|
When I first published WAR, I saw no need to split replacement level into offensive and defensive components. Replacement level is a concept that looks at the whole player. You can be a replacement level player in a lot of different ways.
On the defensive side, it is possible to be a great fielder and still be below replacement level. It can happen if your offense is terrible enough. An extreme example is Billy Hunter, an outstanding defensive shortstop for the St. Louis Browns in their final season before moving to Baltimore. On Bbref, Hunter was a +18 fielder, but still below replacement level. That was because he could not hit. His batting average was .219, with both OBP and Slugging percentage under .260. There is a limit to how good a hitter can be and still be replacement level. In the world of the DH, that limit is very close to league average. If you are above average as a hitter, no matter how bad your glove is you can still DH and have some value. In the days before the designated hitter, there might have been some cases of a player who was above average as a hitter but so terrible on defense that he was unplayable. Well, at least for regulars. Such a player could have been used as a pinch hitter. This might have been the case for some players who ended up with crazy offensive statistics in 200 game PCL seasons, but couldn't field enough for a major league team to want them. Except for the pitcher, there is no way to hide a terrible bat if you are a great fielder. In order to be useful to a team, every position player has to be able to hit at least a little bit. Thanks to the DH spot, there is a way to hide a terrible glove. You could be a -100 fielder at even the easiest position (like a first baseman who can't catch a thrown ball) and still help a team if you can hit as a DH. Before the DH became universal there are a few instances where an outstanding hitter gave away enough runs with the glove to slip below replacement level. Most impressive was Adam Dunn with the 2009 Nationals. That season Adam hit for a good average, .267, which was his career high. He did the things he was known for, with 38 homeruns and 116 walks. His OPS+ was 144, he was 32 batting runs above average. Yet he was below replacement level because he gave back 43 runs in the field. Am I 100 percent certain he was that bad? I can't say that I am, as the DER_OAA has him at -22 runs, still terrible but keeps him above replacement level overall. But at the same time, I can't say the the Bbref number is wrong. I watched Adam Dunn play that year and he is certainly on the list of the worst defensive players I've ever seen, in the mix with Kevin Reimer. Reimer was an outfielder for the 1990s Texas Rangers who is responsible for the only triple of Sam Horn's career. Dunn had been consistently awful in left field for years, but at least he was mostly allowed to stay in a familiar position. The Nationals in 2009 tried him in left field (-12 runs), right field (-8) and first base (-23). Adam Dunn was no Ben Zobrist. Making him move around to different positions probably made him even worse at each one than if he had been left in one spot. The following season they left him at first and he was just a normal level of terrible (-11 runs). Baseball Reference shows us two component stats of WAR, offensive (oWAR) and defensive (dWAR). Offensive WAR is everything except fielding runs. It answers the question, "what would this player's WAR be if he was an average defender at the position he played?" Defensive WAR is fielding runs plus position adjustment. Since the position adjustment is included in both, oWAR + dWAR will not add up to WAR. This is often a source of confusion to people who don't read the fine print. I have never had much use for splitting WAR between offense and defense and it has been my opinion that it should not be done. I have reconsidered the utility of a defensive replacement level comparison. I think of the case of Derek Jeter. By most measures he has more fielding runs below average than anyone in baseball history. Does this mean that Jeter is the worst fielder ever? No it doesn't. He's being compared to the position that has baseball's best fielders, because it's the only position he played. No matter how many negative fielding runs, Jeter was better with the glove than guys like Adam Dunn, or Dick Stuart. I don't think he was even close to the worst fielding shortstop ever. There may have been many worse, but they just weren't allowed to play 2700 games at the position. They were either told to play less demanding positions (if they could hit) or told to find a different line of work (if they could not). Defensive WAR at least gives Derek the position adjustment. This helps in comparison to someone like Prince Fielder. While Derek has more negative fielding runs, after adding position adjustment Derek's dWAR is -9.4, much better than Prince's -20.5. We are still missing something here though - replacement. By using only fielding runs + position adjustment, who are we comparing Derek and Prince to? An average fielder at a neutral position. Let's say center field or third base. Someone like David Wright (0.3 dWAR) or Adam Jones (-0.6). This brings us back to the whole reason we need a replacement level in the first place. Average players have value. If you can be average and stick around for 15 or 20 years, you have a lot of value. If we are only looking at value above average, then the average player who plays 20 years rates exactly the same as the average player who plays one week. David Wright had a lot of defensive value. How can we best show this? One idea that would allow us to show defensive WAR and offensive and have it add up to WAR is just take the replacement level runs and split that into offense and defense. How much do we split them? Let's go back to definitions. Position players get 59 percent of WAR, and pitchers 41. That means that fielders are getting 9 percent of the total, or 15 percent (9/59) of position player WAR. So if we are adding 20 runs for replacement level, then add +17 to oWAR and +3 to dWAR. That's a logical idea, but it still doesn't solve the first base problem. Keith Hernandez is the greatest fielder at first base I've ever seen, but his defensive runs are offset by the negative position adjustment. Adding three runs per year for replacement level helps, but it just doesn't seem like enough. Do that for Eddie Murray and he's still going to have negative dWAR. If the position adjustment is -9, then adding 3 replacement runs back means that a first basemen needs to be a +6 fielder just to avoid having negative dWAR. Eddie Murray wasn't quite at the level of Hernandez, but he was a very good defensive first baseman who played for a long time. He was a 3 time gold glover, and probably would have won a few more if not for Don Mattingly. What I want to see, for this stat to have usefulness, is something big enough to make it a counting stat. I want a good first baseman like Eddie to generate a positive number, and for that number to be fairly big if the player sticks around for 20 years. I also want it to have some logical underpinning, so that I can describe what I'm measuring. I think I've found it with this concept: DVoDV. Defensive Value over Daniel Vogelbach. I don't mean to pick on Daniel here, I hope if he ever reads this he has a sense of humor about it. I like the guy and root for him when he hits. I could have just called it DVoDH, Defensive value over designated hitter, but it seems to work better because Daniel's initials fit, and also because he looks more like a designated hitter than any player I've ever seen. He's too stocky to imagine him playing in the field anywhere but first base or catcher, and he doesn't have the skill for either position. His skills on the baseball field are confined to the batter's box. He's got a great eye at the plate, and when he connects he can hit the ball a long way. Here's how to calculate it: Fielding runs + position adjustment - whatever your position adjustment would be if you were a full-time DH. Let's try a season of Derek Jeter's, 2012 when he was -18 runs in the field. He played every day so would have been -15 position if used as a DH. DVoDV = -18 + 5 - (-15) = +2 That's not very good for a full time player, but at least he had some value. Let's try an Ozzie Season, 1985: DVoDV = 20 + 10 - (-15) = +45 That's going to be about as high as anyone you'll see. Let's try a great fielder at an easy position, like Keith in 1982: DVoDV = 11 - 9 - (-15) = +17 Not bad at all, I think we might be getting somewhere. This is something that can probably be used as a counting stat, the kind of thing that will keep climbing higher as you keep playing, even if you aren't outstanding. We can construct a career leaderboard where an Eddie Murray does not rank somewhere next to Mo Vaughn. It will be something that can easily be understood and explained. This is the defensive value that Ozzie, Keith, or Eddie has above a player who does nothing at all with the glove. I can apply it back through history, because we now have a familiarity with the DH. It may not have made any sense if this concept had been introduced before 1973, starting with the fact that Daniel Vogelbach had not yet been born. ResultsFirst base
Part of the reason Hernandez ranks below Charlie Grimm here is because the position adjustment for first base was different 100 years ago. This is especially true in the deadball era, but things don't change overnight and Grimm, playing in the 1920s and 1930s probably competed with a better crop of defensive first basemen than Hernandez did. A first baseman needed to field bunts much more often than they do today. Grimm was competing against players who were in the lineup as defensive priorities. The position was less likely to be used as a place to hide an aging star, or an awful defender who was in the lineup to hit homeruns, like Dave Kingman. Eddie Murray rates a little below the top 10, but a very solid +148 runs. He is well ahead of Mo Vaughn (-7). By defensive WAR on baseball reference, Murray is -11.6 wins, and Mo is -12.4. Second base
Shortstop
Even Jeter is positive on this, at +205. Out of 2361 people who have played shortstop, he ranks 75th. Most of that value just comes from showing up. Third base
Robinson ranks so far ahead of everyone else because not only was he the best fielder, he came up early, was exceptionally durable, and played for a long time. Catcher
Don't worry, Johnny Bench is the very next guy on the list. Outfield
This is a combined list, players who played multiple outfield positions like Rickey and Max Carey have all of their outfield runs counted here. Next I'll offer a combined list. Ozzie Smith had a large lead over Rabbit at shortstop, but Ozzie never played anywhere else, Rabbit played a few years at second base.
Here is a CSV file with career totals for all players 1912-2023. There are 8,781 players in my database who played at least an inning in the field, not including pitchers. 7,318 of these have at positive or zero value. Of the negative ones, most are only a few runs below replacement. A little over 100 are -10 runs or worse, and only 16 are worse than -25. The players furthest below replacement level are:
This is a list of players who should have been DHs. Some were DHs at the ends of their careers, or for their time in the American League. Some played a little too early to be placed at their ideal position. Not quite this far down, but still in negative territory, are proto-DHs like Dick "Dr. Strangeglove" Stuart and Babe Herman. Frank Howard was mostly a DH in his final season, the first year of the DH rule. Leon Wagner, Daddy Wags, was a top power hitter for the Angels in their first 3 years of existence. He retired 4 years too early to play DH. Wes Covington was a tremendous hitter, including a .330 average and 24 homers in only 90 games for the 1958 Braves. He was also too early for the DH rule. Covington never had 400 at bats in a season. His lack of defensive skill might have been partly responsible, though he had extreme platoon splits and that seems a more likely cause. Jerry Lynch was another excellent hitter who never played regularly. He was a tremendous pinch hitter who even earned some MVP votes for the pennant winning 1961 Reds. That season he hit .404 as a pinch hitter, with a Ruthian slugging percentage and 25 RBI in only 47 at bats. Lynch was a terrible corner outfielder, he never played even one inning at first base. Luzinski might have been the single player who looked most like a DH before Daniel Vogelbach. The problem was, he came up with the National League Phillies. Once they traded him to the White Sox, he became a full time DH and never played the outfield again. I have no problem believing he cost his team more runs in the field than the DH position adjustment. The thing is, the man could hit, so the Phillies had to live with the terrible glove to get his great offense into the lineup. I'm a little surprised that Burroughs comes out worse than Luzinski. They were contemporaries and very similar players, right handed sluggers with great plate discipline (as long as we are talking about home plate). Both came up in 1970. Luzinski played to 1984, Burroughs retired one year later. They had similar counts of defensive innings. Burroughs rarely DH'd early in his career, no more than 3 times a year with the Rangers. He was traded to the Braves, so for the next 4 years there was no opportunity to DH. In his final seasons he finally settled into the role he was made for as DH for the A's and Blue Jays. My recollection is that Burroughs looked less awful as an outfielder than Luzinski, but the stats say otherwise. Tom Tango, when asked about defensive replacement level, has often said there is no such thing. There is just replacement level. You can be a decent hit, no field replacement level player, or a good glove, no bat type, but either way replacement level is determined by the player's overall quality. Terry Vent, the proprietor of BJOL.org, has posted about the need for a sliding scale, that a replacement level defense can vary. I am not sure if I am explaining his point very well and if not I hope he can correct me. As I understand, the idea is that for an Adam Dunn, replacement level defense is very low. He's such a great hitter that he can get away with more incompetence in the field. For a player like Dewayne Wise, who in 11 seasons had an OPS+ of only 68, replacement level on defense is very high. He needs to be a great fielder to keep getting chances. Honestly, I don't see much of a difference, these explanations seem like two ways to explain the same concept. Some things I fully agree with Terry on are these: 1. The position adjustment has no place in offense WAR. Position adjustment is a defensive thing. It does not in any way help or hurt you put runs on the scoreboard. 2. Defensive WAR on baseball reference does not take replacement level into account at all, unlike offensive WAR. It is misnamed, to be accurate it should be called defensive wins above average. It is possible to separate the two sides of the game in such a way that total WAR = offensive WAR + defensive WAR. The problem is that in doing so you need to split the replacement level part between offense and defense. There is no way to do this and have both still represent a tangible concept. If you give too few replacement level runs to defense, then you have the Eddie Murray problem, where a good fielder still shows up as below replacement, and the longer he plays, the worse his rating gets. If you give too few runs to offense, then you aren't really representing replacement level there either. Let's say we gave 15 runs to defense and only 5 to offense. In that case we'd have many players turn out like Pete Rose, where below average play towards the end of their careers would reduce their value, even though teams are still willing to employ such players. In the end, I've come to accept baseball-reference's decision to not make the two things equal. The biggest problem with their offensive and defensive WAR is they are adding the wrong thing in twice. It would be much better to add replacement level into both components but save the position adjustment for defense only. This is very close to the path I have taken for defensive WAR. If I added 20 runs instead of 15 in for every player, the rankings wouldn't change much. Burroughs would be about -40 runs instead of -80. However, it would lack the logical underpinning. It would produce a positive defensive replacement level for pure DHs - I would have to add 20 for David Ortiz and only knock him 15 for position adjustment. By using the -15 figure I have something that we can observe - a player's defensive value above the guy who does nothing in the field. The DVoDV of Ozzie Smith, Yadier Molina, Chipper Jones, or Fred McGriff can be explained. It is the value they have above and beyond what they would have if they had hit and run the bases exactly the same, but never played the field and spent their careers strictly as a DH. I can see some objections to this, the Kyle Schwarber problem. Kyle in 2023 had a bad enough defensive season that he comes out worse in this measure than he would have been if he had played 100% of his games as a DH. Schwarber was bad in the field, playing 103 games in left field, but even so he did in fact catch 166 fly ball outs. A Pure DH - think David Ortiz in 2016, except for one game and 5 innings at first base - would catch zero fly balls. If we measure from zero, as Bill James seemed to argue for in some of his articles, then maybe Kyle is +150 runs. An average outfielder is +170. A great outfielder is +190. And David Ortiz, the pure DH, is a zero. The problem with that is that such a system would have no choice but to tell you that Ortiz's final season (.315 average, .620 slugging percentage, league lead in RBI) could not possibly be as valuable as Schwarber's season. It couldn't even be close. Ortiz couldn't even match up to a left fielder who was a below average hitter with Schwarber level defense. That just is not reality. Despite doing nothing on defense, Ortiz was a tremendously valuable player who was an integral part of 3 world championship teams. We can't get to that conclusion by measuring from zero, that is why we measure from the league average and use the position adjustments. They may not be perfect, but they at least get us to a conclusion that makes some sense. To be fair, in his articles on measuring from zero, Bill James did put in some kind of floor. You have to. A pitcher gives up 100 runs in a season, how many runs did he save? Bill answered this by setting a floor around twice the league average run level. So if an average pitcher allows 100 runs in 200 innings, he also saved 100 runs. You can't get an answer here without setting an imaginary floor, because there is no real limit to the amount of runs an incompetant pitcher can allow. Take a pitcher who just can't throw a strike. He'll walk everybody, and the inning will not end. I've seen this at amateur level games. The losing team gives up 20 runs, and sooner or later the winning team knows they have to do something to show mercy so they either swing at pitches outside the zone or runners walk off the base to get picked off. That's what the position adjustment for the DH is - an artificial floor that allows us to value players at a level that makes sense. In many cases, a bad fielder who spends part of his time at DH and part in the field will have similar value at both positions. The difference in position value will balance out the negative fielding runs. It doesn't happen 100 percent of the time though, sometimes terrible fielders will cost you even more runs than that estimate.
This page was last modified 05/12/2024 |