Saturday, January 22, 2011

Applying the "Candy Maldonado Rule" to dWAR

OK... So first of all: What's the "Candy Maldonado Rule?"

That's just my name for the contradictory philosophy of creating a formula to determine objectively who the best ball players are, but then only accepting the formula if it tells you pretty much what you thought you already knew, subjectively.

In a story I read years ago by Bill James, he described how that was kind of how OPS first came about. Someone came up with the idea of adding OBP and SLG together (which is absurd, mathematically speaking, as each is a fraction with a different denominator, but I guess that's beside the point) and when they looked at who the top 10 were (at the time) they got: Babe Ruth, Ted Williams, Lou Gehrig, Jimmie Foxx, Hank Greenberg, Rogers Hornsby, Mickey Manlte, Joe DiMaggio, Stan Musial and Johnny Mize. IOW: Guys who could reasonably end up on ANYONE'S short-list of greatest hitters. And THAT was how they decided the formula was useful: It told them what they already "knew!" So I wondered what would have happened if instead, it ended up telling them that Candy Maldonado was the greatest hitter that ever lived. Well... for one, we probably wouldn't be talking about OPS anymore! LOL

And really... that's kind of stupid, when you think about it. We come up with a formula meant to TEST our judgement, and then we only keep it if it CONFIRMS what we've already prejudged! (And BTW, no offense is intended to Mister Maldonado. He was a fine ball-player, and just happened to be the first name (of a decidedly non-Hall of Famer) that popped into my head, way back during the time I was having this discussion, and it just kind of stuck after that. Really, what the rule means is just to ask the question: DOES THE RESULT MAKE SENSE?

And given some of the rather odd dWAR results I've come accross doing these teams - guys I remember as being anywhere from very good to phenomenal fielders getting little or no (or even negative!) dWAR, in some cases having enough oWAR to make the team, but not enough dWAR - I figured it might be interesting to apply the old "Candy Maldonado rule" to dWAR. On thge high end, all the top 10 players, all-time, by dWAR are: Brooks Robinson (3B), Andruw Jones (CF), Roberto Clemente (RF), Ozzie Smith (SS), Mark Belanger (SS), Carl Yastzremski (LF-1B), Germany Smith (2B), Willie Mays (CF) and Cal Ripken (SS-3B). The highest ranked Catcher, at 12th all-time, was Ivan Rodriguez, and the highest ranked true first baseman, at 36th, was Keith Hernandez.

OK... that might not be perfect, but I'd say it's pretty darned good. It's clear to see that it definitely recognizes greatness when it sees it. Now... how about mediocrity? Let's take a look at the career dWAR of the "Iron Glove" team that I conjured off of the top of my head in my last post about the futility of doing statistical analysis on defense.

C - Ernie "Schnozz" Lombardi (-2.7)
1B - Dick "Dr. Strangeglove" Stuart (-6.1)
2B - Jose "Errors leader of the 1990's" Offerman (-7.4)
3B - Harmon "Killer" Killebrew (-7.5)
SS - Don "E7" Buddin (-4.2)
LF - Dave "Kong" Kingman (-6.2)
RF - Kirk Gibson (0.1)

For the most part? This seems pretty accurate. I'll admit that I didn't think Kirk Gibson was really all that bad... I just couldn't think of ANYONE nearly as bad in the OF as Kong, and I remembered some of the troubles Gibson had early in his career and figured, along with the injuries he had later in his career, that he'd be a reasonable choice. Turns out he rates better than I though, but then... he doesn't really belong on this team anyway, I just couldn't think of anyone else. Also, there's something worth mentioning about Lombardi: He had a CANNON of an arm. His defensive woes come entirely from his having these short, stumpy legs and being so slow, even for a Catcher, that he often turned Sacrifice Bunt attempts into base hits, if the hitter knew to hustle. In any case though, overall I'd say this passes the test on the other end of the spectrum as well.

So why the goofy results with some players? Bob Boone, for example, failing to make the Phillies due to a mere 0.5 dWAR after 10 years on the team? Well... as I pointed out in my last post, there will always be exactly 27 outs in a nine inning game. No matter how bad a team's fielders are, they'll never get less, and no matter how good they are they'll never get more. So as near as I can figure, in the case of Boone, being on the same team with Mike Schmidt, Larry Bowa, Gary Maddox and Steve Carlton, after a point all these great glove-men (along with Carlton's 200-300+ strikeouts a year) start to cancel each other out. So Boone (along with Schmidt and all the others) ends up getting a bit under-rated. And as further evidence of that? In just SEVEN years with California (with aging and over-the-hill guys like Carew, Grich, DeCinces and Jackson in the field) Boone's dWAR was 7.7. IOW: He accomplished over twice as much every year, defensively in Cali as he did in an entire decade in Philly! Unless you think Catchers get better with age (ha!) I'd say it's pretty clear that he was 'cancelled out' a lot less, later in his career.

So... yeah, while dWAR does recognize some extreme cases, it is not perfect. I'd say it seems to suffer from the same deficiency as every other defensive rate stat: Failure to recognize that no matter how good (or bad) a defence is, they can't manage more (or less) than 27 outs per 9-Inning game. I'm still going to use it, but with many of the borderline cases, there may be a case for making some changes in the future. (And I'll certainly make changes if they refine the formula to correct for what I'm talking about here.)

No comments:

Post a Comment