Baseball Stat Geeks: Wins Above Replacement Question

The Game Room
1

A continuing debate among me and my brothers involves WAR–I’m a proponent and they’re pretty dismissive of it. Like all stats, it’s imperfect, and I realize there are different versions and it has evolved. But my read of it is that it’s a very solid indicator of a player’s value. It’s interesting to me how frequently it validates the obvious (Babe Ruth is the all-time WAR leader, duh) or things that are more difficult to put your finger on definitively (Chase Utley truly was as special a player as he seemed to me over a span of 5 or 6 years). Perhaps that’s confirmation bias, but I don’t think so.

In the last round of the debate, my one brother pointed out that Ozzie Smith has a higher lifetime WAR than Joe DiMaggio, and any stat that results in that is a travesty. (He feels, and I agree, that Ozzie was a great defensive player, but overrated.) I pointed out that career WAR needs to be considered against the length of the player’s career, if that’s the version you use for comparison. A player with a career 5 or 6 years longer than another’s is apples to oranges. A 162-game season WAR stat compared to another is apples to apples.

I did a quick proof of this by taking the top 100 WAR players, and dividing their career WAR by the number of seasons played. This is obviously imprecise because it doesn’t account for shortened seasons, games not played due to injury, 154-game seasons versus 162, etc. That said, it did move Joe D considerably up the list (and well in front of Ozzie) and dropped Ozzie by a lot.

So my thought is that the best WAR measure of a player’s innate greatness (for purposes of comparison) is some form of “average WAR per 162 games.” This accounts for highs, lows, etc. Problem is, I can find career WAR lists, peak WAR lists, and a litany of others–everything except what I’m looking for.

So, a couple of questions: Does this seem like the best way to “equalize” WAR for purposes of career comparisons? Are you aware of such a stat and any rankings of said stats? And finally to fellow baseball stat geeks, what’s your take on WAR?

2

Baseball Reference has them pretty close (slight edge to DiMaggio). But it took Ozzie 6 more years to match DiMaggio. Sounds about right to me.

This is a peak vs. career value debate. DiMaggio obviously had a higher peak.

3

The length of a player’s career should always be taken into account when considering cumulative statistics. This is true for all stats, not just advanced stats. For example, Ozzie Smith had more career hits than DiMaggio (2460 vs. 2214), but DiMaggio had more hits per 162 games (207 vs. 155).

You should know there is more than one version of WAR. The version used by baseball-reference.com gives DiMaggio 78.1 career WAR and Smith 76.5.

I don’t know of any reference that lists WAR per game. I think it’s a good idea, although it might be better for pitchers to consider WAR per 9 innings pitched (or per 7 innings, or something like that).

There are things I like about WAR. It makes it easy to compare players at different positions, or across eras. Its biggest drawback is that it compresses a lot of information into a single number. Knowing a player’s WAR tells you something about his total contribution, but little about how he made this contribution. Was he a power hitter? A slap hitter? A great defensive player at an important position like catcher or shortstop? WAR doesn’t tell you. WAR gives you something useful to consider, but it should be combined with other info to give a more nearly complete picture of a player.

4

There isn’t a ‘best’ way because there isn’t just one way to value a player.

If you’re asking about the biggest buildings in the world, there isn’t just one measurement, you can do total height, highest occupied floors, footprint, volume, floor space, etc. It comes down to the story you’re trying to tell.

WAR/Game, WAR/Season, WAR/Career, it’s not that one or the other is ‘best’ it’s that they are telling different stories.

5

I agree that for your purposes, your measurement (or just a straight WAR per-game measure) is the way to go. It’s not tracked anywhere specifically, although I have seen it show up on some graphics or in articles and so forth.

One thing I would point out: there are a bevy of actual rate stats available already for your deployment, so I’m not sure that taking a cumulative value stat and making it into a rate stat isn’t half reinventing the wheel and half uninventing the wheel. Especially when you’re talking about somebody like Ozzie, whose value is derived essentially entirely from being good on defense at an important position, one of the important components that WAR is capturing is just how useful it was that they actually were out there for all those games. It’s a little bit of a roundabout way to look at Ozzie Smith’s contributions to say that in the games he did play, he was worth X% more than a replacement player, when the real essence of Ozzie Smith’s contribution was that he was worth X% more than a replacement player for that many games. If you see what I mean.

So, in other words, if you want to demonstrate why you’d rather have prime DiMaggio than prime Ozzie, because you think (as I do) that prime production is the best measure of “innate greatness,” you can look at lots of things to demonstrate that – wOBA, WRC+, OPS+, even just his regular old triple crown stats show how awesome he was when he was out there. Those are even more specifically calibrated, because they’re per at-bat and have all different kinds of weighting factored in, depending on the stat. WAR, because it is cumulative, is one of the measures that is guaranteed – designed, really – to be more favorable to Ozzie’s career than to DiMaggio’s, though, because it’s specifically not measuring per-anything, and so trying to convert it back to a rate stat somewhat works at cross purposes.

6

WAR/season seems like a good measure. But yeah, DiMaggio career WAR is 78.1 and Ozzie career WAR is 76.5, so your brother can rest easy that DiMaggio is still better.

7

Thanks for the replies, much appreciated. Jimmy, which specific rate stat do you think best captures the comparison I’m trying to make? (Again, I acknowledge that no single stat does it true justice.)

As an aside, not sure which site my brother pulled his WAR rankings from, but that version at least had Ozzie just ahead of Joe D. His smart phone wouldn’t lie to me! :wink:

8

It is perhaps worth noting that because of World War Two, Joe DiMaggio had a much shorter career than one might expect.

Ozzie Smith was a truly great player but he played 2573 games over 19 seasons. Joltin’ Joe played 1736 games over 13 seasons. His having about 75-78 WAR is incredibly more impressive than Smith. WAR tells me DiMaggio was substantially greater.

Ozzie won a lot of Gold Glovs but he also never helped defeat Nazism.

9

Well sure, but he was never given a chance. I bet if he’d been asked to he’d have given it a go, and a few well timed flips might have done the job.

10

That’s exactly the point I made to my brother. Joe averaged, I think, 50 or 60 more hits per season than Ozzie. He just had a much shorter career.

Yeah, but could DiMaggio do a back flip? Let’s not forget about that.

11

Ozzie probably couldn’t defeat the Nazis by himself, but he could at least be credited with an assist.

12

For one number to express exactly what you want to express, I like your number the best. That’s because this comparison is an extreme edge case. Ozzie Smith, by FanGraphs’ calculation, was worth negative 125 runs with his bat and 375 runs with his glove over his career. Joe D was worth 530 with his bat and a handful more with his glove. (If that seems like it’s no comparison between the two, note that 1. FanGraphs does think DiMaggio was more valuable, but 2. a lot of Ozzie’s value comes from the fact that he played shortstop, a position from which it is comparatively much harder to be a contributor on offense, so the R for Replacement in the WAR calculation gives Ozzie credit simply for being an OK hitter at a position where even that is hard to replace).

Anyway, what I’m saying is, you’re comparing a guy whose value was incredibly high for no reason other than historically great defense (well, plus some nifty baserunning) with a guy who had one of the greatest bats of all time. So you need some kind of absurd all-encompassing statistical conversion to even have a language in which to conduct the conversation. WAR is that conversion. THEN, on top of that, you’ve got a guy whose value is derived from absurd longevity who you’re comparing to a guy with a very short career who was absurdly productive during that short career. Again, you need a wacky translation service. So, taking that all into account, if you want to show that Ozzie Smith, while great, wasn’t as innately valuable at a given point in time as Joe DiMaggio, WAR per game or something like that does a fair job, I think.

I’d be inclined to simply point to something like wOBA and say that DiMaggio was 9th all-time, ahead of Barry Bonds, Stan Musial and Tris Speaker, and say that there’s no possible way in hell that anybody has ever been good enough in their prime at playing shortstop to be more valuable than that bat, but I think your brother would be well within his rights to object to that on the grounds that, well, obviously DiMaggio was a better hitter, but that’s not what the argument is about. That would be less fair to Ozzie than WAR, because it’s so one-sided (even though I think it’s one-sided and also true).

To which I’d probably reply that DiMaggio had like 5 seasons out of 12 or whatever where he was worth 7 WAR, and that was with three seasons ripped right out of his prime, and Ozzie Smith never once had a 7 WAR season. But that’s not a single number, and is basically the same idea as your approach to begin with!

13

Thanks, Jimmy. Good logic even in the absence of my “equalized” WAR…