Exploring Team DefensePosted: November 10, 2010
One of the big banes of Sabermetrics is attempting to quantify team defense. There are nifty tools out there like Ultimate Zone Rating, Defensive Runs Saved, Total Zone (all three of which can be found at Fangraphs) that are the most “advanced” defensive metrics available for Major League Baseball.
However, Colin Wyers of Baseball Prospectus fame has questioned defense with a lot of logical fore-thought. The long and short of it is that the processes of the advanced metrics like UZR or John Dewan’s DRS systems aren’t necessarily the problem; it’s the data being put into it. There’s numerous biases that can enter the mix and that compounds things. Crap in, crap out, so-to-speak. If your data isn’t good, it’s almost going to ensure that there’s nothing of value that will come out of the formula.
Now, since I don’t have any database skills at all, I can’t do a plays made/not made (which is essentially Sean Smith’s Total Zone metric that you find on Baseball Reference) on an individual level like Jeff Sackmann does, but I can use B-Pro’s Def Eff formula and apply it to the Big Ten teams from 2007-2010.
Applying the formula from B-Pro gives me the best way to quantify team-level defense in the Big Ten Conference so we need not rely on Fielding Percentage again. Below is a graph of ten teams in the conference in 2010:
Here, we see Minnesota and Michigan State leading the pack, followed by Illinois and Purdue. The league average DER in 2010 was 0.613 — basically 61.3% of the time a ball was put in play in the Big Ten, it was converted to an out. Below is a chart showing a DER+.
There isn’t that big of a spread between the best and the worst — about 0.05% difference between Minnesota and Ohio State. However, when you factor in the number of balls in play — Big Ten average of 1606 balls in play — the 0.05% becomes substantial.
For instance, Ohio State’s defense had 1510 balls in play last year and their DER of 0.597 means that 901 of those balls were turned into outs. Minnesota had 1718 balls in play with a DER of .646 — meaning that 1110 balls were converted into outs. If we pro-rate Minnesota’s defense to the 1510 balls in play total that Ohio State had, Minnesota’s defense would’ve converted 976 batted balls into outs — 75 more than Ohio State’s defense converted into outs.
Similarly, if we do the same for the Buckeyes and extrapolate their DER to Minnesota’s ball in play total, Ohio State would’ve converted only 1026 into outs. That’s 84 less outs than the Golden Gophers.
It’s a this point that having a run expectancy and play-by-play data available that would make these extra outs into a tangible runs number, but I don’t have those capabilities.
Looking back on the entire 4-year data set that I have, I’ve plotted all of the teams seasons in the graph below.
I tried to break up the colors as best I could. Probably is a big, jumbled mess to everyone, but it’s late as I’m writing this, so it’s the best I can do ‘visually.’ Here’s a table of the same data, if that is easier on the eyes.
The top five defensive teams by DER were Michigan in 2008, Michigan in 2007, Penn State in 2007, Minnesota in 2007 and Minnesota in 2010. I find it interesting that Michigan and Minnesota are in four of the top five spots.
The bottom five defensive teams were Penn State in 2008, Northwestern in 2007, Indiana in 2008, Iowa in 2009, and Purdue in 2007.