) then Cano is the one who is hurt the most.

    Comment by alskor — September 23, 2010 @ 2:52 am

  28. You’re right… Cano *should* be the Yankees’ TEAM MVP.

    Comment by alskor — September 23, 2010 @ 2:55 am

  29. I’m sorry you feel that way, but the facts disagree with you.

    Comment by Aaron/YYZ — September 23, 2010 @ 9:04 am

  30. Ooooh. RBIs and runs. There’s a case that’ll go a long way on Fangraphs.

    Comment by NBarnes — September 23, 2010 @ 9:23 am

  31. Expanding on your “squares up everything” point: It’s really hard to quantify, but he also smashes pretty much everything he hits. I guess it might show up a little bit in his LD%, but that doesn’t really cover it; he has a lot of balls that hit the green monster that are only singles because he hit them so hard that they get out there too quickly, if that makes sense.

    Comment by Steve B — September 23, 2010 @ 9:29 am

  32. I don’t think argument here is that they’re choosing to rest him, but that his injury came at a time when it had little ability to affect the rangers’ season. You could even say this makes his WAR more impressive since all of it came at important times, and he wasn’t simply “compiling stats” in junk time. Games have a leverage index too, and the games Hamilton has missed have not had a particularly high one.

    Comment by Doug — September 23, 2010 @ 9:52 am

  33. Yeah, dude could probably hit home runs from his knees.

    Comment by Eric — September 23, 2010 @ 10:15 am

  34. I think it’s probably closer between Beltre and Hamilton than a lot of people think but I’m not so sure Canu is in the discussion for more than 3rd, 4th, or 5th. What I’m surprised about is that neither Crawford or Longoria have gotten much attention, Rays have been great this year and I think both deserve some attention. I think if Beltre, Crawford and Longoria explode in the last few weeks we can give it to one of them otherwise what Hamilton did deserves recognition.

    Comment by Eric — September 23, 2010 @ 10:21 am

  35. “And this is why WAR’s based on 1 year sample sizes are iffy… I can’t believe Fangraphs does not do a regression on this component in WAR.”

    Tom is promoting a mistaken belief that I see being made on these boards quite often. The sample-size issue is in regards to assessing true talent and not what actually happened. The argument above is the exact same as saying, “well, he’s had 130 hits in 500 at-bats this season, but he was a .300 hitter the last two years, so he should really be given credit for something like .280 in this season.” This is incorrect logic. UZR in WAR describes what has happened. We should expect UZRs to fluctuate yearly, much the same way that batting average fluctuates yearly, so the reason that we regress UZR for projections is because it’s hard to figure out true talent in a small sample size. It’s wrong to regress UZR’s runs saved in WAR, which is a backward looking statistic. The fielder’s play really did save runs on a specific catch, even if he only gets to that ball one time in ten.

    Comment by JFC — September 23, 2010 @ 10:23 am

  36. Excellent point about Choo. This is his first year as a good defensive outfielder. I do think he’s made changes that should stick. For example, he’s always had a strong arm, but in the past he’d continually overthrow the cutoff when he had no chance to throw the lead runner out, allowing multiple runners to advance. He’s not doing that as much now, and his rARM is starting to look more like one of a guy who consistently hits home on a line.

    However, he may be getting an inordinate amount of easy chances this year. Any first good year has to be taken with a grain of salt…time will tell. I just have to be somewhat optimistic about my favorite team, or there’ll be a man on a megaphone telling me not to jump.

    Comment by TribeFanV — September 23, 2010 @ 10:26 am

  37. Until postseason, at least… I hear tell those games are somewhat important too.

    Comment by B N — September 23, 2010 @ 10:54 am

  38. Except we know that a hit is a hit, because the ball landed on grass and the guy is on 1st base. If UZR decides that a certain play was +0.15 because 15% of shortstops wouldn’t have gotten to that ball, there’s more estimation involved. Maybe he spends a lot of time at double play depth because his pitchers are bad and walk everybody.

    Comment by don — September 23, 2010 @ 11:29 am

  39. Seconded. Elite is a big stretch. Inconsistent is the only word I’d use to describe Cano’s defense. On his career, he has a negative UZR/150. This is only the second time he’s posted a positive UZR and he’s still only really qualifying as average on the season.

    Elite defender and Robinson Cano are two words that should only be used in conjunction when saying “and Cano is robbed of a hit by elite defender Adrian Beltre” or other such statements of the sort.

    Comment by Jonathan — September 23, 2010 @ 11:44 am

  40. Hamilton is the MVP of course, but Beltre is a great player when he puts up offensive numbers like this year because of his defense. I still would not sign him if I was Boston, too many down years when there is no contract on the line for my tastes.

    Comment by Phantom Stranger — September 23, 2010 @ 12:12 pm

  41. Exactly!

    UZR is acurately describing what happened each season, but because of the many random things that affect a one year sample (ie. ball distribution) its hard to tell what the player’s actual fielding ability is with real accuracy. This is fundamentally not that different from a player hitting .300 in his first season as a pro and then after three years we find out he’s more of a .280 true talent hitter. Fans don’t have a philosophical issue accepting that like they do UZR… & defensive slumps/hot streaks. Those are all taken immediately as evidence something is wrong with defensive metrics. The guy in this example still DID hit .300 his first season. That actually happened. We just can’t tell how much was luck, random, etc…

    Comment by alskor — September 23, 2010 @ 12:32 pm

  42. Could? Already Did:

    Comment by Steve B — September 23, 2010 @ 12:48 pm

  43. Hamilton, yes. But for possible runners-up, Bautista is never really in the conversation anywhere is he. Maybe no-one can believe it. But he’s not just leading the league in HR, he has hit 25% more dingers than his nearest competitor. That’s pretty good. He is second in the AL in walks. He’s (tied) second in wOBA. He’s second in OF assists despite splitting his time at third. WAR hurts him I suppose and this is a WAR site. But still.

    Comment by Jose's Mom — September 23, 2010 @ 1:40 pm

  44. wow–beltre’s career WAR is over 50. does he have HOF potential? he’ll probably be underappreciated because so much of his value was tied to his health and defense at a position (unlike SS or C), where the cognescentii tend to underrate the value of excellent D.

    Comment by Dudley — September 23, 2010 @ 1:48 pm

  45. Using the xBABIP calculator, a player hitting in Fenway with Adrian Beltre’s AB’s, HR’s, K’s, LD%, popups, FB’s, GB’s and SB’s is expected to post a .313 BABIP. So, Beltre has been a tad lucky in the BABIP department this season, although regressing his current .335 BABIP to .313 yields a .303/.361/.531 batting line (does anyone know of a quick and dirty method to convert a player’s batting line to wOBA?). This removes 8 singles, three doubles and a HR (though 1.84 might be a better number: (HR/H)*12). Regardless, Beltre’s a .900 OPS third baseman playing elite defense. That recipe will garner MVP votes annually, and deservedly so.

    Comment by John — September 23, 2010 @ 2:10 pm

  46. Wrong! career .270 avg, 20-25 hr’s, 65-75 rbi. No… Just No.

    Comment by Jon — September 23, 2010 @ 2:27 pm

  47. Batting average, home runs, RBI….the trifecta.

    You took a wrong turn on the way to ESPN.

    Comment by Not David — September 23, 2010 @ 2:53 pm

  48. Luke Scott is not an excellent hitter. He is the best hitter on one of the worst teams in baseball. I can tell Scott is not an excellent hitter by watching him play. He has a career 11 WAR in over 650 games… Mind-blowing. He’s not even the best outfielder on his team, which is one of the worst teams in the Majors. You take Luke Scott. I’ll take Nick Markakis.

    Comment by Jon — September 23, 2010 @ 3:00 pm

  49. He blows… I can’t stress this enough.

    Comment by Jon — September 23, 2010 @ 3:12 pm

  50. UZR might be describing what happened each season.

    Then again, there’s a good chance it isn’t.

    Because defensive statistics are so subjective, a more reasonable solution is to estimate a player’s current talent using regression, instead of trying to describe exactly what they did on defense in a given season (you can’t, not with the data we have now).

    Also, FanGraphs should not just be using UZR. It has the DRS data, so why not use that as well?

    Because UZR (and everything else, really) is a statistic that relies on subjectivity, defensive statistics are more useful if you look at as much data as possible. Just looking at UZR is like just using one website for pre-season projections.

    Comment by AndrewYF — September 23, 2010 @ 3:28 pm

  51. If you knock down any more strawmen you’ll get arrested for assault.

    WAR = offense + defense

    Only Scott’s offensive production for this season was referenced in the article, a season in which he’s producing an “excellent” wOBA of .385.

    Comment by Not David — September 23, 2010 @ 3:28 pm

  52. I think we have different definitions of excellent. I would call him pretty good but, I wouldn’t call him excellent. Excellent, I define as the best hitters in the game. Scott is not one of the best hitters in the game to me. You may disagree.

    Comment by Jon — September 23, 2010 @ 3:41 pm

  53. Don, I think this is a very smart argument.

    The issue, though, is that it’s basically saying UZR doesn’t report what it purports to describe because it is not accounting for context. This is certainly more true of UZR than other stats, but if philosophically we’re going to choose to use UZR as the fielding stat, I think it would be silly to try to say why a player performed worse or better than he should have on a specific batted ball. To turn your example into uncertainty regarding hitting stats, it’s entirely possible that a guys is hitting more GB singles because the hitters in front of him are getting on base and forcing the fielders to play at double play depth. There is bias in those hits as well, but wOBA still gives him credit for the hits and it shows up in his WAR.

    So WAR isn’t xBABIP corrected, and we use FIP, not xFIP, for pitchers. There’s a reason. Clearly, fWAR is more context neutral than rWAR, but I think if you took it further and regressed its components, it would cease to be a description of what has occurred. This is what projections are for.

    There is a good deal of noise and luck in all baseball statistics–UZR more so than others. But I think that this is the best description we have of what sort of defensive value a player provided to his team in a season.

    Comment by JFC — September 23, 2010 @ 4:01 pm

  54. Alksor – So Gardner should finish higher in the MVP voting than Konerko?

    WAR is instructive when you know the limitations… When you are using a high variation defensive model and that impacts WAR significantly, you need to be careful about this blind devotion to all things WAR as a measure of past performance. Not saying Hamilton isn’t the MVP, just that it’s not the runaway everyone says it is unless you choose to place all your faith in 1 year UZR #’s.

    Has Teixera really been worse defensively than Cabrera this year at 1st base? 1st base UZR’s are a bit sketchy, but they say that is the case (so it must be true?). I’m not saying Cabrera is even average, just that those #’s need to be carefully used and regressed and not blindly stuck into WAR. Cabrera’s career UZR over the previous 2 years at 1st base was -0.5.

    Hamilton’s career UZR in the OF? -0.3 (cumulative)
    LF: +5.5 (almost all ARMR)
    CF: -9.6
    RF: +3.1

    Plus defender? Maybe. Plus arm? Certainly. A +6.0 UZR for a guy who had a cumulative ~-7.0UZR coming into the year? Welcome to the fun world of 1 year UZR sample sizes!

    Comment by tom — September 23, 2010 @ 6:45 pm

  55. Oops Hamilton’s career UZR in the OF is -1.0, I was looking at UZR/150 (which is -0.3)

    Comment by tom — September 23, 2010 @ 6:47 pm

  56. UZR is not describing what happened, it is modeling it (albeit a very good model).

    As soon as you start breaking the field up into zones and assign an average probability across the zone (when obviously not every position in a specific zone is equally difficult), you are estimating defense, not measuring it.

    With a small sample size you are not merely dealing with an issue of measuring what happened vs true talent… you are also assuming the distribution of balls within given zone is representative of the overall distribution the zone is based on. In other words the probability of Hamilton getting to a ball in a specific zone in a given year may not be directly comparable to the average of all balls in that zone as the distribution of balls in that zone may not be the same.

    The zones are based on all fielders, 6 years… obviosuly a pretty decent sample size. With many fielders you are talking 200 or even fewer plays, and once you start breaking those plays into zones, accounting for speed of runner, speed of batted ball, handedness you are now dealing with small buckets that may not reflect the average distribution. Obviously as the zone becomes smaller (or infinitely to a specific vector… like what field F/x is trying to do I think) or the sample size gets larger the model becomes more reflective of what has happened. When you are dealing with 1 year UZR sample sizes, you can’t make this claim.

    Comment by tom — September 23, 2010 @ 9:16 pm

  57. I think UZR can be looked at in a similar way to BABIP. Over the course of a game, week, or even season, balls can sort of find gloves just like they can find holes. In one season, a lot of balls might be hit just inside Teixeira’s range, allowing him to gobble up UZR points in that zone. The next season, a randomly disproportional number of balls are hit just outside his range, but in the same zone. His UZR looks worse. The luck that plays a role in babip, I believe, can be likened to the luck that plays a role in UZR from season to season (which is why multi-season UZR numbers are more valuable).

    Just a thought.

    Comment by Matthias — September 24, 2010 @ 12:56 am

  58. Folks should re-read the UZR primer. An excerpt from the conclusion:

    “Three, a player’s UZR, be it one year, one month or 5 years, ****is not necessarily what happened on the field*** and is not necessarily that player’s true talent level over that period of time either. That is why we regress, regress, and regress! A player can have a plus UZR and have played terrible defense, because the data we are using is far from perfect.”

    Note the part about it being not necessarily being what happened on the field. Variation in the stats is not merely luck or the difference between what happened and true talent. This in my view is the important distinction between looking at variation in this stat and say BABIP, which is something fairly easily measurable and binary on the field (save for things like the subjectiveness of an official scorer on a close play or a blown call).

    The issue is the limitations on the inputs to the UZR model/stat. It doesn’t make UZR a bad statistic, it just means folks need to be more aware of the potential error bar when throwing out WAR #’s (especially when they are used to represent past performance). The variance may not simply be luck, it may be the model (or more precisely the inputs to the model) and thus not simply a difference between what happened vs true talent.

    Mattias – I agree that there is a BABIP like component to the variation (balls finding a specific spot in a zone, etc), but the variation in UZR goes beyond that based on reasons I mention above. The variation is also on an entirely different level of magnitude, and as such can have far more impact on something like WAR then BABIP variation.

    Comment by tom — September 24, 2010 @ 4:02 am

  59. Tom,

    You’re absolutely right that the fact that the model is in doubt is a key difference between this stat and other types of uncertainty. But this, to me, is still not an argument for regressing UZR for purposes of WAR but a good one for combining it with DRS, for instance. This would give us potentially more accuracy if we do not believe that one of the models in fundamentally flawed. The issue with regression to past performance is that in that case you would be giving credit or docking a player for how he’s played outside the given sample. This is outside the purview of the WAR statistic. Also, it mitigates a player’s ability to show improvement or regression.

    Comment by JFC — September 24, 2010 @ 1:18 pm

  60. Well, playing time sure played a factor in the 2008 AL MVP race. Why shouldn’t it this year?

    Comment by 2nd Half Adjustments — September 24, 2010 @ 2:01 pm

  61. Thanks, Tom. In terms of ways to improve UZR and limit its variance, what kinds of things can be done? I know I’ve heard of taking the velocity of the ball off the bat into consideration. Perhaps partitioning the field into more zones?

    Comment by Matthias — October 9, 2010 @ 3:17 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.

0.282 Powered by WordPress