More on College Football Fumble Recovery Luck

Last week, I wrote up a piece on turnover luck in 2015. Specifically, I asserted that the percentage of fumbles recovered and the percentage of passes defended that became interceptions were more luck than skill.

I wanted to double check those assertions, though, and first up is fumble recovery percentage. I used a longer timespan than just once season to see if recovery rates still fit a bell curve or if that was just a quirk of 2015. I went back to 2008 with this because that's how far back the CFBStats.com archives go.

I have 985 fumble recovery percentages to work with this time, and they really do appear to happen randomly:

Before going on, I'll note that this time I threw out games that happened between FBS and FCS teams. I didn't do that last week. But just like last week, these percentages are rounded to the nearest whole percentage point.

The average and median are a perfect 50%, which makes sense because every fumble had to be recovered by someone. None are leaking out to the FCS ranks this time.

A recovery percentage of 50% was by far the most common with 65 teams accomplishing that. The next most common was 53% with 49 teams recovering that percentage of fumbles.

The standard deviation of the distribution was 8.6%. The number of teams one standard deviation below average plus those that hit 50% dead on was 330; the number of teams one standard deviation above was 337. The count of teams between one and two standard deviations below average was 146; the count of teams between one and two standard deviations above was 135. The count of teams between two and three standard deviations below was 19; the count of teams between two and three standard deviations above was 13.

The differences between these standard deviations marks represent 0.6%, 1.1%, and 0.7%, respectively, of the total data points. If these recovery rates aren't randomly distributed, I don't know what is.

I will close with a note that there were five teams with rates below three standard deviations below the mean and none above three standard deviations above the mean. If there is any conclusion to draw from that fact—and there may not be—it would be that it's easier to be incredibly bad at recovering fumbles than incredibly good. But it's not like those outliers came about because they were just horrible teams overall. Three were—2008 Wyoming (97th in S&P+) 2012 New Mexico State (122nd), and 2012 Miami-Ohio (97th)—but one was mediocre (2011 Boston College, 71st), and one was quite good (2014 Auburn, 4th).

After all, these recovery rates really are random.