[49] Excellent post Walt.

Teams do indeed treat a 100-110 OPS+ no glove player as basically worthless.

Every team has to play a DH. (Well, every team in the real league.) It's not as if a guy not-DHing means you get an extra defensive player on the field.

But the player shouldn't get credit for that. If the rules allowed a 4th OF who couldn't bat, we wouldn't assume the average guy in that role was worth 2.0 WAR. We don't assume the average 6th RP must be worth 2.0 WAR.

If you are a +5 R offensive player who can only DH, you are replacement level.

Edit: Here's some recent examples

Sammy Sosa 2007: 101 OPS+, DH only, 2008 out of baseball
Vlad Guerrero 2011: 101 OPS+, DH only, 2012 out of baseball
Jermaine Dye 2009: 102 OPS+, -11 RF, 2010 out of baseball
Shawn Green 2007, 103 OPS+, -9 RF, 2008 out of baseball

Walt,

I draw some pretty different conclusions from your examples. The 130+ OPS+ DHs you mention were all considered stars who made a lot of money. I have no idea who in baseball would think those names are just a rung above replacement level. Molitor was considered the best player on the Brewers in the early 90s as a full-time DH, and the city was devastated when he left for a much larger contract than they had just given Yount coming off a MVP season. If you made the claim in 1992 that Molitor was anywhere near replacement or average, I think most GMs would have thought you were crazy.

Pat Burrell was 15 offensive runs above average the year before his Rays contract and was 32. As a full time DH, 15 runs would make him average using bWAR's positional adjustment. His projection for the next 2 years was a slightly below average DH, and he was paid 2/16, the cost of an average player, in what many at the time thought was a bargain. All Pat Burrell shows is that the bWAR positional adjustment agrees with GMs.

Did you consider that so few teams carry full time DHs because they can't easily find players who are above average hitters unless they pay a lot of money, so they instead cycle through other players? If you can't find a hitter of a certain caliber easily and have to pay millions, then by definition that level of hitter is not replacement.

I don't understand the Adam Laroche comment. He is a first baseman who makes $8mm a year. I don't know what that tells you about freely available talent irrespective of defense.

Which makes perfect sense and is consistent with history. If you're utterly useless in the field, you'd better be bringing a 125 OPS+ bat _at a minimum_ if you hope to win a job.

Who are these true talent 120 OPS+ hitters who can't make a roster? You seem to assert that because teams carry in their 25th roster spot a 5th outfielder/pinch runner or an extra garbage reliever and not a 110 or higher OPS+ DH, teams must value those minor role players more than a 110 OPS+ or better DH. I'd think teams don't carry that DH because you have to pay legitimate money to acquire that quality of hitter, which means the replacement level is lower than you think.

Snapper's examples:
A bunch of old guys who had a 100 OPS+ and poor base running who were at replacement level in their last season bWAR. They would project in the 90s OPS+ the next season -- replacement level or below using bWAR positional adjustments. I doubt any of these guys wanted to play for anything like league minimum. If anything, these examples just confirm bWAR has a reasonable DH positional adjustment assumption.

Do regular position players play better if they DH 5 or 10 games a year or some small number? If not, then it makes no sense to keep the position open even if teams are doing it.

Teams do indeed treat a 100-110 OPS+ no glove player as basically worthless.

But if you're a 90 OPS+ no-glove player the Tigers will pay you $6.75 million!

Every AL team has a DH, and having one does not cost the team a tremendous number of runs -- it only costs the team the defensive value of the worst player remaining in the field who might have otherwise DHed.

Great idea. Given that our computers are really fast and our data are really refined, this sounds like a job for Forman/Tango/MGL/Chone. Conceptually this is an easy problem. Most DH's are not substituted for... so you could probably just use starting lineups as a rough approximation.

A player's overall value already depends on his teammates with respect to WAR -- after all, if a good players is stuck behind another player at the same position, then their WAR will be less than it otherwise would be (because WAR rewards value, not talent). Therefore, it's not that strange to have the DH dWAR depend on the players in the lineup that day, i.e. give the DH the worst season positional dWAR of the worst defensive player for each day. Probably a ##### of an SQL statement, but conceptually straightforward given Retrosheet data that is already available.

The fact that you're a regular DH is showing that you are a worse fielder than everyone else starting on your team.

I don't think this is true, even for "regular" DH's. It might mean you're the guy who can best deal with the different role, it might mean your team caters to a prideful star, it might mean you have position-specific skills and those positions are already filled.

I'm actually (shock! horror!) undecided on this issue, I can see both sides. I was thinking about a similar issue this morning in the shower, how would WAR treat situations like these:
-a world where rule changes never followed Eddie Gaedel, and he became the 50's Lenny Harris. A guaranteed baserunner in a highly leveraged situation has quite a lot of value, but he occupies an entire roster spot for maybe 60-100 ABs a season. Unlike DH's, he requires a PR for every AB, so it's not just defense that is missing.
-the complementary case, Herb Washington. He actually has enough games to produce WAR results, but what is going on there?

dWAR is a joke and it's a shame it's used in the total WAR stat when sorting.

Who are these true talent 120 OPS+ hitters who can't make a roster?

There's a certain kind of AAAA player who can't really play a major-league position: Jason Botts was one from the Rangers organization for a long time, or consider Luis Jimenez, who just got called up by Seattle. Botts was about a .960 outfielder in the minors, .983 at first base; Jimenez has fielded .982 at first base in the minors. Fielding percentage becomes telling at a certain level, because it's a minimum competency; suffice it to say that those numbers are below ML norms, below Manny Ramirez in the OF or Jason Giambi at 1B. Were Botts and Jimenez "true" 120 OPS+ ML hitters? Botts was terrible as a ML hitter, and Jimenez is 1-for-17 this September. But these guys had superior seasons now and then in the high minors; both spent some time in Japan; Botts has been playing in Mexico this year. Botts hit .320/.436/.545 in AAA at age 26; Jimenez hit .328/.399/.591 at AA at age 25. Why don't these guys become their ML team's full-time DHs for a season or two? I'd argue that they are 120 OPS+ hitters for awhile at their peak, but another factor emerges: they're not at that peak for very long. Guys who are true-talent 120 OPS+ hitters over an entire career are very rare, peak higher, and are usually good enough athletes to contribute with the glove. Guys who can hit that well briefly in their mid-20s and are slow and lack gloves are not rare. But it doesn't often pay to promote them to full-time DH because, I suspect, Walt's argument is telling: you need to be better than that to offset the lack of flexibility you bring to a roster, and your total disastrous lack of defensive contributions.

the lack of flexibility you bring to a roster

This is a good point. DH is the easiest position on the team to replace from within. Just give a bench player a start at DH. Teams who are missing their CF or 3B have to call someone up from the minors. Teams who lose their DH for a week just give the 4th OF a bunch of at-bats. DH are not replaced by freely available talent on other rosters. They're replaced by players already on the team.

What does the fielding number look like if you use average instead of replacement as the positional contribution?

Botts hit .320/.436/.545 in AAA at age 26; Jimenez hit .328/.399/.591 at AA at age 25. Why don't these guys become their ML team's full-time DHs for a season or two?

Botts hit 320/436/545 in AAA at age 26. That very same year, he hit 240/326/335 in the majors in a third of a season -- the third straight time he had come up and put up a 75ish OPS+. That's why he didn't become their full-time DH.

Jiminez hit 328/399/591 in a partial season at AA at age 25. In his third try. Unfortunately, that same year, at AAA, he hit 148/231/210. Nobody is going to promote a AA veteran to the majors based on 300 ABs when he flopped in AAA. At the end of the year, he was a minor league free agent; he then signed, inexplicably, with an NL team, making it rather difficult for him to become a full-time DH.

Sure, DMN, but what they'll always tell you is "just give that guy a few hundred unpressured plate appearances, and he'll start to produce." And who knows, it may be true. David Ortiz, Edgar Martinez, and Travis Hafner – three of the best-ever pure DHs – didn't get into ML lineups till their mid-20s. Martinez, who knows why, because as noted he could play a little third base; the other two, because they brought nothing but a bat, and the bat had to get safely into the awesome range (as Walt points out) before they could keep a full-time job. And small samples may not produce awesome, in which case it's back to the minors for a while longer.

Guys who can hit pretty well, field not at all, and don't stay at the 120+ OPS level for more than a few years, do not become full-time DHs at the ML level. Their failures are part of that "brief peak" phenomenon: nobody's got a roster spot to devote to the nurturing of a slugging galumph at the major-league level. If they really brought enough value, every team would be incubating little Hee Seop Chois instead of using the DH role more opportunistically. I reckon, anyway; it may be that the empirical situation reflects poor strategy. I hate to use the "they do it this way so it must be right" argument, though as Walt points out, they do it so pervasively this way that it makes you wonder.

I've been thinking this through and I believe I understand how a player can field so atrociously that he provides LESS defensive value than a DH who by definition doesn't field at all. And I'm coming around a little bit on this issue. Let me see if I have this correct.

I started with the view that if a team has to use a DH, it can literally put anyone who can sit on a bench in that role and that person would provide the same DEFENSIVE value as anyone else playing DH. Obviously, offensive value is a completely different story; how well a DH must hit in order to land a job with an AL team is interesting but is a separate issue that is best dealt with in oWAR. But if we are carving defensive value out of the player's contribution and just looking at that, it actually shouldn't matter whether the guy can hit -- because the DH with an OPS+ of 130 is providing the same defensive value as the DH with an OPS+ of 95. The key here is that this is the way it works in theory. In this theoretical view, the guy who makes a few hundred defensive plays per year in LF, even if that is far less than others playing that same position, is providing more defensive value than the full-time DH who makes zero plays per year. I base this on the supposition that it should be easier to find a defensive replacement for the DH than to find a defensive replacement for the terrible-fielding LF, if only because the LF has to run around the outfield for thousands of innings each year.

However, what a DH contributes defensively in theory runs into the defensive ability that we actually observe in MLB. And that means if we look at the players who actually spend significant time at DH, we observe often enough that DHs tend to be interchangeable with bad-fielding LF/1B. This means that in practice, teams do not find it much easier to replace the putrid defensive value of Ortiz than to replace the nauseating defensive value of Manny -- even if it should be easier in theory. And replacement level, even in defense, is set based on what actually happens in baseball, not on theory.

Okay. This all makes sense on a rate basis. Ortiz and Manny may very well have the same range and fielding percentage in LF if both were playing LF. But Ortiz would likely not be able to play LF 150+ games a season because of the physical demands of the job, which is one reason he is a DH. Even playing statue-like LF requires more running around than sitting on the bench. A number of DHs are like Ortiz in that playing in the field for a whole season is just not practical. So here's my question: In factoring in the defensive value of a DH, does dWAR include the likelihood that the "replacement level" DH could not provide an entire season of defense?

For this reason, I still think DHs don't get a big enough ding in dWAR. DHs should get penalized on rate and on playing time. And so I am on board with the idea that every once in a while Manny (or Derek, etc.) could have a season with the glove that's so atrociously abysmal that it is actually below DH level. But because of the rate and playing-time issue, it seems to me that DHs having higher dWAR than position players happens more frequently than should be the case. Thoughts?

Guys who can hit pretty well, field not at all, and don't stay at the 120+ OPS level for more than a few years, do not become full-time DHs at the ML level. Their failures are part of that "brief peak" phenomenon: nobody's got a roster spot to devote to the nurturing of a slugging galumph at the major-league level.

It's people like you who ruined Randy Ruiz's Hall of Fame campaign.