oWAR & dWAR: You keep using those words. I do not think they mean what you think they mean.


The other day I was having a chat with Graham and Dan about Bill Dahlen, Jack Glassock, and the 2016 Pre-Integration Committee ballot. You know, like normal young men do.

Graham asked me if Dahlen’s case is still strong if you don’t use WAR. I replied…

That’s a neat little fact about my three favorite non-Hall of Fame shortstops… and Ed McKean. So Graham asked about McKean (who I called one of the worst defenders ever):

Ed McKean

Ed McKean: positive dWAR, but terrible fielder. (Image via Wikipedia)

Ugh. Here we go. dWAR.

Sure, McKean’s dWAR was positive. That’s good, right? Well guess what… McKean played a total of 1,655 games. 67 shortstops played that many games. How many had a worse dWAR than McKean?

Rk Player dWAR G
1 Derek Jeter -9.2 2602
2 Shawon Dunston -2.0 1814
3 Ed McKean 4.3 1655
Provided by Baseball-Reference.com: View Play Index Tool Used
Generated 3/6/2014.

That’s it. Two. The truth is that 4.3 dWAR is terrible for a shortstop. The reason the number is positive is that it includes the positional adjustment. So, for a shortstop to get a negative dWAR, he needs to be terrible on a legendary level (sorry, Mr. Jeter). Of the 67 shortstops, the average dWAR was actually 17.7.

Let’s step back for a moment. I was an early user of Sean Smith’s Wins Above Replacement framework. When Sean Forman at Baseball-Reference acquired it for use on the site, it was like watching two of your best friends get married. Seriously, it was beautiful. I could see them having babies together. Babies like Mike Trout.

One thing Sean (Forman) has introduced since taking the reigns is the concept of oWAR (offensive WAR) and dWAR (defensive WAR). It’s not that these stats by themselves are terrible. It’s just that they’re terribly named and as a result, I have never seen them used properly.

Sean actually explains the use for oWAR quite well:

If you think all defensive measure are bunk and no better than random noise, use oWAR. It is every part of WAR, but assumes everyone is an average defender.

But that’s not how people use it. The latest example that got me fired up came from Jayson Stark writing about Joe Mauer yesterday. Stark says:

According to baseball-reference.com’s indispensible Play Index, Mauer’s offensive contributions alone were worth 30.4 Wins Above Replacement from 2008-13. Just four other players beat that: Miguel Cabrera (40.4), Albert Pujols (31.6), Robinson Cano (31.5) and Joey Votto (30.9).

The big problem here is the “Mauer’s offensive contributions alone” part. oWAR absolutely does include part of Mauer’s defensive contribution—his position. And that’s huge for a catcher.

Just to recap, here are the components of a position player’s WAR:

  • Batting, expressed on Baseball-Reference as Rbat and in the Play Index as “WAR Runs Batting”, includes only what the player did at the plate.
  • Baserunning, expressed on Baseball-Reference as Rbaser and in the Play Index as “WAR Runs Baserunning”, includes not only the run values of a player’s stolen bases and caught stealing, but how successful he is at advancing on the base-paths during other plays.
  • Double plays, expressed on Baseball-Reference as Rdp and in the Play Index as “WAR Runs DP”, expresses how much value the player provides in avoiding the double play. Players who ground into a lot of double plays will have a negative value while players who avoid them gain value.
  • Fielding, expressed on Baseball-Reference as Rfield and in the Play Index as “WAR Runs Fielding”, shows how well a player fields his position compared to the average defender at his position. I used bold and italics for a reason there. The player is compared not just to the league average, but to the league average for his position. Obviously, the average shortstop is a much more valuable player than the average left fielder. That is captured in the next component…
  • Position, expressed on Baseball-Reference as Rpos and not searchable in the Play Index, is an adjustment made to give players of important position credit for the fact that they’re filling more difficult roles.
  • Replacement, expressed on Baseball-Reference as Rrep and not searchable in the Play Index, fills the gap between replacement and average.

oWAR is simply all of the above except fielding. So, it may not contain the fielding, but it still contains the positional adjustment. Therefore, oWAR is not just offense. Not only does it include position, it also includes base-running and double plays. Stark would have been better off using the Play Index to search Rbat (or perhaps Rbat, Rbaser, and Rdp combined) from 2008–13. Here’s how Mauer ranks using just Rbat…

Rk Player Rbat
1 Miguel Cabrera 300
2 Albert Pujols 259
3 Joey Votto 251
4 Prince Fielder 217
5 Ryan Braun 207
6 Matt Holliday 200
7 Adrian Gonzalez 191
8 Joe Mauer 163
Provided by Baseball-Reference.com: View Play Index Tool Used
Generated 3/6/2014.

Mauer is still excellent, but he does drop a few rungs.

So, to recap: oWAR is not just the offensive contribution. It is actually a version of WAR that assumes every fielder performed at the league average. Position is included, but fielding is assumed to be average.

Sean actually created oWAR for the people who “think all defensive measure are bunk and no better than random noise”. He wanted them to still be able to use a version of WAR. Honestly, I’m guessing if they think the defensive measures are bullshit, they probably have similar concerns with positional adjustments and park adjustments. I think oWAR and dWAR should disappear. I think they’re hurting WAR more than helping (by spreading false information).

So what exactly is dWAR? My fired up self said this:


I don’t often get fired up and irrational, but perhaps “useless” took it too far. Luckily, when one of Dan and I get crazy, the other is there to help. Dan offered this:


Which led me to this:


Which in turn led me to writing this article. See! Twitter isn’t just for breakfast photos and Kardashians!

So, to recap again: dWAR does actually isolate just the defensive portions of WAR. But you need to remember that it includes the positional adjustment. The fact that Ed McKean has a positive dWAR and Jeff Bagwell has a negative dWAR does not mean that Bagwell was a bad fielder and McKean was good. In their cases, Bagwell was phenomenal at a low value position and McKean was among the worst at the highest value position. In the WAR framework, McKean was still more “valuable” to his team because he filled that very difficult shortstop position.

The main reason I dislike dWAR is that you can’t just add it to oWAR and get WAR. The names imply that they’re split but they’re not. The positional adjustment is included in both. Plus, it is repeatedly misinterpreted.

Let’s just all start using oWAR or dWAR correctly. Or better yet—not at all. Just use WAR or the individual components.

52 thoughts on “oWAR & dWAR: You keep using those words. I do not think they mean what you think they mean.

  1. 1
    Bert says:

    Quite frankly, the real problem is WAR as a universal construct has so many flaws, it’s probably better to avoid using it altogether whenever possible. As for Ed McKean being a terrible fielder, that may very well be true, but I wouldn’t rely on dWAR for that conclusion. Today’s defensive metrics are suspect, so I wouldn’t exactly put much weight in comparisons based on 100-year old data.

    • 5

      It’s not just WAR that dislikes McKean. Bill James gave his defense a rare “F” and he’s pretty much universally considered weak. Otherwise, with his offense he’d probably be talked about for the Hall of Fame.

      WAR does have flaws, yes. But so do standard counting stats. I truly believe that WAR is the best we have. Rather than throw it away, I’d rather acknowledge the flaws and adjust for them when discussing the numbers.

      • 21
        Bert says:

        It’s not an issue of WAR vs. counting stats though. For convenience, WAR can be useful, but otherwise, it’s much better to take the best available offensive metrics (like wrc+) and balance that against a consensus of defensive measures, including metrics and observations.

        In addition to the troublesome definition of replacement level, the flaw in defensive metrics, as well as the subjective assignment of position adjustments, WAR also completely ignores the issue of marginal and relative value that are specific to each team. WAR is a catchall, but, like most things that are so broad, it becomes an oversimplification.

  2. 2
    mosc says:

    oWAR should not have positional adjustments. Problems fixed.

    • 4

      Where should the replacement level be factored in? Both? Just oWAR? Simply do away with dWAR?

      • 32
        Doug says:

        I would prefer a version of oWAR and dWAR without the positional adjustment. Then you *CAN* compare Jeff Bagwell and Joe Mauer (or whatever two players) and have some basis for saying Player A fielded his position better than Player B fielded his position, or that Player B provided more overall offensive value than Player A. And, you don’t lose anything when comparing players at the same position. Ergo, a win-win.

        The concept of “value” is fine and there surely is a place for it in sabermetrics. But, since that concept is so poorly appreciated, why not have a “value-less” metric that means what most people are thinking about when they want to use these metrics.

    • 6
      brp says:

      Agreed. I think a baseball fan who can accept advanced metrics can probably figure out all by their lonesome that catcher is more difficult and defensively valuable then 1B or RF, or that you can live with a weak-hitting MIF more than at a corner infield position. Even managers can figure that out…

      oWAR and dWAR are probably more harm than good for now but they can be useful if explained better and the positional adjustment is made its own category, so, you know, oWAR + dWAR + rPOS = WAR… maybe.

      I have no statistics background so maybe I’m an idiot but that passes the common sense test more than oWAR + dWAR =/= WAR that we have now.

      • 12
        John Autin says:

        I am happy that positional value is reflected in both oWAR and dWAR, and I think the only problem with those terms is that we all need to use them carefully. I’m glad for Adam’s reminder, as I have probably been sloppy at some point.

        brp, it may be true that most people know that the positions have inherently different values. But they also know that about different batting measures, different ballparks, different offensive contexts. Why should oWAR aim to consolidate everything else, but ignore the positions?

        Also, I don’t think the degree of difference in the inherent value of different positions is so widely understood. It wasn’t so long ago that first basemen Ryan Howard and Justin Morneau won absurd MVP votes in the same year. And I haven’t seen any awards love for Chase Utley or Ben Zobrist.

        • 19

          I’m going to jump on board with what Dan and others have said here. I find dWAR very useful in terms of expressing a player’s defensive value as a combination of ability and role.

          Do we have to make the distinction between value and ability? I suppose so, but isn’t that true with most WAR components? Roberto Alomar has a negative Rfield, but a better dWAR than Jeff Bagwell. That’s because Alomar was a more valuable fielder, but I think it also means he was a better fielder. It takes more skill to be close to average defensively at second base than to be above average at first.

          As to oWAR, as you allude to above, Adam, the problem is all in the name. A stat called oWAR should exclude the positional adjustment and should combine with dWAR to make WAR. There’s value to the calculation as it is, but it should be called something like “defense-neutral WAR”.

          I think your suggestion that people who don’t trust defensive metrics probably don’t trust positional adjustments or park adjustments either is flawed, if not completely wrong. Denial that ballparks affect offensive statistics is a flat-earth approach that must be all-but-extinct in baseball writing today. If anything, the anti-SABR set has taken to over-adjusting for park effects, writing off Larry Walker as a creation of Coors Field. Dismissal of defensive metrics is an extreme take on the reality that we have a long way to go in measuring defense. I don’t love oWAR, but if it were called dnWAR and maybe displayed somewhere other than the Player Value graphic, I’d support it.

          • 39
            no statistician but says:


            My problem with park adjustment has two components: 1) it affects both the home and the visiting team, but basically the home team players are the ones who get penalized or rewarded in its metric. 2) players aren’t all alike, and therefore some are more affected by, or take advantage of, park components than others—or else they are affected in differing ways—but the adjustment, unless I’m misunderstanding, is a blanket figure, not something customized to the individual.

            Take Larry Walker, 1999. His BA at Coors Field was .461. Visiting teams batted .328, the Rockies batted .325, a figure that incorporates Walker’s .461. No matter how you look at it, Walker was vastly superior at hitting the ball in Coors that year than his opponents or his teammates. It can’t be discounted or belittled into nothingness by ascribing it to park effect. Coors may have helped everyone, but did it really single out Walker as the guy to be helped most, or did Walker through accident or design take advantage of the park components—or did he just simply hit well at home?

            Park adjustments, as your comment seems to suggest, don’t do Walker justice, true. But do they do anyone justice—specifically? Do they explain, or just explain away?

          • 45
            bstar says:


            Park factors are a combination of park effects for both home AND road games, not just home.

            The numbers listed on the player pages for pitchers and the park factors on team pages don’t just reflect the park effect for the home stadium. They are an average of the park effects for ALL games played that year.

            Take Larry Walker and his OPS+. For his on-base percentage, Walker’s OPS in all games is compared to what an average player’s OPS would have been in all games, not just those played at Coors Field. Same with his SLG. The percentages he’s above the average player in both metrics is combined and added to 100 to get to OPS+.

            For pitchers, I’ll use Whitey Ford. We see on his player page that his 1954 personal park factor is 91.1. PPF is the average park factor for the parks Ford pitched in (weighted by batters faced) in 1954.

            We also see on the Yankee 1954 page that NY’s pitching park factor was 93. Again, that’s not just the park effect for Yankee Stadium. It’s the pitching park effect for 72 Yankee Stadium games AND 72 road games (or, rather, 3 years worth of games since B-Ref uses multi-year park factors in WAR calcs).

            Why is Ford’s park factor (91.1) lower than the overall Yankee park factor for pitching (93)? Because in 1954 he pitched 125 innings at Yankee Stadium and 85 on the road.

          • 47
            robbs says:

            Full disclosure don’t know nothin’ about park effects. But do they take into account left or right handed batters? Many examples but for example Cecil Fielder in Tiger Stadium, known as a hitters park, but it was very different for right and left handed hitters. DiMaggio in Yankee Stadium, etc.

          • 48
            bstar says:

            robbs, I think Fangraphs uses batter handedness for park factors, or at least they list them. They even break them down by each type of hit:


  3. 3

    While I totally agree with Adam that dWAR is almost always incorrectly used, my other point was it can be argued that dWAR (if you buy into its validity) is an indicator of who was/is a better defender. Any player who plays shortstop regularly is a better defender than any player who plays first base regularly, for instance.

    I’m sure we could argue that Doug Mientkiewicz is his prime was a better defender than Derek Jeter (who should no longer be a SS) is now, but otherwise, every player on this list was a better defender than Jeff Bagwell (not relative to other players at his position, just better…period): http://www.baseball-reference.com/play-index/share.cgi?id=y5ri7

    • 7
      Brendan Bingham says:

      “Any player who plays shortstop regularly is a better defender than any player who plays first base regularly…”

      I might agree if we restrict the discussion to righthand throwing defenders. A lefthander who has all the defensive skills to play shortstop is still going to be playing either first base or the outfield.

    • 9
      Joseph says:

      Re: “dWAR is an indicator of who was/is a better defender.”

      My understanding is that dWAR is an indicator of who was/is a more valuable defender, not better.

      • 11

        Ding ding ding! I think that’s an important distinction.

        • 16
          birtelcom says:

          Indeed, this is the critical distinction I always try to make when I use dWAR (which is rarely). I think it is fair to describe dWAR as a reasonable attempt to identify which player was the more valuable on the defensive side of the game. Trying to explain oWAR in words is harder because it combines offense with a defensive consideration — positional adjustment — and in general as a result I almost never use oWAR in my writing.

      • 15

        Yes, dWAR is intended to be an indicator of value more than ability, but (and you misquoted me here) I said “… it can be argued that dWAR…is an indicator of who was/is a better defender.”

        My point is the positional adjustments are kind of a stand-in for a rough measure of the base level of defensive ability required to play the position. Besides Brendan’s excellent point in #7 above, you don’t play shortstop in MLB without being a good defender.

        Miguel Tejada (-46 Rfield for his career, mostly at SS) was a better defender than Jeff Bagwell (54 Rfield for his career, almost entirely at 1B). That is at least roughly reflected in their dWAR (Tejada +5.9, Bagwell -7.9).

        • 25
          Joseph says:

          Sorry for taking your statement out of context.

        • 42
          bstar says:

          I agree with Dan. If you’re comparing first basemen vs. other first basemen, you can just use fielding runs. But if you’re comparing a first baseman to a shortstop (which is usually rare except when comparing at a career level), you have to include the positional adjustment.

          If you actually think Bagwell was a better defender than Tejada, consider how Bags might have done as a shortstop and that idea should disappear.

  4. 8
    Mike L says:

    Sorry to break in with yet another Obit, but Frank Jobe, the inventor of Tommy John surgery, just passed away. Interesting story about the procedure–Jobe had used the technique before, but only on polio patients to improve joint-use and functionality. No one had any idea what it would be like in an arm trying to throw a baseball hard.

  5. 10
    wlcmlc says:

    8 comments in and no one mentions the funny typo in one of the chat boxes?

    • 13
      John Autin says:

      And the funny typo was … ?

      I mean, Jayson Stark misspells “indispensable,” but I don’t see the humor.

      And “Glasscock” is all too correct.

      • 14
        wlcmlc says:

        Ahh I thought his name was Glassock. Now I see it actually was spelled wrong the first time and right the second time.

        • 22
          Joseph says:

          Glas Sock? That’s funny. Nobody could ever wear a glass sock. On the other hand, I’ve seen a particular artificial body part made out of glass.

          Yes, I’m fully expecting this comment to get deleted.

  6. 17
    Jeff says:

    I’m not sure where this will take the discussion, but…..

    Why does there need to be a positional adjustment to batting numbers? Nobody goes up the plate as a shortstop or first baseman. They go up to the plate as a hitter. On the flipside, nobody goes to the field as a hitter (even Ted Williams swung his glove in the outfield).

    My point is that each player should be evaluated as a hitter at the plate and as a fielder in the field. How those numbers are mixed to try to come up with an absolute number for each player would then be the big discussion.

    • 20
      Joseph says:

      I think the positional adjustment is because WAR is supposed to indicate value.

      Most people who study the game would agree that a SS who bats .280 with 13 HRs a year and an OBP of .350, is more valuable overall than a 1B with the same stats (assuming all other batting stats are equal). Indeed, on some teams, a 1B with those batting stats might be sitting on the bench, but the SS might make the all star team.

      • 26
        Mike L says:

        Joseph @20, the first baseman you are describing is Wally Joyner, who played 16 years with a 162 game average of 16 HR, .288/.362. Or Chris Chambliss (17 years, 162 game average of 14HR, .279/334

        I think GM’s see a value in players who can play the position competently and stay on the field. It’s one less thing they have to worry about on a year to year basis, and a true upgrade can be very expensive.

      • 27
        paget says:

        The principle behind positional adjustment is unassailable. My question is how the precise figure gets determined. Does anyone know, for example, why a shortstop is given precisely the amount of Rpos he is given? I’ve never been clear on this. Also, I’m unclear as to why this number changes over time. Why does a CF in the 90s get an Rpos bump of 3 runs or so, while a CF in the 50s loses a run?

        A long time ago I raised the (admittedly a bit strange) comparison of Willie McCovey and Lou Whitaker; two players with roughly the same OWAR and PAs. Understanding of course that a 2B’s offensive value needs to be measured differently than a 1B, I had a hard time wrapping my head around them being equally valuable from an offensive perspective given the enormous difference in Rbat. Basically what that boils down to is me doubting that the difference between a 2B and a 1B is -before they step on the field- around 15 runs.

        (Parenthetically, just to be clear about where I’m coming from, this doubt of mine as to the extremes of Rpos has translated into me consistently voting for huge Rbat guys, like McCovey, in the COG and being more reluctant to give votes to many of the glut of 2B candidates.)

        • 30
          Jeff says:

          I’m not sure positional adjustment is unassailable.

          If you want to say that Ozzie Smith is x number of runs (or x number of wins) better than Dick Stuart in the field, I’m sure we can agree and there’s good evidence to prove it. But if you want to say that that has anything at all to do with Ozzie Smith’s value relative to Dick Stuart’s value at the plate, I can’t get there. Their values with a bat in hand don’t change because of the glove they wear in the next half inning.

          Or.. Miguel Cabrera in 2014 can’t possibly be as valuable at the plate as he was in 2013 or 2012 just because he moved across the diamond? He’s the same player at the plate that he has been. He’s just contributing less in the field.

          • 31
            paget says:

            How about this then: the principle behind positional adjustment is unassailable as a component of *total* value. Personally, I’m indifferent as to where that adjustment gets placed: part of OWar, part of DWar, or part of both – in the end it doesn’t really make a difference, since what one truly wants to know (most of the time, I think) is what a player’s total value is. If one is statistically minded, one wants to come up with some kind of number that will allow you to express the fact that it’s more valuable to be a catcher than it is to be LF; my question remains, though, how that number gets determined.

          • 34
            brp says:

            The logic is that he’s less valuable overall, because Cabrera is, let’s say, 50% “better” offensively than the average 1Bman, but was 75% better offensively than the average 3Bman.

            I.e., Cabrera + league average 1B IS MORE LIKELY to generate better offense than Cabrera + league average 3B. That would be the reasoning for putting the positional adjustment in oWAR, but I agree that it makes no sense.

            @12 above: Why should oWAR aim to consolidate everything else, but ignore the positions?
            Semantics argument; Bryan @19 sort of explained this, as does Jeff @17… if you’re calling it offensive WAR, then your position is irrelevant.

            It’s just a measure of positional scarcity; if there were 15 great 2B in the league but only 5 great 1B, then maybe for that year the 1B positional adjustment is less of a penalty than normal. Let it be its own column – oh wait, there it is – rPos. So just leave it out of one or the other since we can already see the value if we want.

        • 36
          Richard Chester says:


          paget: You can get somewhat, but not completely, of an explanation about how Rpos is calculated by going to BR, click om “more”, click on “About”, scroll to WAR Position Players and click it and then scroll down the page to Rpos, Positional Adjustment Runs.

        • 43
          bstar says:

          paget: positional adjustments originally got determined by comparing how the fielding runs of players changed when they went from one position to another.

          So, how do a CF’s fielding runs change when he moves to a corner spot? How about vice versa? How many fielding runs does a third baseman get when he moves to first? Etc, etc. All except for catcher, who rarely switches positions. Catchers are put at the top of the defensive spectrum by definition (who would argue against that?).

          I found this old thread very useful. It’s harder to argue against positional adjustments if you hammer through this. I wouldn’t get too caught up in the final numbers, though. I think this was the first part of a long process.


          Also, average offensive numbers at these positions help fine-tune the positional adjustments from year-to-year.

        • 46
          Michael Sullivan says:

          I don’t know the gory details of the calculation, but the principle behind how rPos is calculated is by looking at all the players at any given position and what their average offensive value is relative to the average of all players.

          So if the average shortstop’s bat is worth 9 less runs in 600 PAs than the average of all batters, then someone will get 9 runs of rPos for 600 PAs as a shortstop.

          As the league changes makeup, this will vary somewhat.

          The principle here is that we are trusting the GMs, managers and coaches who make the lineup decisions, and just measuring them. If they are playing people at 2B who on average are 4 runs worse than the average of all batters, that’s probably because those are the best batters they can find who play an acceptable 2B. Etc. for every position. Presumably if it were easy to find 2Bs who were at least average bats, then more teams who have such, and the positional value would go down. If it became harder, the value goes up.

          I think this is actually an excellent method for determining the value of being able to play a particular position.

          The only position that isn’t based on this empirical measurement is the DH. The reason is that DHs are often players coming back from injuries and other situations that don’t involve excellent bats. The people who did the stats weren’t willing to have the DH position be treated as more valuable than 1B or any other actual field position, so they arbitarily chose a negative number for rPos/PA that would exceed any other position.

          The one place I think you could quibble with rPos is to suggest that the DH penalty should be somewhat smaller or larger than it is. I definitely have to agree with the stat-makers that making the DH more valuable than 1B would be a mistake.

  7. 18
    Joseph says:

    So, spurred on by this discussion, I decided to do a little research on a couple of other positions. First base stood out immediately. The play index shows that there are only 7 players ever who had more than 3000 plate appearances as first basemen (80% of playing time) to have a dWAR greater >= to .2–only 77 all time (regardless of plate appearances) to have a dWAR >=0.

    Someone above brought up Bagwell. He had a -7.9 dWAR. I do not think you can conclude that means he was a bad fielder, and here’s why (at least the way I understand it): Just like every first baseman, he gets assigned an Rposition value of -9 or -10 every single year. In other words, every year he plays first base, he gets about one win deducted because of the lack of position scarcity. Over his career, Bagwell had a -121 Rposition.

    Indeed, I think you could argue that Bagwell was a good fielder for a first baseman, because his Rfield is 54–and only about 17 first baseman ever had an Rfield of at least 54.

    I think that makes sense to me.

    • 23
      Joseph says:

      Following up–if you look at Jeter with an Rposition of +130 and an Rfield of -234 and wonder what he would have done if he had played 1B–consider that his Rposition would have been something like -150, so if he fielded as well as an average replacement first baseman and no better, he would end up with about a -15 dWAR. Even if he had an Rfield of 100, which very few players have done at 1B, he would still have a negative dWAR.

  8. 24
    tunatuna says:

    This is a great discussion and very helpful for novices like me to get a little better understanding of advanced metrics. I have a question.

    It seems to me that WAR has a rival in WAA (wins above average). Is that right?

    If so I feel that WAA is a better way to evaluate the players. Thanks for all you fellas do and keep up the great work.

    Also Adam made me realize just how good of a defensive first baseman Bagwell was. He was outstanding and underrated! IMO.

    • 28

      WAA is WAR minus the replacement level adjustment, so it’s not really a rival statistic.

      At first blush, you might think WAA is a more useful tool, and it might be better at evaluating careers, but the problem you run into is it penalizes marginal players who get playing time versus players not even good enough to get in the game.

      For example, Mike Aviles in 2013 was good enough to play in 124 games and get 394 PA. He was worth 0.6 WAR, but -0.8 WAA.

      His teammate and fellow infielder Jose Ramirez spent most of the season in the minors, but in 14 major league PA was worth 0.1 WAR and 0.0 WAA.

      Now, I don’t know much about Jose Ramirez, and he may turn out to be a better player in the long run than Aviles, but there’s no way he was more valuable than Aviles in just 14 PA. However, WAA says he’s was, which essentially rewards him for not being worthy of major league playing time.

      I’m sure there’s a better example, but I hope that got the point across.

      • 35
        Michael Sullivan says:

        WAA+ is a reasonable answer to that problem — if you believe that value between replacement and average is completely unimportant for evaluation.

        WAA+ would be zero for both those players.

        I think WAA+ is a good thing to look at for candidates with high peaks but shorter careers, to see how much they soared above the rest during the highs of their career and possibly justify a HoF/COG/HoM/etc. selection.

        That said, I definitely don’t support the idea that players who continue to play above replacement level add nothing to their resume if they aren’t at least average. The ability to contribute *at all* in your 40s is significant and worth lauding. 3-4 years of average level play in your 40s is worth a *lot* relative to the zero or even negative value that 98% of players produce at that age.

        WAR also specifically penalizes you for hanging on and hurting your team. Play beyond when you can contribute at least as much as a callup, whether it be to hit certain counting stats, or just because — it puts your team in a bad situation. Nobody wants to bench the future HOFer, or even HoVGers that often are fan and clubhouse favorites. Continuing to play beyond your ability to produce at least replacement level value hurts your team, and I believe should count against you.

      • 40
        tunatuna says:

        Definitely gets the point across Dan, thank you.

    • 29
      Lawrence Azrin says:

      WAA is better for identifying peak candidates. Someone could have a very long career without much of a peak (like Nolan Ryan) and pile up a lot of WAR, but have relatively less WAA.

      • 33
        RJ says:

        Lou Brock is an extreme example of someone who accumulated a lot of WAR (but not much WAA) simply by turning up every day.

    • 37

      My very first HHS article was about the difference between WAR & WAA. Hope this helps! http://www.highheatstats.com/2012/07/wins-above-replacement-war-vs-wins-above-average-waa/

      • 41
        tunatuna says:

        Thank you all for your input as I try to get a handle on this. I am starting to really appreciate the offensive side but don’t know if we can ever get a completely true evaluation for the defensive side. In the meantime I will check out Adam’s suggestion above and I am most definitely interested in peak performance above all else.

  9. 38
    Doug says:

    Some Top 10 Lists.

    Highest career WAR Batting Runs per PA (min. 3000 PA).

    Rk Player Rbat PA From To Age
    1 Babe Ruth 1335 10622 1914 1935 19-40
    2 Barry Bonds 1127 12606 1986 2007 21-42
    3 Ted Williams 1067 9788 1939 1960 20-41
    4 Lou Gehrig 967 9663 1923 1939 20-36
    5 Rogers Hornsby 858 9480 1915 1937 19-41
    6 Mickey Mantle 801 9907 1951 1968 19-36
    7 Jimmie Foxx 757 9676 1925 1945 17-37
    8 Dan Brouthers 682 7676 1879 1904 21-46
    9 Albert Pujols 662 8546 2001 2013 21-33
    10 Shoeless Joe Jackson 438 5693 1908 1920 20-32
    Provided by Baseball-Reference.com: View Play Index Tool Used
    Generated 3/7/2014.

    Highest career WAR Fielding Runs per PA (min. 3000 PA).

    Rk Player Rfield PA From To Age Pos
    1 Mark Belanger 240 6601 1965 1982 21-38 *6H/45
    2 Andruw Jones 236 8664 1996 2012 19-35 *89H7D/3
    3 Jim Piersall 175 6592 1950 1967 20-37 *89H7/65
    4 Jesse Barfield 161 5394 1981 1992 21-32 *9H/8D7
    5 Brian Jordan 153 5646 1992 2006 25-39 *978H/D35
    6 Charlie Bennett 142 4310 1878 1893 23-38 *2/5874963
    7 Rey Sanchez 141 5246 1991 2005 23-37 *64/H5D
    8 Adam Everett 117 3070 2001 2011 24-34 *6/H54D
    9 Clint Barmes 113 3827 2003 2013 24-34 *64/H5839
    10 Lee Tannehill 113 4184 1903 1912 22-31 *56/34
    Provided by Baseball-Reference.com: View Play Index Tool Used
    Generated 3/7/2014.

    Highest career WAR Fielding Runs per PA (min. 5000 PA).

    Rk Player Rfield PA From To Age Pos
    1 Mark Belanger 240 6601 1965 1982 21-38 *6H/45
    2 Andruw Jones 236 8664 1996 2012 19-35 *89H7D/3
    3 Joe Tinker 180 7152 1902 1916 21-35 *6/5H49
    4 Jim Piersall 175 6592 1950 1967 20-37 *89H7/65
    5 Paul Blair 174 6673 1964 1980 20-36 *8H/974D56
    6 Jesse Barfield 161 5394 1981 1992 21-32 *9H/8D7
    7 Brian Jordan 153 5646 1992 2006 25-39 *978H/3D5
    8 Craig Counsell 142 5488 1995 2011 24-40 465H/37
    9 Jack Wilson 141 5339 2001 2012 23-34 *6/H45D
    10 Rey Sanchez 141 5246 1991 2005 23-37 *64/H5D
    Provided by Baseball-Reference.com: View Play Index Tool Used
    Generated 3/7/2014.
    • 50
      BryanM says:

      Doug , great lists and they should put to bed the assumption that people who play shortstop are automatically better fielders than those who play other positions. Not only is there the issue of handedness, raised elsewhere on this thread; Catchers and rightfielders in particular require very different skills than middle infielders . I do share the wariness of some others about the position adjustments,. Although the logic of finding the difference in average batting runs for each position and using it to develop the adjustments seems fine on the surface; the timing of the changes in the adjustment factors seems chunky and arbitrary , sometimes staying the same for several seasons at a time and it doesn’t deal properly , as I understand it with replacement level.

  10. 49
    bells says:

    Adam, this is a great post, and has inspired some good discussion. I’ve always had a problem with the misunderstandings around oWAR and dWAR (even if I didn’t understand them all myself), and this discussion has made me think of other tools of player evaluation such as park factors and WAA in ways that I haven’t considered before. I don’t have much to add to the discussion right now, but wanted to say thanks.

    Unrelated question – whatever happened to the HHS prospect updates for OOTP14? I enjoyed those posts in the fall, but they seem to have just stopped. Do you have any intention or resurrecting them, or are they on the HHS list of things to get to right after the last 8 Mount Rushmores?

    • 51

      Hey bells, thanks. Sadly, I doubt I’ll continue with those. They were fun, but a series of life changes have limited time for such things (as seen by the fact that this is my first post in a good long while).

  11. 52

    […] more defense stuff?! Adam Daworski, known on Twitter as @baseballtwit, offers an interesting take on oWAR and dWAR. I say interesting because he hates both of them, or at least how they’re often […]

Leave a Reply

Your email address will not be published. Required fields are marked *