Before moving on to this week's team, I need to finish up Ole Miss with this one last post. It's delayed to today thanks to the conference re-draft project, which kicked off on the SBN back channels this weekend.
The theory behind this series of posts is cataloged on the Vandy one, so hit that if you're confused. Basically, I'm running correlations on most major stats with winning percentage to see what is best correlated with SEC teams' winning. The baseline for consistency's sake is +/-0.264 for relevancy, and either the raw values or rank is used (but not both) depending on which is stronger.
For this one I used Houston Nutt's history going back to 2000, the year the NCAA stats archive begins. Nutt has shown in his three years in Oxford that he's largely the same kind of coach he was in Fayetteville, so here go the correlations for 11 years of Nutt head coaching.
Stat | Correlation |
Scoring Defense Rank | -0.789 |
Turnovers Gained Rank | -0.673 |
Passing Efficiency Defense | -0.663 |
Total Defense | -0.539 |
Passing Efficiency Rank | -0.440 |
Net Punting | -0.435 |
Rushing Defense | -0.420 |
Yards per Pass | 0.395 |
Scoring Offense Rank | -0.392 |
Turnover Margin Rank | -0.315 |
Turnovers Lost | -0.311 |
Seven of the 11 stats that hit the threshold for relevancy are related to defense and turnovers. That's somewhat surprising given that Nutt is known as an offensive coach, but not really when you think about it. More on that later. Anyway, I find most interesting among those is that turnovers gained had with a much higher correlation with Nutt's winning than turnover margin or turnovers lost did. I think that's because his teams in this span were, with two exceptions, always within three of 22 turnovers lost on a season. The exceptions were 26 lost in 2007 and 31 lost in 2009, but he won eight and nine games respectively those years.
If you've been following along, you'll also note that rushing defense has fallen in strength of correlation compared with Vanderbilt and Kentucky. That makes perfect sense given that Nutt's teams are not perennial underdogs as the other two are. For VU and UK, rushing defense (in yards per game) is almost like a proxy for how often they're down big, resulting in a strong correlation with winning percentage. With Nutt, that's not the case.
What jumps out at me most in this table is that not a single rushing offensive stat made the list, while two passing stats did. How can that be when rushing is Nutt's bread and butter?
It's an artifact of what correlation is and the fact that correlation and causation aren't the same. Nutt's teams tend to be good running the ball regardless of whether the overall team is good or not. Because rushing offense and yards per carry don't rise and fall with winning, the correlation for them is not very strong. That's the reason why there aren't many offensive stats period compared with defense and special teams, to answer the question from three paragraphs ago: because Nutt tends to have good offenses in about the same range year in and year out, offense doesn't correlate strongly with his winning.
The passing stats do correlate a bit with winning, and that does make sense. Passing success is certainly something that varies over time with Nutt's offenses. When the passing game is good, it gives the team a boost on top of its normally good running game. When it's not good, that boosting effect is lost. In that light, it's exactly what you'd expect from Houston Nutt teams.
Conventional wisdom is not always right, but here it is.