Monday, July 29, 2013

Pace and Injuries: How Dangerous is the HUNH?

Looks like random blue dots in a white square to me.

How long has this been going on?

Hurry Up, No Huddle Offenses did not just appear last year.  In 1997, Gus Malzahn was the head coach at Shiloh Christian in Arkansas when he decided to run his offense at a 2-minute drill pace for the entire game.

"We started running two- and three-play drives at the first of games the year before," Malzahn said. "We'd get great momentum, and then we have to go back to huddling and we'd lose it. So we decided in 1997 that summer to revamp everything and see what happens."
Malzahn was hired as the offensive coordinator at Arkansas for the 2006 season after running his offense at the high school level for nearly a decade.  He then held the same position at Tulsa and Auburn, where he continued to use pace as a weapon against defenses.

Chip Kelly became the offensive coordinator at Oregon in 2007, bringing his fast paced offense from the University of New Hampshire.  Oklahoma, Oklahoma State, West Virginia, Baylor, Texas A&M, Ole Miss, and Clemson (not to mention several non-BCS schools like Houston and Marshall) are examples of teams that have run their offenses at a hurried pace within the past few years.

So *now* we're worried about the HUNH?

Yes, the HUNH is being used at more and more schools, but it is not new to football.  What's new is that a few coaches are becoming increasingly frustrated by it.


Nick Saban first broached the subject in October of 2012.  West Virginia and Baylor had just combined for 1507 yards and 133 points and while Hugh Freeze and Ole Miss were only able to scrape out 14 whole points against Alabama's defense, it was about 50 points more than what Saban deemed acceptable.  So when asked about the proliferation of up-tempo offenses, he said, "I just think there's got to be some sense of fairness in terms of asking is this what we want football to be?"  The only game Saban's Alabama team lost last year was to the fast-paced offense of Texas A&M.

At SEC Media Days, Bret Bielema answered Saban's rhetorical question. "Yeah, we wanted to play a little bit of normal American football."  But he also went "late night local news scare tactic" (do you know what's in your soup?) when he raised concerns over HUNH offenses causing more injuries.
"Not to get on the coattails of some of the other coaches, there is a lot of truth that the way offensive philosophies are driven now, there's times where you can't get a defensive substitution in for 8, 10, 12 play drives.  That has an effect on safety of that student-athlete, especially the bigger defensive linemen, that is really real."
So in one corner, you have Malzahn, Freeze, Dana Holgorsen, Mike Leach, and Mike Gundy in favor of up-tempo offenses.  In the other corner, you have Saban, Bielema, Lane Kiffin, and Will Muschamp, all coaches who suddenly have a concerns that a style of offense (one they don't use and they can't consistently stop) causes more injuries, despite any actual research or study.

Are you asking for a CHALLEEEENGE?

No actual research or study was available when Bielema made his comments, but at least one person took that as a statistician's challenge.  Just a few days after the assertion that fast football causes more injuries than slow football, Dave Bartoo of cfbmatrix.com presented a post that tried to find a correlation between plays per game and offensive starts lost.  What he found was interesting to say the least.
"For all of FBS football in 2012, the ‘fast’ teams averaged over 17 plays per game more than the bottom 20 ‘slow’ teams. This is 26% more plays run per game than a ‘slow’ teams. Even though this adds up to over 340 more plays run in a season, the ‘slow’ teams still lost 8 more starts to injury than the ‘fast teams. Additionally, the average number of starts lost per play was 33% HIGHER for the ‘slow’ teams. Although this is all FBS programs and just the 2012 season, that is a huge argument in favor of ‘fast’ play."
Simply put, fast offenses lost fewer starts than slow offenses.  Actually the exact opposite of what Bielema was arguing.  In fact, the followup post shows that the SEC as a whole ran the fewest plays but suffered the most starts lost.  Perhaps lining five offensive linemen and two tight ends up against 4 defensive linemen and three linebackers to then slam a fullback and a halfback into two safeties isn't the model of perfectly safe football after all.

Pace and starts lost for every team

Bartoo's analysis compared the 20 fastest and 20 slowest teams, but what about all the teams in between?   Thanks to cfbstats.com and philsteele.com, I was able to generate the following graphs.



Counting only runs and passes in the plays per game count, Auburn's offense had the fewest plays per game of any team in 2012.  What a change 2013 should be.  Marshall ran the most plays per game.  A couple teams lost no offensive starters during the season, but LSU lost the most.  Regardless, the scatter plot seems to show no correlation between number of plays per game and number of offensive starts lost.  An r-squared value of 1.0 would mean there is a perfect correlation and an r-squared value of 0.0 means the plot is completely random.  Which number is this r-value closer to?




Maybe a team that runs more often is more at risk?  Maybe a passing-crazed team puts their players in harm's way?  Well, not really.  Both of these plots look mostly the same as the total plays per game version.  The r-value of the runs per game plot is a touch stronger, but in favor of faster play.  The r-value for the passes per game plot is laughably small.  These graphs have convinced me that, at least in 2012, plays per game had no affect on offensive injuries.



What about the pace a defense faces?  That seems to be what Saban and Bielema are most concerned about.  In 2012,  Alabama defended the fewest runs and passes per game, while Houston defended the most.  A few teams lost no starts on defense, but Colorado State's defense suffered the worst of the injuries.  Though the trend line shows an increase in injuries with an increase in plays per game, the r-value is still miniscule.




The best case for pace causing injuries comes from the runs defended per game plot.  Compared to the other graphs, at least you can see a trend.  But again, the r-value is too small for this to matter much.  Also, I can't prove this, but which teams must defend the most runs?  The teams that spend the most time behind.  If your team is losing, your opponent is going to be running the ball over and over to kill the clock.  Could the teams that lost starters early in the season be more likely to lose games and therefore see more runs?  Could teams that are as mismatched talent-wise also be more likely to be injured?

A few disclaimers.  For one thing, I'm no statistician, so I could be making terrible assumptions with this data.  But I do have a bit of common sense and some basic math skills, so at least the plots are based in actual data.  Second, the Phil Steele only counted injuries that resulted in a lost start and if a player was out for multiple games, then one injury counted as multiple lost starts.  It's not the best tabulation for this purpose, but it shouldn't be completely useless either.

When this debate over HUNH offenses arose, there was no data readily available to support either argument except for vague worries and personal anecdotes.  We now have a few pieces of information that, so far, show there is no obvious evidence that increased pace of play causes injuries.  Bartoo's averages and these scatter plots won't end the debate, but in this concussion-aware era, studies will be done in time.  And I'm willing to bet it is not the time between plays but the size of the players that makes football a more dangerous game.

1 comment: