Skip to content

A critique of WAR

Sep 6, 2011, 12:30 PM EDT

savingsaccountcalculator

No, not war. WAR. As in, Wins Above Replacement level.  Hippeaux at IIATMS has a lengthy column up today looking at the stat (it’s actually multiple stats complied via different methodologies by different folks) and pointing out a potential flaw. Specifically, that one of its components — the defensive metric, Ultimate Zone Rating or UZR — that Hippeaux argues distorts WAR.

I am quite obviously not a stats guy, so I can’t say that I have any real insight here. I’m mostly waiting to watch the arguments and counterarguments to the piece, which is how most folks who don’t have their own aptitude with this stuff should probably proceed. Over time you get some clarity as to whether the critique makes sense, if the counterarguments make sense, etc., and then eventually you get something approaching an advancement in real knowledge.  Kind of like science, you know.

But this article is getting a lot of play this morning, pro and con, so it’s worth throwing out there.

One warning: I’ve already seen a couple of decidedly non-statty writers link to this thing and say something to the effect of “see, those holier-than-thou stats types don’t know WHAT they’re talking about.”  To those people I offer a hearty “shove it.”  A critique of a component of an analytical tool does not constitute a repudiation of the analytical tool any more than a faulty component in your engine constitutes a totaled car.  If it makes you feel better to say “see, those statheads aren’t God,” so be it. But do know that the statheads have never been interested in being God. They just want to understand stuff, and this sort of intellectual give and take is how that occurs.

141 Comments (Feed for Comments)
  1. Jason @ IIATMS - Sep 6, 2011 at 12:38 PM

    Thanks, Craig. I dig your explanation.

    Not an invalidation of the outputs involved, just a questioning of the inputs and their inherent vagaries.

  2. acheron2112 - Sep 6, 2011 at 12:39 PM

    On one hand, I was disappointed that this post was not titled “WAR: What is it good for?”. But I guess that’s too obvious a joke, though that’s never stopped Craig before.

    • klbader - Sep 6, 2011 at 12:44 PM

      I was thinking the same thing.

      • bigharold - Sep 6, 2011 at 1:08 PM

        I was thinking more along the lines of WAR is hell, .. but I’m rather dubious about WAR to begin with.

      • nolanwiffle - Sep 6, 2011 at 1:16 PM

        It is well that war is so horrible, lest we grow too fond of it….

  3. zakharovsa - Sep 6, 2011 at 12:41 PM

    Could this be the article to displace the latest Phillies-related post from the “Most Commented” list? It depends if halladaysbiceps shows up or not.

    • heyblueyoustink - Sep 6, 2011 at 1:32 PM

      Nah, it’s a masked attempt by Craig to discredit Steve Carlton’s 1972 season as lucky, being I believe Carlton has the highest WAR of all time in that season, if i’m not mistaken.

      ( Paranoia, cha cha cha )

      • thefalcon123 - Sep 6, 2011 at 4:10 PM

        It’s the highest of the live ball era and the 2nd highest since 1900, but it only ranks 28th all time. This is becase pitchers became huge pussies and stopped throwing 650 innings a season. If Old Hoss Radbourn could do it, why couldn’t you Mr. Carlton!

  4. alfreddigs - Sep 6, 2011 at 12:47 PM

    I frequently check the WAR totals on player pages at Baseball Reference, and it’s always struck me as odd that, while oWAR is fairly consistant (ie. a good offensive player will consistently put up positive oWAR totals), dWAR fluctuates like crazy. A player can have positive 0.5 one year and -0.5 the next and a positive total again the year after that. I’m not a stats guy either, but I imagine these dWAR totals, which are often fairly close to 0, are wavering within (or close to) the margin of error. This really has some effect when a guy with a long career puts up multiple -0.5 dWAR type seasons, and can end up ultimately losing quite a few overall WAR points.

    • drmonkeyarmy - Sep 6, 2011 at 1:21 PM

      Single season defensive metrics are ridiculously filled with error and deviation. Anybody who treats them as gospel is mistaken in my opinion.

      • Alex K - Sep 6, 2011 at 1:30 PM

        Anyone who treats single season defensive metrics as gospel is also mistaken in the eyes of the people who created the statistics.

      • drmonkeyarmy - Sep 6, 2011 at 1:38 PM

        I agree. The point is then why is it given so much weight in the single season calculations of WAR?

      • Ari Collins - Sep 6, 2011 at 1:42 PM

        It isn’t. Even by FG.

      • Alex K - Sep 6, 2011 at 1:45 PM

        I guess it’s because it is the best measurements available and defense has to be accounted for when determining a players value for the year.

        WAR is not the be all end all. Anyone who says it is is mistaken.

      • drmonkeyarmy - Sep 6, 2011 at 1:47 PM

        Well, then I think you and I are in agreement Alex.

      • normb11 - Sep 6, 2011 at 1:50 PM

        Isn’t a multi year UZR sample just to try and determine the true talent of the player?
        With WAR, they’re just trying to describe what happened, so the true talent isn’t all that important.
        A hitter might get lucky and bat .340, but we know his true talent is much less than that. But, the fact remains, he did hit .340 that year.
        Isn’t that why UZR is included in WAR? It might not be true talent, but it did happen?
        Or am I wrong?

      • Alex K - Sep 6, 2011 at 1:50 PM

        I don’t think we are in total agreement, Doc. I still think WAR is a really good tool. I don’t consider it worthless or anywhere close to worthless. I do, however, acknowledge that is is not without it’s flaws.

    • normb11 - Sep 6, 2011 at 1:31 PM

      Prince Fielder can go from 50 homers to 34 homers, back up to 46 homers, down to 32 homers….
      His batting average from 276 to 299 to 261 to 293…
      -
      why is it so hard to believe the same can happen for defense?

      • drmonkeyarmy - Sep 6, 2011 at 1:39 PM

        It isn’t….but those stats are concrete data. The defensive calculations are too subjective for my taste.

      • Alex K - Sep 6, 2011 at 1:41 PM

        Your point is valid – there are swings in performance year to year. But the numbers you gave are context free and incomplete.

        Also, the swings in defensive outputs are much wider.

      • normb11 - Sep 6, 2011 at 1:47 PM

        I’m as skeptical of defensive metrics as the next guy, but it’s not because of swings in data….swings happen with a lot of statistics.

      • ftbramwell - Sep 6, 2011 at 5:04 PM

        Because offensive statistic are a function of how pitchers and defenses attack a batter and how well the batter adjusts to those approaches. But do you really think that anyone not named Chuck Knoblauch forgets how to field his position?

    • paperlions - Sep 6, 2011 at 3:04 PM

      It is worth noting that dWAR doesn’t fluctuate like crazy, and it doesn’t fluctuate nearly as much as oWAR even for individual players. The example above is a difference of 1 dWAR, which is not much at all, actually. The best oWAR season ever is 13.7; the best dWAR season ever is 4.0. The WAR component represented by defense (at least for BR WAR) is relatively small compared to the offensive component, indicating that 1) those calculating BR WAR think offense is much more important than defense, or 2) that the current version of dWAR under-estimates the importance of defense.

  5. Chris Fiorentino - Sep 6, 2011 at 12:50 PM

    “While I admit the difficulty of building a model that accounts for the effect a pairing like Braun/Fielder or Pujols/Holliday has on the rest of the lineup, this is one area in which I find the conventional wisdom to be irrefutable. While I applaud WAR (and other metrics) for aiding in our appreciation of defense and baserunning, it’s beyond asinine to conclude that Ellsbury is twice as valuable as Fielder. Too often WAR is used as a means of comparing oranges to apples. One of the things that makes baseball great is the diversity of the fruit basket. WAR give incredible weight to scarcity of shortstops, but no weight to the scarcity of pitcher-intimidating, strategy-altering cleanup hitters, which I see as a form of reverse discrimination.”

    I just have to say that as I read this paragraph coming from a sabremetric writer, I jumped up from my desk and yelled “AMEN BROTHER!!! HALLELUJAH!!! PREACH IT!!! PREACH IT!!!” I’m not going to say that this repudiates all stats…but this single paragraph does in fact repudiate the WAR stat and I want to thank the writer from the bottom of my well-beaten anti-WAR heart.

    • manifunk - Sep 6, 2011 at 12:58 PM

      Way to confuse “hey this stat could use some work and has some flaws” with “totally repudiated”.

      • Chris Fiorentino - Sep 6, 2011 at 1:06 PM

        Sorry to burst your bubble, mani, but when a stat guy writes that “It’s beyond asinine to conclude that Ellsbury is twice as valuable as Fielder” it is saying exactly what I have said everytime somebody comes on here and posts WAR to compare guys…which Matthew does almost everytime he wants to compare players.

        This article is a total repudiation of the WAR stat as it currently exists.

      • Ari Collins - Sep 6, 2011 at 1:28 PM

        Yes, because one scientist disagreeing with a theory means that that theory has been repudiated and is worthless. It’s in no way a continuation of the scientific method.

      • thefalcon123 - Sep 6, 2011 at 4:13 PM

        Yes Ari, that is exactly how it works. You know, how global warming is fake because 1% of scientists say so and how 9/11 was an inside job because 1 out of every 3,000 structural engineers say so. That’s just how science works!

    • drmonkeyarmy - Sep 6, 2011 at 12:59 PM

      That might be the single greatest paragraph I have read in quite some time. I completely agree and pretty much summarizes my feelings about WAR. It should be noted that the author is way more eloquent in his delivery than me…but it is what I’ve been trying to say for months. Thank you author.

    • evanhartford - Sep 6, 2011 at 1:02 PM

      Yeah, I wouldn’t call it complete repudiation. Its just been taken off its pedestal so to speak.

      • Chris Fiorentino - Sep 6, 2011 at 1:07 PM

        It is a repudiation of WAR as it exists today…and that’s good enough for me.

      • manifunk - Sep 6, 2011 at 1:15 PM

        The thing to realize is that, even though WAR has some significant flaws, it’s still lightyears ahead of other stats like batting average, RBIs, pitcher wins and saves. Even with these flaws, WAR is the best measure available.

      • Chris Fiorentino - Sep 6, 2011 at 1:24 PM

        While I will disagree with you on the value of WAR, mani, there are a lot of stat heads who will say that WAR is it. They will put an article together where they ask who is the MVP…or who is the Rookie of the year…and the major stat they will use to sort guys is WAR. Why? Why should a guy who is a shortstop be given such a large advantage over a first baseman because of the position he plays? Isn’t assigning value positionally just as stupid as assigning value based on playoff position? It isn’t Bautista’s fault he is on a losing team anymore than it is Fielder’s fault he is a first baseman and there are a lot of first basemen.

      • The Dangerous Mabry - Sep 6, 2011 at 1:56 PM

        I’d have to argue that it IS Fielder’s fault that he’s a first baseman. If he could play above-average shortstop, you can bet your ass that’s where he’d be. You put a guy at first base because he can’t get the job done elsewhere. Very few first basemen play the position due to their defense, and for those that do, their team is generally giving something up by playing a defensive specialist at first base, rather than a big bat who can get the job done.

      • Chris Fiorentino - Sep 6, 2011 at 2:30 PM

        That’s silly Mabry. I’ll ask you what I asked below…what game are we playing…I’m a fan or I’m a GM. Because they are two totally different games. GM’s worry about building a roster. Fans worry about whether one player is BETTER than another…not his “replacement value”.

      • The Dangerous Mabry - Sep 6, 2011 at 2:53 PM

        I’m simply arguing with your statement that it’s not Prince Fielder’s fault that he’s a first baseman. It’s entirely his fault. Teams have always known that they needed to get more offense at the easier defensive positions, thus the reason that first (and to a lesser degree third) basemen and corner outfielders have typically been the premier hitters in baseball history. And fans have always known to value “up-the-middle” defenders differently when it comes to their offense, which is why you’ll see players like Ryne Sandberg considered hall of famers in the same discussion as players like Mike Schmidt, who was a substantially more productive hitter. Sandberg was a second baseman. Ozzie Smith was a shortstop. You don’t require the same kind of offensive production out of those positions, simply because it’s not likely that a player can give you that kind of offense and also be gifted enough to play those positions in the field. It’s the same reason that so many call Willie Mays the Greatest Ballplayer of All Time instead of someone like Ted Williams, even though Williams was quite simply a better hitter. Mays was a centerfielder, and that’s worth a whole lot.

        So when you happen to see a player that can give you substantial offensive output in a premium defensive position, you’ve got to grab him, and realize that you’ve got something special.

        If fans are worrying about who is “better”, then you can have goofy exercises like “who’d be a better team, 7 Jacoby Ellsburys or 7 Prince Fielders? You might score a few more runs with the all-fielder team, but your defense would be so atrocious that you’d be in a world of trouble. Of course that kind of discussion is silly, but if that’s where you want to go, that’s the question I’d ask.

      • Ari Collins - Sep 6, 2011 at 3:03 PM

        Chris… the shortstop IS better, whether you’re a fan or a GM. Defense is part of the game too, and the shortstop is way way way better at it than the first baseman.

      • Chris Fiorentino - Sep 6, 2011 at 3:08 PM

        Nah, I wouldn’t play those stupid games. I’ll just say that Fielder is a better player than Ellsbury and leave it at that. Even if WAR tells me a totally different story…and by a mile.

      • skipperxc - Sep 6, 2011 at 3:38 PM

        Care to give us a reason? Or is it really just “zomg home runs!”? I hear Russell Branyan is still kicking around the league if a slugger’s all you want.

    • Ari Collins - Sep 6, 2011 at 1:12 PM

      That single paragraph might repudiate WAR.. if it weren’t flat wrong. Ellsbury is nearly the offensive force that Fielder is this year, and instead of playing the easiest position poorly, plays one of the toughest very well. That he’s been nearly twice as valuable this year isn’t that surprising if you’re looking at the big picture instead of “omg homers and RBIs!!”

      The traditional stats community often fails to properly weigh defensive ability and defensive position. Which is always surprising given the number of words they’ve spilled about how pitching and defense wins ballgames.

      • drmonkeyarmy - Sep 6, 2011 at 1:26 PM

        The SABR community has the same short comings when trying to assess defensive value and positional adjustments. Just because people have a means in place to gauge defensive value doesn’t mean it is any good. Nor is a faulty system of measurement necessarily better than nothing at all.

      • Ari Collins - Sep 6, 2011 at 1:32 PM

        Yes, but when you have several different statistical measurements of fielding, and you understand what they’re measuring, their strengths and weaknesses, and then you add additional information from scouting reports, that’s a much better approach than “Fielding stats suck. Jeter is a gold glover.”

      • drmonkeyarmy - Sep 6, 2011 at 1:36 PM

        But when you are adding together a bunch of subjective information does that not make the final supposition exponentially more subjective?

      • Ari Collins - Sep 6, 2011 at 1:45 PM

        “subjective” doesn’t mean “useless”. Defensive statistics have different strengths and weaknesses, as some are better at measuring range, some arm, some positioning, etc. If you’re just throwing the stats together without any understanding of what they measure, then yeah, you’re doing it wrong. Same as using OPSBIs.

      • Chris Fiorentino - Sep 6, 2011 at 1:54 PM

        But no amount of subjective beats watching day in and day out. And I believe that’s what the author is going for…if a stat is subjective, it is no better than watching the game every day.

        There’s nothing subjective about BABIP. That’s why I like the stat. It is a common sense OBJECTIVE way to see how a guy may be playing above his head and whether he may see his #’s go down in the coming months or next season.

        But WAR…bleh. Too much subjectivity goes into that one. It gives a shortstop too much credit for doing half as much as a first baseman. Why would I, as a fan, care whether there are less really good SSs or CFs than there are First Basemen? When I see Prince Fielder and I see Matt Kemp, I wonder…how the f**k is Kemp TWICE as good as Fielder? Their numbers are almost identical. So because Fielder plays first base and is less of a, er, fielder, while Kemp okays center field and is more of a fielder, his WAR is double Fielder’s? Ugh.

      • seanmk - Sep 6, 2011 at 2:46 PM

        well fielder isn’t exactly a gold glover at first base, so yes i’d rather have the non gold glover in the harder position that hits like a first baseman, all things being equal. plus fielder isn’t the best base runner, while i think you might agree kemp is at least an average base runner.

        add it all together

        fielder = elite bat for the position + below average at 1B + below average runner

        kemp = elite bat for position + average to below average CF + average runner

      • Chris Fiorentino - Sep 6, 2011 at 3:12 PM

        But sean, you leave out the most important part…I didn’t say that Kemp merely beats Fielder in WAR…he DOUBLES Fielder in WAR. So “all else being equal” isn’t relative to what I wrote.

    • clydeserra - Sep 6, 2011 at 3:13 PM

      but the article doesn’t repudiate anything, its just some crank yelling get off my lawn.

  6. dailyrev - Sep 6, 2011 at 12:54 PM

    A good analogy probably applies here: in traditional statistical research theory there is an animal known as “meta-analysis”, which involves a logarithmic amalgamation of disparate studies (usually randomized control trials or RCTs) of putatively similar subject matter into a broad (“meta-”) analysis meant to derive generalized conclusions through the identification of common statistical trends.

    The most interesting thing about meta-analysis, however, is that it unwittingly exposes the underbelly of the very God upon whose name most statistical analysis is based — probability theory itself. And what I have seen of modern baseball sabermetrics stands upon the same foundation of shadows as any other statistical methodology based on the century-old Fisher test and its theoretical progeny. I was very fortunate in studying statistics with a professor whose focus was on teaching critical thinking toward probability theory rather than blind obedience to its terms and dictates. Because any theory that goes unquestioned quickly becomes a religion.

    • cur68 - Sep 6, 2011 at 1:57 PM

      Amen. Now try getting your thesis supervisor to see it that way. Save your breath, do the damn re-write, and good luck, that’s my advice.

  7. proudlycanadian - Sep 6, 2011 at 1:07 PM

    There are lies, damn lies and statistics. I have never believed that those so called advanced metrics actually measured what they were said to measure. Garbage in leads to garbage out.

    • manifunk - Sep 6, 2011 at 1:15 PM

      Much like your posting!

    • cur68 - Sep 6, 2011 at 2:47 PM

      That has to be my favourite stat saying PC. A more appropriate ‘in/out’ statement though is “assumptions in lead to assumptions out”. The biggest mistake stats people make is couching things in terms of certainty. The trouble though with uncertainty is that, while its correct to be uncertain, its cumbersome to express.

      WAR’s not a bad stat. It has its flaws and is based on certain assumptions and shouldn’t be the last word on ranking players. It is however a pretty good metric. You rarely find a comprehensively bad player (Adam Dunn this year, for instance) with good WAR either defensive or offensive. As such, when you want to know if a guy is good or not, his WAR is reliable indicator. Its when you have to choose to between guys with similar WARs that the trouble starts.

      It breaks down when you get guys who are good at their position and at the plate. Fans always want to rate one guy against another guy. It’s like the fascination with lists and ranks that we have. But why anyone would want to rate a short stop against a 1st baseman is beyond me. 2 different beasts. You need at least one of each to make a team.

      Rather than ranking Ellsbury against Fielder what we should do is say “If I had to fill out a roster and I had to pick, who would I take?” Then, if you did it by highest WAR, what you get would be something amazing with few arguments, because of course it be Jose Bautista in right field, the Phillies starting rotation, and the usual assortment of rogues you find in the highlights. WAR’s real cool like that. It’s not perfect, but it’s useful position by position.

      • Chris Fiorentino - Sep 6, 2011 at 3:15 PM

        EXACTLY Cur…it all depends on what game you are playing. Are you playing the “I’m a fan” game or “I’m a GM” game. If I were starting a team, I’d take the best shortstop first, even though there are about 8 first basemen better. However, if there are 2 outs in the bottom of the 9th, a SS is probably not in the top 20 guys I want at the plate needing a hit.

      • cur68 - Sep 6, 2011 at 3:44 PM

        And the wonderful thing about it all is that it breaks down with certain guys. Take Pedroia for instance. He would be a SS you’d want up with 2 outs in the bottom of the 9th. WAR is 24.2/6 seasons and you can depend on it really, because this is Pedroia: he gives you as good a chance as you can call “the best chance” to score runs in that situation. WAR you can count on, right? But is he really better than Bautista? Would anyone reject Bautista with his 13.8 WAR/8 seasons over Pedroia @ 24.2/6? Trick question of course. The real answer is that Brett Lawrie with his 2.3 WAR is who you want. I’m certain of this BTW. Just saw it at work yesterday.

      • Chris Fiorentino - Sep 6, 2011 at 4:02 PM

        Pedroia plays 2B and he would definitely be in my top 20, along with Utley and Cano. But I would certainly take Bautista over any of those three.

      • cur68 - Sep 6, 2011 at 4:14 PM

        Hah, busted! :) You said ‘short” and I immediately thought “Pedroia”! Just a slave to my prejudices I guess…but you take my meaning.

  8. vince9663 - Sep 6, 2011 at 1:08 PM

    It would have been polite for Hippeaux to cite Fangraphs own Primer on UZR (http://www.fangraphs.com/blogs/index.php/the-fangraphs-uzr-primer/) for the same reliability issues with small samples that they have already outlined. More polite than trying to pass the ideas off as his own.

    • drmonkeyarmy - Sep 6, 2011 at 1:24 PM

      Are you implying the author is a plagiarist? If you are just come out and say it…don’t beat around the bush.

      • Ari Collins - Sep 6, 2011 at 1:39 PM

        He’s just saying that it’s a little misleading to act as if Fangraphs themselves don’t discuss what their tool can and can’t do.

      • vince9663 - Sep 6, 2011 at 1:41 PM

        I’m not sure “blog plagiarism” is a real thing, but no. I meant it with sarcasm where the unwritten implication is that he was ignorant of the Primer on UZR and didn’t bother to do even basic research on the stat that he’s criticizing. I think he genuinely believes he’s made a unique discovery of a heretofore unknown weakness in UZR.

        If you’d like to think he’s a plagiarist, fine. If you’d like to think he’s an ignoramus, that’s fine too. I’ve left it open-ended for you.

      • drmonkeyarmy - Sep 6, 2011 at 1:41 PM

        Sorry Ari, but saying “more polite than trying to pass the ideas off as his own,” is clearly calling the author a plagiarist. That quote isn’t saying misleading, it is saying that his ideas are stolen.

      • Chris Fiorentino - Sep 6, 2011 at 1:47 PM

        Much like you guys love to see “Phillies fan on Phillies fan” crime…I LOVE to see stat head on stat head crime. I believe that whoever Hippeaux is, he is going to see his membership to SABR revoked, his Excel Spreadsheet renewal for 2012 cancelled, and he may even get thrown out of his mother’s basement :)

        And for you thin-skinned stat heads out there, that was a joke…mostly.

      • vince9663 - Sep 6, 2011 at 1:50 PM

        @Chris

        Pretty sure they also break his protractor and melt down all his many-sided dice when that happens :)

  9. zakharovsa - Sep 6, 2011 at 1:13 PM

    Not to interrupt all the stat-hater jumping for joy, but this article is really a great example of how the stat-based crowd approaches the game. We’re constantly looking for ways to improve metrics, which involves pointing out flaws in the stats that currently exist. That’s how we’ve moved forward from runs created, VORP, and now WAR. We’re not content to wrap it up once we’ve found one stat that looks pretty, which is why we’re not still using batting average and RBIs to evaluate players.

    WAR isn’t the perfect stat, and this article is a good argument for why. Statheads will probably look to refine the formula or even create a new stat, rather than declaring that the field of sabermetrics is worthless. Some commenters in this thread should consider doing similarly.

    • Ari Collins - Sep 6, 2011 at 1:17 PM

      Well said. And I’d add that FG and BR both have put a lot of work into the defensive side of things while acknowledging that it’s not perfect.

      However, acknowledging that the defensive numbers aren’t perfect isn’t to say that they’re not the most useful tool we have.

      • Chris Fiorentino - Sep 6, 2011 at 1:26 PM

        Ari, if a tool is not good, then why use it? Isn’t that what you guys say about ERA, RBI, Pitcher Wins, etc.? Don’t get me wrong…I don’t like Pitcher Wins either, but to say that Pitcher Wins is a stupid way to look at a pitcher’s ability because it is misleading…then to use a misleading stat like WAR because it is the “best we have” is just as misleading…isn’t it?

      • Ari Collins - Sep 6, 2011 at 1:37 PM

        A sledgehammer may not be the best tool for hammering nails, but if you don’t have a standard hammer around, it’s way better than using your fist.

      • Chris Fiorentino - Sep 6, 2011 at 1:39 PM

        I’d rather wait to get a hammer then use a Sledgehammer or my fist :D

      • Ari Collins - Sep 6, 2011 at 1:47 PM

        And if you don’t feel the need to build a house (i.e. understand this game we love), then continue waiting for the golden hammer.

        Me, I’m going to go listen to Maxwell’s Silver Hammer now.

      • Chris Fiorentino - Sep 6, 2011 at 2:29 PM

        I’d rather wait for a stat that doesn’t depend on so much subjectivity.

    • Chris Fiorentino - Sep 6, 2011 at 1:18 PM

      Hope you aren’t referring to me as I never said nor implied that sabremetrics is worthless. What I said is that WAR is worthless. I have always said that and I will always say it. Until it does what the authors says in his piece, it will always be worthless. It gives too much worth to positional advantages and it also hurts sluggers by not taking into account things like clean-up hitters.

      AGAIN, I WANT TO WRITE IT IN CAPITAL LETTERS SO THE STAT HEADS WHO COME ON DO NOT MISQUOTE AND MISUNDERSTAND ME…I DON’T LIKE WAR BUT THAT DOES NOT MEAN I DON’T LIKE THE FIELD OF SABREMETRICS.

      • manifunk - Sep 6, 2011 at 1:19 PM

        Man why do you hate Sabremetrics so much

      • Alex K - Sep 6, 2011 at 1:29 PM

        What do you mean, doesn’t take into account clean-up hitters? Are you saying that a stistic should take line-up construction into account? I’m not trying to be a jerk, I’m really curious.

      • Chris Fiorentino - Sep 6, 2011 at 1:34 PM

        Alex, I’m not sure if you read the article, but I was referring to this specific quote…

        “WAR give incredible weight to scarcity of shortstops, but no weight to the scarcity of pitcher-intimidating, strategy-altering cleanup hitters, which I see as a form of reverse discrimination.”

      • Alex K - Sep 6, 2011 at 1:48 PM

        I did read the article. I was just a little confused by your use of the term clean-up hitter, that all. Thanks for clarifying.

      • Joe - Sep 6, 2011 at 1:52 PM

        But WAR does give weight to the scarcity of slugging 1B/DH-types. Those are less scarce, therefore they get less weight.

        The author seems to be tripped up on the concept of “replacement.” It’s easier to find a replacement slugging first baseman than a 900+ OPS, slick fielding center fielder. So replacement value is lower for the center fielder.

        It’s very likely that the pool of replacement first basemen is better collectively than the pool of replacement center fielders, and therefore the premier first baseman could be a better player than the premier center fielder even with the same or lower WAR. If that’s the case, that doesn’t mean that WAR is useless. That just means that WAR doesn’t tell us what most of us thinks it tells us when comparing players across positions.

        All of the above with the caveat that I understand UZR ratings seem funky on occasion.

      • seanmk - Sep 6, 2011 at 2:27 PM

        positional adjustments are a wonderful thing

      • Chris Fiorentino - Sep 6, 2011 at 2:27 PM

        OK, Joe, then why do I care about replacement value? If the top 15 First Basemen are all better than the best Shortstop, then who gives a rat’s ass what their WAR # is? The problem is that stat heads want stats like WAR to become more mainstream but WAR can’t become mainstream because it is too biased toward position played. It really is.

        Now, if you said “Pretend you are a GM and you are starting a team” then yeah, sure, I would probably take the first SS over the 6th best 1st baseman, for the simple reason that I can grab the 7th or 8th first baseman. But I’m NOT a GM. I’m a fan. And if the 7th best 1st baseman is better than the best SS, then the 7th best 1st baseman is just BETTER than the best SS, no matter what their value is above replacement.

        I guess it just depends on the game we are playing…are we playing GM or are we playing fan.

      • seanmk - Sep 6, 2011 at 3:07 PM

        chris are you only looking at hitting when you say “best” first baseman is better then the best SS? The questions i’d ask you are “is a run saved = to a run scored?” “is it harder to play first base or third base?” and “it it harder to find a elite center fielder or elite corner outfielder?”.

        How a team full of shane victorinos at every position fair against a team of prince fielders at every position, if they both had the same pitchers. does the plus plus bat of fielder’s outweigh the negative glove to victorino’s plus bat and plus defense?

  10. halladaysbiceps - Sep 6, 2011 at 1:16 PM

    ” But do know that the statheads have never been interested in being God.”

    Could have fooled me. I thought they were trying to attain religious cult status or something.

    • manifunk - Sep 6, 2011 at 1:18 PM

      Thus speaketh the head pastor of Philliestianity

    • zakharovsa - Sep 6, 2011 at 1:18 PM

      We’re more interested in fighting against willful ignorance, but thanks for the compliment!

    • Ari Collins - Sep 6, 2011 at 1:20 PM

      Which sounds more like cult dogma: the appeal to tradition and “we’ve always thought this way,” or an attempt to figure out how things work?

      The science-as-faith argument is one of the most illogical ideas in history.

    • halladaysbiceps - Sep 6, 2011 at 1:24 PM

      Gentlemen,

      I will say this. Any sports stat that requires that I perform a quadratic equation or an f(x) function to achieve an answer I have no use for. Give me the brass tax. That’s all I need :)

      • zakharovsa - Sep 6, 2011 at 1:27 PM

        It’s complicated and requires a little more thought than ribbies, therefore it must be wrong! As someone who got B-minuses in high school math, you don’t have to be a math-head to appreciate the value of advanced stats.

      • halladaysbiceps - Sep 6, 2011 at 1:32 PM

        And that’s cool. If you like advanced metrics, I have no problem with that. All I always state is they don’t tell me anything more than I can see in traditional stats. And you are talking to a guy that took up to Calc 3 in college. I know the value of statics, performing f(x) functions and their derivatives, etc. Saber stats just muddy up the waters for me, so to speak, and give me nothing tangible to wrap my head around.

      • kopy - Sep 6, 2011 at 2:43 PM

        “Hey white boy, what’s the square root of this room?!” -Tron, Mad Real World, Chappelle’s Show

  11. sabathiawouldbegoodattheeighthtoo - Sep 6, 2011 at 1:22 PM

    War is a nice guide, but it should not really be counted as a stat. If I recall, doesn’t baseball-reference use a different defensive metric to calculate their version of WAR? Often with vastly different results.

    So we have unreliable and conflicting defensive metrics, suspect positional weighting, and a system of scaling individual stats into ‘wins.’ Again, I am not saying WAR is a bad way to assess a player’s total contributions, but I would certainly not look at it as a piece of concrete information, like say RBI. ;-)

    • seanmk - Sep 6, 2011 at 2:04 PM

      While i agree with the premise that WAR isn’t perfect because it has to calculate defense and we haven’t come up with the best way to quantify it, I can’t help but notice that this comes from a yankee website, which is the team Granderson plays for and that his “defense” is what’s limiting his WAR number. Because of the lower WAR number people write him off as an MVP candidate.
      More information is always better so if we look at different defensive numbers we’d see baseball reference doesn’t love his defense this year with a -0.4, dewan +/- has him at -12 and UZR has him at -5.9. could all three metric be wrong? sure. could they be right? sure.

      from tangotiger

      “I’m all for criticizing WAR. I have no problem with criticizing WAR. If you want to say that WAR is limited because of its reliance on UZR, then fine. It’s when someone CONCLUDES that “WAR doesn’t work” that the person will show himself to be illogical, irrational, or myopic. (Which is another part of the statement that you removed.)

      WAR is a framework, not an implementation. fWAR, rWAR are implementations. To not make the distinction between a framework and an implementation is ONE way of many ways to be myopic. “

  12. clydeserra - Sep 6, 2011 at 3:03 PM

    That was was a bad article. The problems with UZR and TZ and the like are that there are not enough chances in a year to evaluate how good a player is on defense.

    The article ignores the fact that Fenway is a known problem, and alludes to a Dunn effect and never really susses it out. Further as someone pointed out in the comments, the article claims that the Giants positve UZR is evidence of the metric doesn’t work because they rolled out Pat Burrell and Aubry Huff for over 1000 innings without noting that the Torres and Scheirholtz are very good and Cody Ross is passable.

    As someone noted above, WAR is a measure of what happened not a tool to see who is a better player.

  13. paperlions - Sep 6, 2011 at 3:15 PM

    Part of this refers to a comment way up above, but I’m putting it down here in hopes that it will not get lost.

    People need to stop taking WAR out of context….as soon as you forget about the AR of WAR, you will start making silly comparisons and arguments. Ellsbury may be worth twice as many wins above replacement as Fielder this year, but that is NOT THE SAME THING as being worth twice as much. That is NOT what WAR means, and replacement level is not equal for every position.

    Using the above example it should be easy for people to see how Ellsbury could be worth twice the WAR of Fielder because Ellsbury has been nearly the same hitter as Fielder this year, with great base running and defense, while Fielder is a horrible defender/runner. In addition, Fielder plays a position of sluggers and Ellsbury plays a position much farther up the defensive spectrum and plays it fantastically.

    • Chris Fiorentino - Sep 6, 2011 at 3:19 PM

      OK, paper. Then my question to you is, as a fan, who cares about what a guy’s replacement value is? If, as a fan, I’d rather watch Fielder play the game over watching Ellsbury play the game, then why am I wrong? Why is a replacement value important to ANYONE except a General Manager of a team deciding whether to bring a certain player back again next year? Why is that stat important to MVP? Or Hall of Fame voting? Or rookie of the year voting? WAR is a meaningless stat for anyone other than a General Manager of a baseball team. It should not be used to decide MVP, ROY, Cy Young, or, and especially, the Hall of Fame. Why? Because it assigns to much extra value to position, which again, has no real value unless you are building a team.

      • clydeserra - Sep 6, 2011 at 3:25 PM

        I really really hope general managers have better proprietary data on the players than Fangraphs and B-Ref.

      • halladaysbiceps - Sep 6, 2011 at 3:27 PM

        Chris,

        Seriously. These saber people are like cockroaches. Notice they don’t generally talk about a game the night before like we do in the ATH thread? But, once a saber-related article appears on this site, they come out of the woodwork and try to convert you. They are not real fans of the game. They could care less about what you or I think. You are basically debating a group of people that could care less what happens on the field. All they care about is numbers that no logical baseball fan would have a frame of reference with.

      • paperlions - Sep 6, 2011 at 3:29 PM

        As a fan, you don’t need to care about it. Many of the people that work on these types of stats work for teams (with the team’s versions of these stats typically being proprietary)….and teams care, because they want to know how hard/expensive it is to replace particular types of contributions.

        Nonetheless, much of the development of stats that advance understanding have been led by fans….and you you love something so much to put in the kind of work many have done for free (or close to it), it is because you love baseball….not because you love stats/number. There are many ways that the love of baseball may manifest, every way isn’t for everyone.

      • paperlions - Sep 6, 2011 at 3:31 PM

        No offense biceps, but I’m not going to waste my time typing in a conversation about last nights game. Give me a beer and some wings, and we’ll talk….I’m not doing it sitting at my computer at work. I’ll exchange ideas, but I won’t idly chat.

      • Chris Fiorentino - Sep 6, 2011 at 3:41 PM

        HB, I disagree. I don’t think they are died in the wool fans like we are, but they like what they like and that’s all fine with me. Sometimes, they can be a little pretentious with their “stats rule…eyes drool” talk. But for the most part, I enjoy the debate and I usually learn something whenever I debate this issue.

        Paper, OK, as a fan I don’t really care about it to be honest. But the problem is that too many people think that WAR is important to things like MVP, etc and it just isn’t and shouldn’t be. Because the 7th best first baseman, who is an average fielder, shouldn’t be penalized for being a 1st baseman as opposed to the 3rd best SS who gets extra credit for being a SS but his hitting #’s aren’t nearly as good and he is not Ozzie Smith. He’s average defensively and offensively, but his WAR is higher than the 7th best first baseman who has better hitting #’s and is average defensively himself.

      • halladaysbiceps - Sep 6, 2011 at 3:42 PM

        paper,

        You said ” but I’m not going to waste my time typing in a conversation about last nights game.”

        Dude, that’s what baseball all about. It’s not about debating statistics. It’s about what you saw the night before with the particular game that you watched, whether you are a Rays, Brewers, Rangers, or Yankees fan. My god, game to game is what baseball is all about. It’s not about these stats debates. In the offseason, I guess I can understand, since there in no active baseball games. But, its in the here and now.

      • halladaysbiceps - Sep 6, 2011 at 3:50 PM

        Chris,

        I respectfully disagree. Pretentious is not the word. The proper phrase is “I know sabermetrics and you don’t. I believe that I am superior to you in every respect and have greater intellect than you”. That’s what I get from these nuts. I stated it before. These people are so fanatical that they were tweeting Craig a few weeks ago saying I was dumb and to ban me from the site. Some of them run their own paid blogs and come on here to start sh*t.

        Chris, I know you love to debate, like myself. You give some of these people too much credit, though. Your baseball IQ makes you 1000 times smarter than these stat-heads could ever achieve.

      • Chris Fiorentino - Sep 6, 2011 at 3:55 PM

        Well HB, I will just say this…while I will NEVER go for the WAR stat as it is currently figured, there are some pretty good stats out there that are objective and use common sense…like BABIP. But these defensive metrics, where it is a bunch of people watching every play from every game and clicking a mouse when they think the player should or should not have made the play, thus creating Dewan…bleh. These same stats will tell you that Ken Griffey Jr. was just an average fielder when we all watched him play and know he is one of the greatest fielders of all time.

      • halladaysbiceps - Sep 6, 2011 at 4:01 PM

        Exactly, Chris. You know baseball like I do. We watch it for the love of the game, not some numerical value that we put on a player. That’s why I enjoy discussing baseball with you and the Good Doctor. We talk baseball games, not stats, although we may throw in a stat from time to time. I love talking baseball with you and the Good Doctor and we will have fun through the next two months doing so.

        Chris, I appreciate you because YOU ARE THE MAN!!!!

      • paperlions - Sep 6, 2011 at 4:04 PM

        Bicepts, that is what baseball is all about to you; not to everyone. Unlike some on each side of discussions, I don’t try to force everyone to view things as I do….though I do make efforts to point out what I perceive to be errors in presentation, fact, or interpretation….the intent is for edification and enhancement of one’s understanding and enjoyment of baseball.

        If you derive the greatest amount of joy from yelling at the TV and rehashing events online, great. But that is no reason to view with derision those than consume their baseball through alternative mechanisms.

      • halladaysbiceps - Sep 6, 2011 at 4:14 PM

        paper,

        I have no problem with the saber guys. They have an opinion like the rest of us. But, what I don’t understand is the insistance of, when someone such as myself, Chris F. or the Good Doctor throw out something like “Ryan Howard is having a great season” several of the saber guys have to come on this site and tell us that we are wrong, provide advanced metrics to prove they are are right and fight and belittle us. I have a problem with that. If they watch Howard play everyday, please critique him. But, don’t come at us with his WAR, etc. and that’s all you got. Baseball is more than these trivial stats. It’s like we are dumb in their eyes because we don’t subscribe to Saber Digest.

      • Alex K - Sep 6, 2011 at 4:16 PM

        biceps, Why can’t we have it both ways? I watch most Cubs games because I’m a Cubs fan. I watch random other games because I am a baseball fan. But I also enjoy getting on HBT and talking about who is better than who, and using stats like OBP, WAR, wOBA, etc to make my case.

      • halladaysbiceps - Sep 6, 2011 at 4:36 PM

        Alex K.,

        We can have it both ways. I have no problem with Saber guys like yourself throwing the advanced metrics equation out there, as long as it’s not to belittle a commenter and tell them they are stupid, not that you have personally done it. I’m sensitive to this because it was a few weeks ago when I got into a heated discussion with a saber guy over the Ryan Howard thread and he (who is a paid blogger at another site) tweeted to Craig that I was dumb and should be banned. That stuff cannot be tolerated. Disagree with me? Fine. Don’t go behind my back and tweet to the guy that runs HBT calling to ban me. I don’t think that’s fair and I have a chip on my shoulder because of that, along with earlier beratings that I took from the saber crowd when I brought up traditional stats. That’s my issue.

  14. Mark - Sep 6, 2011 at 3:16 PM

    I think there are some good points and negative aspects in the article.

    If you’re going to use UZR for WAR, it should be based on the past 2-3 years and included the current year. Averaging the data, with an emphasis on current year. While a one year sample size can fluctuate, it’s likely to be more consistent over several years. I don’t agree that Ellsbury is a 15 UZR player, and that adds a win and a half to his value. Averaging his UZR from the past 2 years and this year gives him 2.5, which would make sense given his arm and the fact he often takes poor routes. Slightly above average, but not an elite defender.

    This would also eliminate the problem with Carlos Lee – using my method he’d be at -5.2 UZR, which drops him down by 1.4 wins. Which makes more sense, given his poor production at LF/1B is identical to what he had last year.

    The point about flyball % is interesting, and I don’t know if that’s been addressed. I would like to see that incorporated in the future as well, because more opportunities to lead to a potentially higher UZR IIRC.

    The funny thing about UZR or other stats that people disagree with, is if there’s an odd result people go “THE STAT IS FLAWED, SEE I TOLD YOU SO!”. But that’s not always the case. If a hitter goes from being a 290-300 hitter to 220, we don’t say batting average is flawed. When Jose Bautista hits 40-50 HR’s, we don’t say HR’s are flawed. Yet people do that with UZR or WAR all the time. Additionally, a full season of defensive innings at one position is like half a season offensively. I wouldn’t judge a guy based on 250-300 at bats (because even a lousy hitter can stay hot for that period of time), so I shouldn’t judge one based on one season of defensive innings. Which again, is why it’s important to average it out over 2-3 years.

    WAR doesn’t hate sluggers – it’s just that there are so many dangerous bats at 1B/LF, that in order to be considered an above average slugger you have to do something to separate yourself from the pack. Of the qualifying 1B, 7 are above a 900 OPS – and the next best is 859. Shouldn’t be shocking then, that there are currently 8 players at 3 WAR of greater for 1B. The guys below them are middle of the pack, and graded accordingly at 2-3 WAR. Which is fair, because an 825-850 is the average for qualifying 1B. Exactly what 2-3 WAR is designed to show.

    I did like the comment about how Pedroia has a higher WAR despite similar traditional stats to Melky. That 50 point OBP edge is just a TINY difference :) .

    I’m not trying to suggest that it’s perfect (as no stat is), but I don’t believe it’s as flawed as he suggests. There are improvements that can and should be made to improve the effectiveness (Specifically related to averaging UZR’s to eliminate sample size error). But it is an effective tool, when used properly. Unfortunately, a lot of people just go “PLAYER X HAS 7 WAR HE > guy with 6.9 WAR”. And that is about as helpful as the people who say “WAR IS FLAWED I LIKE MY RBIZ THEY ARE SO MUCH BETTER!!1!”.

    I think what’s important here is to try and figure out ways to improve it. Conceptually it’s a good idea, and a lot of the right pieces are already there. It’s just about fixing up some of the defensive issues, which seem to be the most troubling part of WAR (especially for 1B/C).

  15. Chris Fiorentino - Sep 6, 2011 at 3:22 PM

    “If a hitter goes from being a 290-300 hitter to 220, we don’t say batting average is flawed. When Jose Bautista hits 40-50 HR’s, we don’t say HR’s are flawed. Yet people do that with UZR or WAR all the time”

    Because HRs and BA are tangible numbers. UZR and WAR are subjective stats, and WAR especially, gives extra credit for position. If you hit 40 HRs as a SS, then you hit 40HRs. You are behind a 1st Baseman who hit 55 Home Runs, no matter how much extra credit WAR wants to give you for playing SS.

    • clydeserra - Sep 6, 2011 at 3:28 PM

      your interpretation may be subjective, but the data* and formula’s used aren’t.

      *I realize that stringers plat what zone a fielder fields the ball in and that can vary from person to person, but I really don’t think that is any more subjective than hit/error.

    • Mark - Sep 6, 2011 at 3:31 PM

      Right, but a 40 HR SS is more valuable than a 55 HR 1B. There isn’t a single SS who’s hitting 40 HR’s.

      You gotta figure a 40 HR SS is around a 575-630 slugging (just based on Granderson/Bautista’s slugging to give you an idea). There are exactly 0 SS’s with a 600 slugging, 1 with 550, and everybody else is 500 or significantly lower. There are 3 1B above 550, and 8 above 500.

      Please tell me you’re able to see the difference here. A 40 HR SS is Alex Rodriguez in his prime. A 40 HR 1B is above average, but not in the discussion for the best player in the game.

      So again, the reason is pretty simple. The average SS has a 700 OPS. The average 1B has an 825 OPS. A SS that is posting a 900 OPS (AKA a 40 HR SS) is significantly more valuable than a 900 OPS from a 1B.

      This isn’t subjective, this is common sense. If your team could have a 900 OPS SS and a league average 1B (825 OPS) or a league average SS (700 OPS) and a 900 OPS 1B, which team is getting more production? That’s why positional adjustments make too much sense. You may disagree with the values of positional adjustment, but they are very real and very essential to any discussion with or without WAR.

      • Chris Fiorentino - Sep 6, 2011 at 3:48 PM

        But Mark, again…are we building a team or are we evaluating one player vs another player.

        Like I said, sure I would take the second scenario every time. But if I asked you who is the better player…the league average 1B (825 OPS) or the league average SS (700 OPS) I would say the 1B, but I guarantee WAR would either be equal or the SS would probably be a little higher. And that is where WAR fails to me as a fan.

      • Mark - Sep 6, 2011 at 4:43 PM

        A league average SS is just as valuable as a league average 1B regardless of whether you’re evaluating one player or building a team. This has nothing to do with WAR. Honestly, it’s like asking what’s heavier between one ton of bricks and one ton of feathers. A league average SS is just as valuable as a league average 1B.

        I think what you’re arguing is that the 1B is better because he hits for a higher OPS. So yes, his offensive contribution is more valuable. But we can’t simply ignore that the SS’s defensive value is far greater than that of the 1B. I’m not talking about WAR here. Look at Rollins for instance – his value isn’t just from his bat, but from his glove.

        It’s tougher as a fan to assign value to a defensive play. And I think that’s the problem that a lot of people run into. But if we’re going to stick to arbitrary terms such as “league average 1B” and “league average SS”, that means that both players are creating the same value. The 1B is just doing it with his bat, and the SS is likely doing it with the combination of his glove and his bat.

        So while you may feel that’s where WAR fails for you as a fan, I think it’s one of WAR’s greater strengths. Too often we simply ignore the defensive contributions of players and just focus on their OPS. WAR balances that out, by accepting that value comes from base running, from defense, and from the different offensive standards of each position. Because we shouldn’t just evaluate hitters on their offense.

      • Ari Collins - Sep 6, 2011 at 8:15 PM

        Very well said, Mark.

  16. thefalcon123 - Sep 6, 2011 at 4:06 PM

    The coorlelation between WAR in team wins isn’t great. It’s not bad, but sometimes, you’ll see 86 win teams leading the Majors in collective WAR and 95 wins teams finishing 8th or 9th. I think WAR is a really great approximate snapshot of how good a player was, but it’s not a good exact metric.

    For example, if player A had a 7 WAR season and player B had a 3 WAR season, it’s pretty clear player A was significantly better.

    On the other hand, if player A was 5.5 WAR and player B was 4.7 WAR, it seems kind of silly to say player A was better based on WAR alone do to the correlation issue above.

    • Chris Fiorentino - Sep 6, 2011 at 4:17 PM

      “For example, if player A had a 7 WAR season and player B had a 3 WAR season, it’s pretty clear player A was significantly better. ”

      But falcon, this really isn’t always true. Look at this season with Matt Kemp. His WAR is 8.6. Adrien Gonzalez WAR is 6.0. Has Kemp really been 33% better? Sure he steals more bases and has hit a few more home runs. But A-Gon has higher BA and OBP, has 25 more hits, and is a great fielder. Yet he gets punished because he does all of that as a first baseman, whereas Kemp is a CF. So he gets extra credit. It’s the whole extra credit thing that annoys me with regards to WAR…when comparing player A vs player B. Now, if you asked me who would I rather have if you gave me the first pick of this year’s players, then I would take Kemp because CFs putting up these #’s are more valuable. But that is me playing GM. If you asked me who is the better player this year, I would still take A-Gon and his 6 WAR over Kemp player for player.

      • skipperxc - Sep 6, 2011 at 4:25 PM

        The trouble is that you would never put Adrian Gonzalez in center field. Sawks Nation pitched a fit when he played in right for a couple games this year — he’s just not an outfielder.

        And that by its very nature makes him less valuable. The offense between him and Kemp is pretty close overall, so being unable to play him anywhere but first makes him a lesser player. Marginally, maybe, but lesser nonetheless.

      • churchoftheperpetuallyoutraged - Sep 6, 2011 at 4:44 PM

        Has Kemp really been 33% better? Sure he steals more bases and has hit a few more home runs. But A-Gon has higher BA and OBP, has 25 more hits, and is a great fielder.

        You do know that WAR is a combination of stats, and you can take out/regress the ones you don’t like but you’ll have to do that for everyone, and you’ll just end up calculating them and creating your own version of WAR?

        Regardless, let’s look at the two players:

        MK – .320/.400/.570
        AG – .339/.404/.548

        Very similar, I’d give the edge to Kemp and here’s why:
        MK – 11.6 BB%
        AG – 9.4% BB%

        Gonzalez’s OBP is inflated due to his higher BA as Kemp is walking in far more plate appearances. Kemp also has a higher SLG%, and a far higher ISO (.250 to .208). This leads to a wOBA advantage for Kemp of .414 to .405 and a wRC advantage for Kemp of 166 to 154 (8% better based on wRC)

        So we could say Kemp’s “batting runs” are better than Gonzalez.

        Now onto base running. Kemp is 37/46 in stolen bases and Gonzalez is 1/1. Considering body type I can assume we don’t need to argue who’s better going first to third. So we have 0.7 BSR for Kemp vs -6.3 for Gonzalez?

        Just adding those two up you could get your own mini version of WAR (Kemp 47.9 BAR + 0.7 BSR = 48.6 to Gonzalez 41.6 -6.3 = 35.3. Roughly 38% better).

        But now you run into the problem parts with WAR. People don’t like replacement value. You could choose average player, but that brings about two issues. One, you’d have to find out what the average person at each position would do and then rerun the numbers. Grab an excel spreadsheet and b-ref.com data and get to work! Second issue is that you’re going to get a lot more people below average (at a negative number) than with replacement. Think about it like a bell curve. You need just as many F’s and D’s as B’s and A’s. So you come up with your own replacement value.

        But wait, playing CF is a lot more difficult than playing first base. Shouldn’t Kemp get a benefit to playing CF as it’s a harder position? So now you’ve added positional adjustment.

        Last but not least is defensive value.

        Add those all up, and you get WAR. Throw your initials in front of it if you want, but you’ve done essentially the same thing as fangraphs/bref.com do.

  17. thefalcon123 - Sep 6, 2011 at 4:33 PM

    Yes, I will certainly argue Kemp is 33% better. Gonzalez should be punished because he does it at first. Why? Because it is easier to find a 1st baseman who will hit at an acceptable level than it is to find someone capable of playing center field who can hit at an acceptable level.

    Go through baseballreference year by year. Pick out the 5 best hitting 1st baseman, catchers, center fielders, shortstops, ect. You’re top 5 first baseman with all probably be .300 hitter with 30 homers and a .900 OPS. You’ll maybe get 1 or 2 center fielders with those numbers. Maybe, MAYBE 1 shortstop and probably not any catchers. This is because 1st base is a less demanding defensive position, therefore, there are more good hitters who can play there.

    Another example. In 1991, Cal Ripken hit .323/.374/.566. Frank Thomas hit .318/.453/.553. Thomas’ OPS was 66 points higher. Who would you rather have had on your team? Cal Ripken, unless you were a f**king idiot. No offense to the greatness of Frank Thomas, but there was zero chance you would find another shortstop who hit nearly as well in 1991.

    The 5 highest OPS’ for 1991 shorstops were:
    .940, .884, .757, .756, .747

    The 5 highest OPS’ for 1st baseman were:
    1.006, .922, .895, .890, .888

    This is why you measure each position based upon it’s replacement level. It’s how much value can you add over your next best option.

    FYI, the defensive spectrum goes (pretty much) like this C, SS, 2B, CF, 3B, RF, LF, 1B

  18. phillyphreak - Sep 6, 2011 at 4:34 PM

    OMG those stat guys…..they don’t even like baseball! OMG OMG.

    Did anyone point this out (taken from a post in the comments section that the author made)…

    “This is good point. I slightly misrepresent how the stat works here, in favor of making the statement more hyperbolic. “

    • skipperxc - Sep 6, 2011 at 4:54 PM

      I wanted to reach through the internet and strangle him for that comment. Stuff like that is completely out of line in a useful discourse. If your point isn’t strong enough that it needs that kind of underhandedness to support it, maybe consider not making such a point.

  19. seeingwhatsticks - Sep 6, 2011 at 4:43 PM

    In all fairness, depending on the component and the age of the car, a bad part in the engine can mean the car is totalled.

    I don’t care enough about this argument to wade into the details except to say this: if you build a house and the foundation is faulty it really doesn’t matter if 95% of what you did was right when the whole thing collapses.

  20. diehardcubbiefan4life - Sep 6, 2011 at 7:24 PM

    Say what you want about Sabermetrics, but nothing is more accurate then actually watching the game and making your own judgement.

    • phillyphreak - Sep 6, 2011 at 7:35 PM

      Yes. And wouldn’t it be fun if we could take what we watch and use those numbers for something…hmmmm.

    • jwbiii - Sep 6, 2011 at 8:02 PM

      Which exactly describes UZR, except the people who do that watch a lot more games than we do.

  21. pitperc - Sep 6, 2011 at 7:56 PM

    @Chris

    I’ve read through the entire set of comments. I think you’re misinterpreting one *really* important thing about WAR.

    WAR has nothing to do with who the “best” player is. (For the record, I’m not even sure I know what “best” player means)

    You’ve asked several times if we’re playing the “who would I rather take on my team” or “who’s the best player” game (or some variant of that kind of proposition).

    WAR won’t tell you who is the “best”. It’s a tool used to determine the value of a player. It can help you determine which player is more valuable, it can help you determine which player is more difficult to replace, it can help you determine which player is having a greater impact on their team’s overall success, it can even help you figure out which player is having an off-the-charts kind of a year. But it has nothing to do with helping you determine who’s the “best”.

    I read earlier some critiques of advanced statistics being too “subjective”. It can’t get much more subjective than asking two baseball fans which player out of two is the “best”. I can use WAR to tell you who was more valuable, harder to come by, or a greater contributor to team success, but it takes some truly subjective work to make the jump to “best”.

    Many have already made that point above me, so I won’t rehash. I did want to point out though that you seem to be misunderstanding the core purpose and intent of WAR.

    You’re not wrong if you prefer to watch Fielder player over Ellsbury. Having that preference is totally legit. But you’re wrong if you tell me Fielder is a better ballplayer than Ellsbury because you like to watch him play more.

    • Chris Fiorentino - Sep 6, 2011 at 10:41 PM

      Right, pit…WAR WON’T tell you who is the best. Yet, people will use it to say Player A is better than Player B. Not everyone does this…but more than you may think. Whenever they put up an article on HBT and ask “Who is the MVP so far” or “Who is the Rookie of the Year so far” they ALMOST ALWAYS start with a listing of players ranked by WAR. That kind of thing is where my beef is. WAR may tell you a player’s value to his team, but it doesn’t tell you anything about the best player so it shouldn’t be a part of the discussion of who is the best player.

      Put it this way…if I am starting a team tomorrow, I am taking Albert Pujols #1 no matter how much higher Matt Kemp’s WAR is.

      • phillyphreak - Sep 7, 2011 at 7:21 AM

        I think you just don’t understand WAR.

      • gcil - Sep 7, 2011 at 4:25 PM

        I think most people that have a problem with WAR only have a small problem with the actually statistic. The majority of their complaint stems from the way a lot of people have started to use it, and how they discuss it, and other sabermetric stats.

        You see more and more “sabermetric” people, who have some understanding of the stat, use it as if it is the only statistic that matters. They make it appear as if it is an all encompassing, debate ending metric that is infallible. The truth, as actual proponents of sabermetrics would probably tell us, is that it is a very good stat that needs to be refined, and that it is most certainly not without it’s flaws.

        I’m pretty new to advanced statistics, but I enjoy them and I enjoy reading articles and books that explain them, and use them in real life instances. What I’ve come to realize though, is that there is the actual sabermetric community that invents these statistics, explains them in simple terms, discusses them with people who want to learn more, and then refines them to make them better where they can. And then you have the people (not everyone, but definitely the vocal bunch) who populate message boards, that take these stats as gospel and degrade anyone who questions them. Which I find ironic considering these stats are born out of people questioning the state of current statistics, and what those stats actually tell us. People who are unwilling to accept criticism of these statistics are just as thick headed as the people who are unwilling to give these statistics a chance. There probably worse to be honest, because they should be the ones questioning and looking for ways to improve, rather than resting on their laurels.

        When I first started getting into these stats, I was ridiculed by people on a different board for questioning whether a pitcher could have an actual affect on his BABIP. I thought that the idea of a pitcher having no control on how the ball go into play made little sense, considering if you throw the ball in the middle of the plate it’s more likely to get hit than a ball up and in, or low and away. As a Phillies fan I watched Moyer, Glavine, and Maddux induce weak contact on a regular basis, which I thought, would decrease their BABIP. When I stated this, it was as if I had just said “Hey guys, I’m pretty sure the Sun revolves around us.” It was like I was some caveman who was too ignorant and moronic to understand these glorious sabermetric statistics. I think the smugness exhibited by the vocal bunch is what can turn a lot of people off.

        There are a lot of people who use some of these stats, WAR specifically, in the wrong contexts, but act as if they are professional statisticians or engineers. They use WAR to compare players at different positions, and to say player X is better than player Y. That’s not really what WAR is used best for, but you rarely see them corrected by someone who knows this.

        It is used to show how much better a player at that particular position was better, in wins, than the waiver wire pickup would be at that position. A SS to a SS, a 1B to a 1B, not a SS to a 1B. It also has shortcomings in the defensive aspect, and it should never be used in the vacuum that a lot of people use it in. Its a good stat, but like any other statistic, you can’t only look at WAR, you need to dig deeper.

      • Chris Fiorentino - Sep 8, 2011 at 10:11 AM

        gcil, thank you for perfectly putting into words my exact feelings on the sabremetric community as a whole. And also for explaining much better than I ever could, exactly what I was trying to say about 50 times but never could make as clear as you just did. SS vs SS…YES!! 1B vs 1B…YES!!! SS vs 1B…NO!!!

        Thanks again!!

      • churchoftheperpetuallyoutraged - Sep 8, 2011 at 1:40 PM

        A SS to a SS, a 1B to a 1B, not a SS to a 1B.

        It does this perfectly, in two separate ways. You can first just do it by adding all the components together, and second by including the positional adjustment.

      • gcil - Sep 8, 2011 at 3:25 PM

        Statistics like this don’t really do things perfectly. And the problem with comparing it across positions is the actual positional adjustment number itself. It may give us a good idea, but its not perfect. If a 1B and a SS have identical offensive numbers, the SS presumably gives you more because of his defense. However, the difficult part is quantifying how much more. What happens when their offensive numbers aren’t that close? Positional adjustment numbers are not concrete, and are based on a small sample of players who have played multiple positions.

        Furthermore, is a poor fielding SS worth more than a poor fielding LF? Is he actually worth less because he’s hurting his team more by being in a valuable defensive position and playing poorly?

        There’s a definite gray area in here, and I prefer to leave it out by not using WAR and WAR alone to compare players of different positions.

      • churchoftheperpetuallyoutraged - Sep 8, 2011 at 5:02 PM

        Except, as some of us have mentioned numerous times, you can take apart “WAR” into it’s components, remove/regress the ones you don’t like, and rebuild it. Call it gcilWAR if you want.

        Take the two players you mentioned, a 1b and SS, look at their batting runs. Look at their base runs. You are making comparison’s against each other that haven’t factored in position, yet. Feel free to add a positional adjustment after. You’ve now created 60% of WAR by yourself.

      • gcil - Sep 8, 2011 at 11:21 PM

        That’s all well and good except that I’m discussing the actual WAR statistic, and not its components. If you take it apart and reconstruct it differently then it is no longer the WAR that this article is discussing. I always prefer to look at multiple statistics as the more you use the clearer the picture usually becomes. Taking “WAR apart” is usually what I essentially do when looking at players. However, we are discussing WAR how it exists currently and how it is used by many people. To me, that WAR is best used for comparing players at the same position.

      • Kevin S. - Sep 9, 2011 at 12:06 AM

        Go into the Fangraphs glossary and read the entry on how Tango derived the positional adjustments. I think that will clarify a lot of the confusion you seem to have here.

      • churchoftheperpetuallyoutraged - Sep 9, 2011 at 8:28 AM

        If you take it apart and reconstruct it differently then it is no longer the WAR that this article is discussing. I always prefer to look at multiple statistics as the more you use the clearer the picture usually.

        Except the article is a hatchet job on UZR/defense and uses WAR as hyperbole to get page views. The author said so him/herself in the comments section. And this is just as much of a problem as people who blindly use WAR as a tool.

      • gcil - Sep 9, 2011 at 2:12 PM

        Kevin,

        I did read how Tango derived his positional adjustments, and I know that they are generally ok, and something he just posted did clear up a question I had. They follow an order of difficulty that makes sense to the general baseball fan. However, the specific numbers is where the errors can come in. They are based off of a “relatively” small sample size, and also off of UZR. I’m not a particularly huge fan of UZR. I think its a good step in the right direction, but I don’t put much stock into it because I feel like it varies too much from year to year.

        I know people say that offensive stats can vary a lot, but no one questions them. However, there are reasons for that. Offensive statistics are mostly straightforward, are much easier to measure, and are far more concrete. They are also dependent on the batter recognizing the spin, location, and velocity of a pitch in milliseconds. Infielders have more than milliseconds to read the ball off the bat, and outfielders have much more time. So you would think that UZR results would be more consistent than offensive results, especially for outfielders.

        The problem is they can vary an awful lot (and apparently with fly ball percentage). That is why it is hard for me to use the specific positional adjustment numbers based on a statistic that will no doubt be refined in the future. If you’re trying to compare players that are relatively close in their WAR but not in their offensive numbers, how can you definitively say that the CF with a 5.2 WAR is better than the 1B with the 4.8 WAR? The difference is attributed to the CF’s positional adjustment, and to his defense. Those two areas, IMO, are where the murkiness in WAR creeps in.

        This also brings me back to what my point really is. The problem isn’t so much WAR itself (defensive aspects aside) but more so how it is used by the mainstream. I’d much rather use WAR as a starting point and go from there, than use it as the final word on the subject.

        Church,

        Would you agree with the point I made above? That the way people are now using WAR as a singular, all encompassing statistic that definitively proves player A was better than player B is the actual problem?

        I get what your saying about how you can make your own WAR because its technically a framework, but many people have turned it into the be-all-end-all of baseball statistics. It doesn’t help that Fangraphs lists players in descending order of WAR when you look at their leader boards.

      • churchoftheperpetuallyoutraged - Sep 9, 2011 at 5:16 PM

        So you would think that UZR results would be more consistent than offensive results, especially for outfielders.

        Not sure why you feel this way, because may of the biases inherent in defensive statistics cause the results to vary. The stringers generally don’t see the players positioning prior to the at bat, so if someone is shaded far to his right and can’t make a catch to his left, that he would have made if he was in a normal position, he’ll get docked for it. But should he have?

        If you’re trying to compare players that are relatively close in their WAR but not in their offensive numbers, how can you definitively say that the CF with a 5.2 WAR is better than the 1B with the 4.8 WAR?

        Because you shouldn’t, and we’ve been trying to get people to not do so blindly. When the numbers are that close, look at all the factors involved. Is it merely due to defense? Or positional adjustment? If everything is the same except batting runs, I’d feel more comfortable about a .4 difference in WAR. But if not, a general error bar of .5 WAR is what most are using now.

        I agree that people shouldn’t blindly quote a statistic as the be all and end all, but just because fangraphs sorts by WAR as a default doesn’t mean people can’t look at the leaderboards and do their own sorting, by wOBA, by wRC, etc.

  22. President Miraflores - Sep 6, 2011 at 10:50 PM

    I don’t really have a dog in this fight, but according to Baseball Reference, Mark Texeira has a dWAR of 0.0. Brett Gardner, Alex Rodriguez, and Nick Swisher have higher dWAR’s than Texeira. Am I to believe it is easier to replace Texeira defensively than the other three? Is Texeira a merely average defensive 1B?

    Can someone make some sense of this for me?

    • churchoftheperpetuallyoutraged - Sep 7, 2011 at 8:36 AM

      A couple of things to note:

      1. You generally need three full years of defensive statistics to stabilize before making a judgement on whether a player is good or not.
      2. Different systems use different measures to evaluate defense. b-ref.com uses DRS(a) and fangraphs uses UZR(b)
      3. Because of (1), in season UZR is highly volatile so be careful making any conclusions based off just a few months of data.

      As to your question, the defense component measures the player against other players at that position. According to the glossary at fangraphs, DRS measures how often a player makes a player vs an average player(a). If the person has a positive rating, he’ll make more plays than the average, and a negative vice versa.

      So if those three players all have positive ratings, it is merely saying they are better than average at the position.

      As for whether it’s easier to find a better defender at 1b than the others? If they are using the true meaning of average, it shouldn’t. As with a bell curve, you need just as many on the right side of the line as on the left.

      Also note, 1b and C are the two positions that the defensive systems have difficulties calculating. Far too much happens that’s different from all other positions. It’s a known issue and they are trying to rectify it.

      (a) – http://www.fangraphs.com/library/index.php/defense/drs/
      (b) – http://www.fangraphs.com/blogs/index.php/the-fangraphs-uzr-primer/

      • President Miraflores - Sep 8, 2011 at 6:25 PM

        Thanks for the detailed response.

  23. Bryz - Sep 7, 2011 at 9:39 PM

    I find it interesting that the argument is about defensive stats like UZR and how subjective it is. Well then, what are errors? How about a player’s range? Certainly comparing Carlos Lee to Brett Gardner in the outfield is an apples and oranges argument, but what about Jacoby Ellsbury to Brett Gardner? Line up some scouts, and will they have an objective way to rate one of these guys over the other? No.

    Errors are the same way. Some are obvious, but others determine the subjective opinion of an official scorer. So while UZR and WAR may be considered subjective, it’s not like traditional statistics are immune themselves.

Leave Comment

You must be logged in to leave a comment. Not a member? Register now!

Featured video

Pujols deal still worth it to Angels?
Top 10 MLB Player Searches
  1. I. Davis (4022)
  2. K. Farnsworth (2683)
  3. M. Harrison (2618)
  4. M. Minor (2600)
  5. D. Robertson (2594)
  1. M. Perez (2403)
  2. I. Nova (2398)
  3. C. Gomez (2364)
  4. M. Cuddyer (2363)
  5. J. Mejia (2331)