Announcement

Collapse
No announcement yet.

Defensive WAR is ridiculous.

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Defensive WAR is ridiculous.

    This is an example of why I think WAR is a VERY unreliable stat.

    First of all, first base is a position where nobody really cares about defense. let's get that out of the way. Guys are there for their offense. if two first basemen have similar offensive stats, then they had similar seasons. period. Yet, look at this:

    2005 Carlos Delgado 160 OPS+ 4.6 Offensive WAR.
    2007 Albert Pujols 157 OPS+ 5.4 Offensive WAR.

    The two seasons are comparable, if not identical. In fact, Delgado actually had a HIGHER OPS+ than Pujols, so you can make a case that he had the better season. What amazes me is that, somehow, Delgado ends up with 2.5 WAR for the season, and Pujols ends up with 8.5 (!!!)

    That's right, somehow, after defense is added, at a position where nobody cares about defense, Delgado ends up a mediocre player, while Albert ends up the best player in the league.

    Is there enough first base defense in the world to turn a player from an average player to the best in the league? Apparently so.
    Last edited by willshad; 12-17-2012, 11:01 PM.

  • #2
    Originally posted by willshad View Post
    This is an example of why I think WAR is a VERY unreliable stat.

    First of all, first base is a position where nobody really cares about defense. let's get that out of the way. Guys are there for their offense. if two first basemen have similar offensive stats, then they had similar seasons. period. Yet, look at this:

    2005 Carlos Delgado 160 OPS+ 4.6 Offensive WAR.
    2007 Albert Pujols 157 OPS+ 5.4 Offensive WAR.

    The two seasons are comparable, if not identical. In fact, Delgado actually had a HIGHER OPS+ than Pujols, so you can make a case that he had the better season. What amazes me is that, somehow, Delgado ends up with 2.5 WAR for the season, and Pujols ends up with 8.5 (!!!)

    That's right, somehow, after defense is added, at a position where nobody cares about defense, Delgado ends up a mediocre player, while Albert ends up the best player in the league.

    Is there enough first base defense in the world to turn a player from an average player to the best in the league? Apparently so.
    False assumptions. Of course people care about defense at first base. Of course defense matters.


    EDIT:
    You are somewhat right though...defensive metrics are not as precise as offensive metrics.

    One thing you can do is average the current season with the previous 2 seasons and use that defensive number.

    Doing that you'd get:
    Pujols ((2.2+.4+.8) / 3) = 1.1 dWAR instead of 2.2. That'd reduce his total WAR by 1.1 to 7.4 WAR
    Delgado ((-2.9+-.1+-1.5) / 3) = -1.5 dWAR instead of -2.9. That'd increase his total WAR by 1.4 to 3.9

    Another thing to remember is that just because they had similar OPS+ numbers that doesn't mean that they had the same offensive output. Pujols had an extra 63 PA's which also makes a difference.

    I'd recomend using FanGraphs instead of Baseball-Reference. Their WAR breakdown is easier to understand and better presented so you can see exactly where the value is coming from

    Pujols
    Delgado
    Last edited by filihok; 12-17-2012, 11:57 PM.

    Comment


    • #3
      How could you say that???? Next time I see a low or high throw flummoxed, a 3-6-3 turned down, a grounder go past the bag and runs come in I'll realize that they are all imaginary and relax.

      Comment


      • #4
        The fact that 1B is not a key defensive position is already considered in WAR - in the positional adjustments. Pujols lost 1.2 WAR just for playing 1B for that reason. But under the umbrella of 1B, there is a massive difference between a great fielding 1B and a terrible fielding 1B - as we can see from Pujols and Delgado.

        But WAR isn't saying 1B is as important as SS. A great fielding shortstop could have +4 defense plus positional WAR, where as a great fielding 1B could have +1 defensive plus positional WAR. Ozzie's best year was +4.7 and a bunch over 3 and Pujols only has one season over 1.

        And don't tell me that a 30 run saved gap between two 1B is too big. That is only 40 plays, or Pujols making one extra play that Delgado couldn't every fourth game. Seem perfectly reasonable. Watching the two play, I am surprised it isn't more.
        1885 1886 1926 1931 1934 1942 1944 1946 1964 1967 1982 2006 2011

        1887 1888 1928 1930 1943 1968 1985 1987 2004 2013

        1996 2000 2001 2002 2005 2009 2012 2014 2015


        The Top 100 Pitchers In MLB History
        The Top 100 Position Players In MLB History

        Comment


        • #5
          Defensive metrics have taken the route of determining runs saved, which I believe is going down an impossible road. Defense serves to put out players, which indirectly prevents runs. I think these metrics should shift to who is getting the most outs the most efficiently as opposed to saving an indeterminate amount of runs. We don't who is going to be stranded or reach home, so why not focus our efforts on telling what we do know, which is who got whom out?
          "Allen Sutton Sothoron pitched his initials off today."--1920s article

          Comment


          • #6
            Originally posted by Tyrus4189Cobb View Post
            Defensive metrics have taken the route of determining runs saved, which I believe is going down an impossible road. Defense serves to put out players, which indirectly prevents runs. I think these metrics should shift to who is getting the most outs the most efficiently as opposed to saving an indeterminate amount of runs. We don't who is going to be stranded or reach home, so why not focus our efforts on telling what we do know, which is who got whom out?
            Runs saved is just "who got who out" x the run value of those events. UZR and DRS and TZ ALL start with putouts per number of opportunities at the very base.
            1885 1886 1926 1931 1934 1942 1944 1946 1964 1967 1982 2006 2011

            1887 1888 1928 1930 1943 1968 1985 1987 2004 2013

            1996 2000 2001 2002 2005 2009 2012 2014 2015


            The Top 100 Pitchers In MLB History
            The Top 100 Position Players In MLB History

            Comment


            • #7
              Whenever I see 1B put down as "a necessary evil" as far as defensive value is concerned, I simply remind myself of this very simplitic and simple-minded approach to addressing defense and creating a metric to measure it:

              Remove the 1B from you lineup, then play the game with nobody in that function. See what effect this has on the game and the final score. Then, if someone tells you how obtuse you are in presenting such a stupid argument, remind that person that when you remove your first baseman from the game, you also lose the player ... and his bat.

              Historically, first base was critical in dead ball/small ball baseball, defending against the bunt. As that function declined, so did a critical portion of the 1B expectation and challenge. However, lets not toss the baby out with the bathwater:

              1. Take a typical season of outs:

              -25+ outs/game [allowing for DP and incidentals] * 162 games = 4,100 outs;

              -1,000 pitching staff K's;

              -1,100 OF putouts;

              -2,000 outs involving all infield positions;

              -110 [3B]; 250 [SS]; 240 [2B]; 100 [C] = 700 PO at positions other than 1B;

              That leaves 1,300 outs recorded at 1B, +/- 100 of which are inassisted, in that the 1B fields the batted ball and steps on the bag himself. This has the 1B somehow involved in 1,300 of 3,100 outs involving batted balls in play [61.3%]. Add to this involvement the number of throws over [by a pitcher to hold a runner on base] or snap throws [by a catcher as attempted pick-off]; and we have added involvement, which demand the 1B to cede potential range on batted balls for the more "important" function of keeping baserunners honest.

              Any metric that incorporates + or - values to a player for there mere consideration of the position played is, to me, incredible. A metric, to be worth its salt, for me, must incorporate limitations as well as challenges into the fabric of each position expectation. You don't add or subtract before anyone shows up to play.

              If the metric places values +/- on the types of challenges the position player confronts, and exacts - credit for plays not made or made plays blundered, the position metric recipe should reveal defensive value for you, without extraneous gadgetry.

              Comment


              • #8
                Originally posted by leewileyfan View Post
                Whenever I see 1B put down as "a necessary evil" as far as defensive value is concerned, I simply remind myself of this very simplitic and simple-minded approach to addressing defense and creating a metric to measure it:

                Remove the 1B from you lineup, then play the game with nobody in that function. See what effect this has on the game and the final score. Then, if someone tells you how obtuse you are in presenting such a stupid argument, remind that person that when you remove your first baseman from the game, you also lose the player ... and his bat.

                Historically, first base was critical in dead ball/small ball baseball, defending against the bunt. As that function declined, so did a critical portion of the 1B expectation and challenge. However, lets not toss the baby out with the bathwater:

                1. Take a typical season of outs:

                -25+ outs/game [allowing for DP and incidentals] * 162 games = 4,100 outs;

                -1,000 pitching staff K's;

                -1,100 OF putouts;

                -2,000 outs involving all infield positions;

                -110 [3B]; 250 [SS]; 240 [2B]; 100 [C] = 700 PO at positions other than 1B;

                That leaves 1,300 outs recorded at 1B, +/- 100 of which are inassisted, in that the 1B fields the batted ball and steps on the bag himself. This has the 1B somehow involved in 1,300 of 3,100 outs involving batted balls in play [61.3%]. Add to this involvement the number of throws over [by a pitcher to hold a runner on base] or snap throws [by a catcher as attempted pick-off]; and we have added involvement, which demand the 1B to cede potential range on batted balls for the more "important" function of keeping baserunners honest.

                Any metric that incorporates + or - values to a player for there mere consideration of the position played is, to me, incredible. A metric, to be worth its salt, for me, must incorporate limitations as well as challenges into the fabric of each position expectation. You don't add or subtract before anyone shows up to play.

                If the metric places values +/- on the types of challenges the position player confronts, and exacts - credit for plays not made or made plays blundered, the position metric recipe should reveal defensive value for you, without extraneous gadgetry.
                Technicaly, positional scarcity WAR is based on offense per position, not defense, fwiw.
                1885 1886 1926 1931 1934 1942 1944 1946 1964 1967 1982 2006 2011

                1887 1888 1928 1930 1943 1968 1985 1987 2004 2013

                1996 2000 2001 2002 2005 2009 2012 2014 2015


                The Top 100 Pitchers In MLB History
                The Top 100 Position Players In MLB History

                Comment


                • #9
                  Also FWIW, I ran my metric for Delgado [2005] vs. Pujols [2007].

                  I can't/won't relate the comparison in WAR terms [because I have not use for WAR]. However, there is a comparison basis for the comps: League average, offense and defense.

                  2005 Delgado:

                  1B: Innings 1206.0 of 1,442.1 team innings = 83.63% of playing time

                  Rating: the digested adjustment of all 1B defensive inputs [2005] converted to rating uniformity [resembles fielding percentage]: .954 vs. League 1B average[2005] at .947.

                  Defense Runs: seasonal rate: 1.6; adjusted for playing time = +1.33 above average;

                  Batting Runs Created = 121 vs League average @ Delgado's PA [72] = 49 batting runs;

                  Comps for Delgado = 1.33 DR + 49 BR = =50.33 Runs>AVG = 5.03 Wins

                  Pujols, same drill:

                  Defense = Rating [.984] vs. LG [.947] = +8.4 DR>AVG * .924 [Playing Time] = 7.76 DR >AVG;

                  Batting = 132 Runs vs LG AVG [82; run production higher in 2007 than 2005] = +50 BR > AVG;

                  Pujols total: 7.76 + 50 = 57.76 Runs> AVG = 5.8 Wins.

                  Pujols had more PA and more playing time at 1B in 2007 than Delgado; and the RC atmosphere in 2007was higher than in 2005.

                  By my metric's concept of competent utility Player X the DR would jump from 1.33 to 9.5 for Delgado, and from 7.76 to 16 for Pujols, making their totals: Delgado [59.5]; Pujols [66].

                  I imagine in WAR the numbers would be higher for each, with the edge going to Albert.

                  Comment


                  • #10
                    Goodness Pujols crushes Delgado by a nearly 6/1 ratio defensively but the offense is almost a draw so they come out nearly identical. Is the offense really such a high proportion of the overall value? I wouldn't assume 50/50 or anything close but this makes it seem so meager. Is it just that their offense was so great and a crappy hitting slick fielding SS might skew closer to 50/50.

                    Comment


                    • #11
                      PVNick: There are several factors at work here.

                      1. We are comparing 1B for their play in two separate seasons, 2005 and 2007.

                      2. The batting runs climate was different between the two seasons, a run production expectancy of
                      .1161 per PA [2005] and .1211 per PA [2007]. Moreover, Delgado had fewer PA in 2005 than Pujols did in 2007, so it compounds itself.

                      Delgado, at .1964 and 616 PA produces 121 runs. The average NL production is .1161 * Delgado's 619 PA = 72. Yes, it's 49 runs > average in a season with lowe runs scoring and run generation.

                      Pujols, at .1944 [.002 lower than Delgado in a higher batting run creation climate] is at 132 RC vs. LG [82] by 50 runs [and in a higher number of PA].

                      Their seasons are of virtually exact comparable value in the batting area. However, the departure is, indeed at 1B defensively. The disparity between them, with the glove, is notable on a rating basis but is discounted by built-in position risk-reward factors consisting of relative athletic challenges to executing expected types of plays at each position [+] and the relative injury done, by position for errors on plays that others would have turned into outs.

                      This is a distinction with a genuine difference. The position carries NO minus points like an albatross around its neck; but thedemands of the position do have observable limitations, of both risk and reward.

                      In my metric, Delgado at 1B rates .954. The rating intentionally resembles fielding percentage because it is a familiar basis of comparison. The NL LG average at 1B was .947 in 2005, as it was in 2007. At a glance we can see that Delgado was slightly above average. The formulas for 1B in my metric convert that difference into +1.6 DR. I could leave that as an absolute; but it is also desirable to provide context, since Delgado did not play every inning of 2005 for his team. Adjusted for playing time, the 1.6 is diluted to +1.33.

                      By comparison, Pujols rated .984 in 2007, a large gap from Delgado's .954; but internal metric governors at 1B convert that to +8.4 DR, which is diluted to 7.76 DR considering playing time.

                      Such a disparity would have more dramatic impact on +/- DR at other positions, each considered unique. A similarly rated .954 SS in a .947 LG would be at +2.6 DR for full time play and +2.2 DR adjusted. A .984 SS in that same .947 LG would be at +13.84 DR, 12.8 DR adjusted for playing time.

                      I don't use replacement but rather a competent utility player probably 8-9 DR runs below average on a full-time basis. Add 8 or 9 runs to my > AVG rating for Delgado and Pujols, and you see the numbers climb.

                      Defense does have its positional limitations, with very shared and distilled impacts when compared to hitting and its impact of the mechanics of winning.

                      Comment


                      • #12
                        Just a very blurry snapshot of one difference between an excellent and a below-average fielding 1B:

                        As far as righty-lefty, BIP, and GBIP go, Mark Teixera and Prince Fielder pretty much share the same distributions. But Teixeira over his career has averaged one ball fielded per 17.5 defensive PA, Prince Fielder about 1 per 20 PA.

                        Over the course of a season, say 6000 PA, that's an edge for Tex of about 42 balls fielded, assuming there is nothing biasing the distribution in his favor. That's not the same as one player reaching base 42 more times; some of the balls Tex fields are hits or errors, and some of the balls Prince doesn't field are outs. But that's still a huge swing.

                        And I would guess that the difference would be indiscernible: about a quarter of a play a game, a little more than two balls a game vs a little less.

                        Another reason 1B defense seems unimportant is that almost all the plays are routine. Tex averages about 1200 PO, of which 1100 are force outs at first. (Another 200 are unassistedPO or 3-1 assists at first.) Almost all of these are routine. Smead Jolley could jandle them. First base fielding averages are distorted the same way catcher fielding averages are by K PO.

                        So not only is the per game differential small, it's lost in a sea of routine plays. It must be so easy for management to figure something like, "Prince isn't as slick as Tex around the bag, but he's not BAD. he'll miss a few that Tex would get, but we can live with his glove."
                        Last edited by Jackaroo Dave; 12-18-2012, 05:25 PM.
                        Indeed the first step toward finding out is to acknowledge you do not satisfactorily know already; so that no blight can so surely arrest all intellectual growth as the blight of cocksureness.--CS Peirce

                        Comment


                        • #13
                          In defense of first basemen, the presumption of routine plays there discounts greatly the numbers of bad throws hastily made by infielders, or catchers firing away to make a pickoff ... compounded by the absolute fact that the player's range is severly impeded by the higher demand that he assist the pitcher in holding runners on [or close to] 1B.

                          The unassisted PO at 1B is a direct product of that "holding" restriction; and if a difficult smash toward the hole is fielded my an agile 1B, it is "relegated" to being just another 3-1 play.

                          Yes, 1B is a haven for hitters who may not be able to play 2B, SS, C, or 3B specifically ... maybe a bit slow for the OF. However, their in-game involvement in play-by-play events warrants at least a minimum degree of respect. They are certainly given more than an average share of opportunities to screw up in the field and with the high numbers of plays [routine or not] it's surprising how few errors are actuall y committed at first base.

                          Comment


                          • #14
                            Originally posted by willshad View Post
                            This is an example of why I think WAR is a VERY unreliable stat.

                            First of all, first base is a position where nobody really cares about defense. let's get that out of the way. Guys are there for their offense. if two first basemen have similar offensive stats, then they had similar seasons. period. Yet, look at this:

                            2005 Carlos Delgado 160 OPS+ 4.6 Offensive WAR.
                            2007 Albert Pujols 157 OPS+ 5.4 Offensive WAR.

                            The two seasons are comparable, if not identical. In fact, Delgado actually had a HIGHER OPS+ than Pujols, so you can make a case that he had the better season. What amazes me is that, somehow, Delgado ends up with 2.5 WAR for the season, and Pujols ends up with 8.5 (!!!)

                            That's right, somehow, after defense is added, at a position where nobody cares about defense, Delgado ends up a mediocre player, while Albert ends up the best player in the league.

                            Is there enough first base defense in the world to turn a player from an average player to the best in the league? Apparently so.
                            Pujols made 6 fewer errors and had 41 more assists and 178 more putouts.

                            Adjusted for playing time that would still be about 6.5 fewer errors, 36 more assists and 158 more putouts. That would all have to be adjusted for team strikouts, and actual opportunities, but an extra putout at first base is very clearly worth about half a runs. If opportunities were equal it would mean that Pujols basically picked up 158 putouts by doing one of the following: snared a line drive, fielded a pop-up (fair or foul), dug a throw, or fielded a grounder and stepped on the bag. Except for the foul pop-up all of those things basically are equivalent to robbing a base hit, and sometimes robbing an extra base hit down the line.

                            Delgado outhit an average first baseman but gave back some 30 hits a year on defense while Pujols saved maybe 80. Good hitting and poor fielding versus good hitting and great fielding produce much different value. Also consider that in terms of total wins, Delgado was worth about 9 total wins not couting defense, and Pujols about 9 as well. Pujols picks up about 3 for defense and Delagdo loses about 1 so in terms of total value Pujols only produced about a 12 to 8 edge in total wins, but with a "replacement" level player being worth about 4 wins, the WAR edge is about 8 to 4 or therabouts.
                            Last edited by brett; 12-19-2012, 07:10 PM.

                            Comment


                            • #15
                              Here's another thing: Why do we seem to see more outlier defensive seasons that batting seasons when a first baseman is involved in over 1200 fielding plays, versus a batter only around 600? Because WAR and FRAA are on the margin. A 30 FRAA fielding season may only mean that the player made 1320 plays versus 1240 for a 10 FRAA season. A 300% difference in FRAA but only a 6.5% difference in plays made. About the same difference as a guy batting .320 one year and .300 the next.

                              Comment

                              Ad Widget

                              Collapse
                              Working...
                              X