QUOTE (ptatc @ May 7, 2014 -> 11:07 AM)
This article is one of the primary problems I have with "advanced" stats. Both of these players are being paid to produce runs. You can determine the probable value of this all you want but Abreu has scored 22 and driven in 35 for 57 runs minus the HR as you don't want to count the run twice for a total of 45. Rizzo has 19 scored and 16 driven in for 35 minus the 6 HR for a total of 29.
There is always the comment of RBI being useless because it depends on the teammates getting on base for the opportunity. In that case the percentage of runs driven in by opportunity works.
Don't flame me because I know many of you like the +RC stat and all but when it comes down to it, I think it is all about producing runs from these guys and I really don't care what his weighted on base percentage is. It's great to debate and discuss but it's not as relevant.
First of all, I think it's good that you approach it this way and do it without the "this newfangled stuff is terrible, you nerds are ruining everything." It allows the SABR folks to answer the primary criticism of what they advocate, which is that sometimes sabermetrics seem to be at odds with what is observed (I should say that sometimes sabermetrics do a better job at explaining what it is we see).
Obviously, if Abreu finishes with a 110 wRC+ and knocks in 150 RBI, I'm probably going to want to say he was a better run producer than just 10% above average. If he were to do that, though, it would suggest something that these kinds of statistics don't really take into consideration (and for good reason) - it would mean that the hits, walks, and outs that led to 150 RBI happened in a way that was biased towards runs being scored. Generally speaking, the assumption that underlies these statistics is that most of your at-bats are no different than any other at-bats; not on the micro level, but on the macro level. Over time, few people seem especially clutch or especially not clutch. When a guy looks awesome in some aspects of the classic, counting stats and not so great with SABR stats, there might be some reason to believe that this particular player is better or worse in the most pivotal situations than what is statistically typical.
For instance, let's take two pitchers.
Tom Glavine was criticized by many as a first-ballot HOFer, including myself, for his not-amazing FIP over his career (3.95). It would suggest that his accomplishments were more about longevity than ever being truly dominant. His xFIP (same stat, but assuming league average HR per fly ball) was worse at 4.59. First of all, it's obvious that Glavine didn't luck into not giving up homers to that large of an extent for 19 years. He was just better than we assume at suppressing homers. There's more than that, though. I'll let FanGraphs explain some more:
Basically, peripheral stats tend to be better at predicting future performance than counting stats. Furthermore, these "peripherals" are better at saying how well a pitcher truly pitched in a smaller sample. David Purcey threw something like 10-15 innings last year and had a low ERA but he walked a batter per inning and hardly struck anyone out. We know he was just a lucky SOB, especially since his walks weren't clustered in one outlier appearance.
Rarely, you get a guy like Glavine who is an exception.
Then we have Javier Vasquez, who is the opposite sort of exception. His career FIP was 3.91, better than that POS Glavine! His xFIP was even better, 3.75. His ERA? 4.22 (Glavine's was 3.54). An FIP-based WAR suggests that Javy was a good season or two away from matching Glavine for career WAR. His RA9-WAR for his career was 43.3, though, compared to Glavine's 88. That looks a little better, Glavine being twice the pitcher. It turns out that sabermetricians tend to agree that RA9 (which is based simply on the amount of runs surrendered) is much better for long-term evaluation of pitchers than FIP, which is better for evaluating small samples.
We know what was wrong with Javy - his bad outcomes weren't randomly distributed. He liked to cluster all his walks, hits, and homers in the 5th-6th inning. This means he'll give up more runs than the accumulation of walks, hits, homers, strikeouts, etc. would suggest.
FWIW, we see this kind of variability much more often with pitchers, who are more difficult to evaluate in ways beyond measuring runs allowed. Batters are easier to measure. We have a clearer idea of what every batting outcome is worth, run-wise. There is just a bunch of research that repeatedly demonstrates that players performing better in clutch or run-producing scenarios than other players do is a fiction - that is, they do it, but it is not because they are better in those situations. It's just random variation.
