FanPost

Make love not WAR

Reading through the "Jordan Danks nails the first half of his audition" comments, I decided to check and see what we’re really arguing about.

Every statistic in baseball was essentially developed to answer one question, "How good is this guy?" Obviously some of them are going to do a better job of that than others and their utility in terms of measuring talent or longevity has to be considered. But when we discuss the relative value of WAR or OPS or wRC+ what are we really discussing. Generally bigger numbers are better (at least for position players) but beyond that how much more do I need to know in order to judge the relative value of two (or more) players?

In order to answer this question, I ran correlations between commonly used statistics available at www.fangraphs.com for the 166 players with 400 ABs as of Saturday night. (h/t to colintj who ran a correlation between OPS and wOBA in 2009.) I stuck with 2013 stats through about 5:00 pm central time on Saturday. I included the presumed "gold standard" statistic of WAR (actually the offensive portion of it which is wRAA if I read Wikipedia right) as well as the stat with the greatest fall from grace since the advent of the personal computer, RBI.

Here’s what I found:

Correlations

OPS - wOBA

0.996

wRAA - wOBA

0.989

OPS - wRAA

0.986

OPS - wRC+

0.979

wRAA - wRC+

0.976

AVG - wOBA

0.745

AVG - wRAA

0.738

AVG - OPS

0.735

RBI - OPS

0.672

RBI - wRAA

0.666

RBI - wRC+

0.619

So what do we take from this?

  • All of the advanced statistics (i.e., wRAA, wOBA, and wRC+) are essentially the same. This is no surprise in the case of wRAA (and its contribution to WAR) and wOBA since wOBA is the primary variable used to calculate wRAA.
  • Interesting, OPS has a near perfect correlation with wOBA, wRAA, and wRC+. As colintj noted in 2009, "OPS was never designed with [finding the value of a batter’s plate appearances] in mind, but it turns out that wOBA and OPS correlate very very well." The same holds true for the relationship between OPS and other advanced statistics.
  • RBI is indeed a statistical outlier and (if advanced statistics can be trusted, which I think we all presume to be true) not to be trusted as a measure of value.*

Essentially, other than RBI and Batting Average, they all say the same thing. Each of these statistics are aligned to the point that their statistical difference is much smaller than their statistical similarity. If we want to fight about which stat is better, we better have a pretty good reason for selecting one over the other and knowing how that reason is driving a meaningful difference in terms of answering the question, "How good is this guy?"

P.S. I tried this for pitching stats and the correlations are all over the place. Generally over 0.5 but beyond that there’s a lot of room to disagree on the relative value of WAR, ERA, FIP, xFIP, and K/9.

* For those interested, I started this analysis at the team level for simplicity, but the RBI correlation to the advanced statistics didn’t look right. It was much too high. I suspect this is because RBI is largely a function of opportunity, and the standard deviation of team opportunity may be lower than the standard deviation of individual opportunity, which would consolidate the data at the team level.

SouthSideSox is a community-driven site. As such, users are able to express their thoughts and opinions in a FanPost, such as this one, which represents the views of this particular fan, but not necessarily the entire community or SouthSideSox editors.

Recent FanPosts

In This FanPost

Teams