The Statistical Reason Why The Packers Defense Has Declined


While doing research on my last article, I noticed one very interesting fact: Dominant 3-4 defenses tended to have a star 5-technique defense end.  The 3 best 3-4 defenses in terms of Advanced NFL Stats’ dEPA (defensive expected points added) in the NFL right now are San Francisco, Arizona and Houston and each team boasts impact 5-technique defensive linemen like Justin Smith, Calais Campbell and JJ Watt, each of which is among the top five 5-technique defensive linemen according to ProFootballFocus.  This got me to thinking: everyone knows that the quarterback effects offensive success more than any other position on the field (hence why Aaron Rodgers and Peyton Manning can keep winning games without good offensive lines and running backs), but is there a position on a 3-4 defense that is most important to defensive success?

Traditionally, the hallmarks of a good 3-4 defense has been it’s nose tackle and outside linebackers; indeed in 2009 when Green Bay switched from a 4-3 to a 3-4 defense, general manager Ted Thompson drafted nose tackle BJ Raji with the 9th overall pick and then traded up back into the 1st round for outside linebacker Clay Matthews III.  The argument has always been made that a dominant nose tackle that can eat up multiple blockers and outside linebackers who are athletic enough to rush the passer are the keys to a dominant 3-4 defense.  You could argue that Green Bay seems have both positions covered, both Clay Matthews III and BJ Raji are both dominant players but while that seemed to have translated to success in 2009 and 2010, it didn’t seem to matter much in 2011 and 2012.

What I’ve done is a correlation analysis using ProFootballFocus’ player grades and comparing them to overall defensive efficiency measured in dEPA.  I’ve flipped the signs for dEPA to just to avoid making it an inverse correlation.  I’ve included both Pearson’s r and chi2, I’m not really much of a statistics guy so I have no idea what the difference is between them, but if you happen to know more about this, leave a comment and I can adjust my analysis if needed.  Overall, the way to read these figures is that a value of 0 means there is no correlation at all while a value of 1 means that there is perfect correlation.  So for this case, the higher the number the more “valuable” that position is to defensive efficiency.  I’ve also included a positive control by correlating dEPA vs. dDVOA (from Football outsiders) and they are 91% correlated, which basically means this analysis holds for both metrics.  Finally, I’ve included a negative control by looking at the correlation between how well the offensive center plays versus how well the defense does; presumably how well the center plays has no relationship to how well the defense plays.