There is a heated argument some people I know need resolved - OBP vs SLG.  It has be belief of many that OBP is more important for a team scoring runs than SLG.  If you take the last 5 years worth of data, SLG (r-squared of 77) correlates better to runs a team scores than OBP (r-qsquared of 73).  Has there been a shift towards power being more important or is OBP still king?  We know that other values like OPS+ and wOBP take both of these measures into account, but we want to answer the debate on SLG vs. OBP.

--Jeff in “Hey, Bill”

November 10, 2008

I took the batting records for all major league teams since the last strike, giving us 13 years of data (1996-2008).   The first thing I did was to test the basic Runs Created formula

(H + W) * TB/ (AB + W) = runs

I ran that for each team and each year, to see whether there was some sort of shift in offense.   For those 13 years, the formula was more accurate in 2008 than in any other season.   It's second-best year (among the 13) was 2007, it's third-best was 2006.

So the basic runs created formula, for some reason, has been working better in the last few seasons than it has in years. . .not that it was ever bad, but. . the standard errors have been going down.   Here’s the 2008 data:

I suspect that the essence of this problem is that the phrase “more important than” is imprecise, and is capable of different interpretations when you put your hands on the data.   It certainly is true that one point of on-base percentage is more significant than one point of slugging percentage, which some of us may sometimes translate loosely as “on base percentage is more important than slugging percentage.”   But on the other hand, the standard deviation of slugging percentage is essentially twice as large as the standard deviation of on base percentage.  The number of “points” is not the same.   You could generalize the data somewhat imprecisely as “one point of on base percentage is twice as important as one point of slugging percentage, but the differences between teams in slugging percentage are twice as large as the differences in on base percentage. . .therefore, the net impact is essentially the same.”   

In recent years the standard deviation of on base percentage on a team level has been very low (.011 over the last three years), which makes the “importance” of the category seem small on a team level.   But whether three to five years of data is enough to draw any conclusion here. . .I’d be skeptical.