Originally Posted by RedsManRick
It's cool -- sorry if I came across as combative. I'm not saying it's 50/50. I'm saying that you can't figure out what ratio it is by looking at actual performance of a given player. I guess the easiest way to describe it is that the ratio that matters is between the amount of possible contributions a guy can make, not how much he actually does. The denominator, not the numerator.
I think that the ratio of possible offensive contributions to possible defensive contributions is something like 2:1. That ratio would say that the game is 50% hitting, 33% pitching and 17% defense. That strikes me as reasonable and I believe is in line with what TangoTiger and others have found (and is reflect in WAR). And practically speaking, because players are shifted from position to position to maximize their defensive value, it's functionally more like 3:1. That is to say, the distribution of offensive performance across the range of possible performances is fatter than it is with defense.
Because of they way WAR is calculated, using a set of replacement levels and adjustments, it's hard to see how the two pieces stack up. The denominators, the range of possible contributions, isn't really evident. It's not like you can see that a guy has 60 possible runs for offense and 30 possible runs for defense. But when you net out all the math, including replacement level that's basically how it shakes out. Also, consider that replacement level for offense is a really bad hitter. However, because there are so many good fielders who can't hit, replacement level for defense is average. So you can look at a guy who's a "poor" hitter and he'll still be above replacement. You look at a guy who's a "poor" defender and he'll be below replacement.
Regarding Soriano, a few things:
1.) It's important to remember that UZR isn't a single scale from a WAR perspective, it's a different scale for every player based on their position. If you want to compare one player's UZR to another's, you need to include the positional adjustment that is part of the WAR calculation. Doing that makes Sori's 2007 a 25.1 defender in 2007, still very high, but a bit closer to reasonable expectation.
2.) That 2007 figure is just a massive outlier, driven in large part by Soriano's arm -- 14.3 runs. That was Soriano's 2nd year in the OF and guys were running on him non-stop. He had a bunch of assists. Over time, guys stop running on guys like that and the assists come down. So in a sense, he had access to opportunities to save runs that most other LFs didn't get. There's really no comp on offense -- your opportunities to produce are limited by your PA and there's no way to create more -- it's like being allowed to PH with men on base in a game where you're already in the lineup.
And further, some guys just have years where they perform way above (or below) an established talent level. Trying to assess the reasonableness of UZR by looking at Soriano's 2007 compared to his other years is akin to assessing OPS by looking at Bret Boone's 2001.
Anyways, I'm rambling now. But I think as you dig in to it, you'll find that while UZR does vary a lot more from season to season than offense (imagine if you compared everybody's offensive performances after 200 PA -- you'd have a bunch more guys who are bad hitters putting up great "seasons" and vice versa than you do over 600 PA), the numbers are still entirely reasonable.