Personally, I think that it would take beating VT, UGA, and FSU, with FSU and UGA in the top 12 and VT in the top 20.  What sayest thou(s)?  And, since this is a two-part fanpost, after the break you'll find fumble total analysis from CPJ's teams to see if we'll improve, inspired by this comment by lil broey. 

So I made this handy dandy spreadsheet to assist in the telling of this story.  I include only 02-10 because I couldn't find any data from his days at GASU from 97-01.  Now, out of these years, the worst was 02.  This was also the year PJ inherited a miserable Navy team, incomparable to any of his others, so I excluded this year from maximum/minimum consideration.  Thus, our worst years were as far as fumbles AND fumbles lost was...surprise surprise, last year.  however, after such miserable years, PJ usually turns it around.  For example in 03, we put 9 less on the ground and gave up 11 less total!  This is an encouraging sign, but even more encouraging is that after other egregious years (06 and 08) PJ has reeled it in, giving up 13 and 8 less fumbles (respectively) the following years!  This really shows me that PJ's "fall on the ball, play's over, pick up the ball, get a touchdown" drill that he has talked about before really works. We gave up less fumbles than we did the year before 5 out of 8 years, an even further positive for this offense.  All in all, it looks like we'll give up significantly less fumbles (5.8 less, on average) next season!

Miscellanea: the data seems almost periodic, where his teams have horrid fumble numbers one year and great ones the next.  This may be why our recovery numbers every other season are so good.  That is, the numbers are so horrible in a previous season so it's pretty hard to be worse!   We give up about 16 fumbles on average per season, which seems high, but this is a secret devil of the system in my opinion.  "YOY FL Pavg" and "YOY FL Navg" are the averages of year-over-year turnover differential for positive vales of this number and negative values of this number, respectively.