How interceptions changed for Florida football in 2018

One of the major findings of the football analytics movement is that turnovers have a lot of luck embedded in them. After that result was firmly established, there has been a lot of work done to try to figure out where the luck ends and what, if anything, can be predicted in the category.

An example of this effort is the demonstration that interceptions happen more often when a team is behind. Chase Stuart showed this fact with NFL data a few years ago, and I was able replicate the result with college football data. The following is from all college football data from 2005-18. The Margin column is the scoring difference in the game relative to the offense, meaning that first four rows are when the team throwing the ball is behind. It only includes passes thrown by FBS offenses.

This should make sense on a couple of levels. First, teams that are bad should both throw more interceptions and fall behind a lot. Second, teams that are behind take more chances and throw more aggressively than those that are ahead.

Using this knowledge, here’s what I found about Florida and interceptions in 2018.

Defense

To begin, the Gator defense actually declined in its interception percentage from 2017 to 2018. In each year they swiped 14 passes. However with two extra games played, the ’18 unit defended more passes than the ’17 unit did. Therefore, the interception rate last fall was 3.7% compared to 5.2% the previous year.

It’s pretty easy to see where the difference was when you look at the interception rates by margin. To keep consistency, the Margin column still refers to the offense. So, when the margin in negative, that means Florida was leading. When positive, the opponent was leading.

I will note that it’s pretty eye-popping that the Gators went from defending eight passes with a 22+ point lead in 2017 to 86 in 2018. That became the largest category last year. However, most of those were against the two FCS opponents, whereas the ’17 team didn’t play an FCS opponent thanks to the Northern Colorado cancelation.

The 2017 defense did most of its air theft in one-score games. They were about on target with the national averages while ahead by a score, and they were a little above expected while tied and way ahead when down a score. That unit can thank Jake Bentley for the lift, as he supplied two of the four interceptions while UF was down a score. CJ Henderson grabbed two of those four as well, one from Bentley and one against Michigan.

Meanwhile, the 2018 defense didn’t snag a single interception while the team was down. The only pick during a tie game came against FCS Idaho, so the interceptions against real competition all came while the Gators were ahead.

Some of this could just be bad luck on the part of last season’s team. The prior year’s squad intercepted more passes, but the 2018 defense broke up 14.7% of the passes it faced versus 13.3% for the 2017 defense. Todd Grantham’s Gators were better at getting their hands on the ball than Randy Shannon’s were; they just couldn’t come up with as many interceptions.

The rest is probably about coaching and scheme. Everyone knows, for instance, that Grantham’s defense was more aggressive. That could account for the difference in pass breakups. I would have to go back and really study the two units to be sure.

That said, I will keep it in the back of my mind to see if Florida’s defense can intercept a pass while the team is trailing in 2019.

Offense

Here I’m going to keep it to just Feleipe Franks since no one else threw a pick in 2018 and I don’t think any of y’all care about Luke Del Rio or Malik Zaire anymore. All of this is moot looking ahead if someone else ends up the starter.

Franks did get better at taking care of the ball in 2018. His interception rate fell from 3.5% in 2017 to 1.9%. He threw two fewer picks despite attempting 93 more passes.

Franks’s issue as a freshman was throwing picks while the team was behind. Here is how he did in 2017, and I’ve eliminated margins of over positive seven because he threw no interceptions (and barely passed at all) when the team was up that much.

His interceptions while down one or two scores all happened against either Texas A&M, Georgia, South Carolina, or FSU. They were never going to beat UGA with everything that was going on then, but the other three hurt badly. The Gators lost to the Aggies and Gamecocks by one score each, while the interception against the Seminoles was a backbreaking pick-six that pushed the game to 24-7 shortly before the half.

Last year, Franks was much better and a lot more in line with the overall numbers I showed at the top.

The rate when down two scores is higher than you’d like to see, but it’s only half a percentage point above the expected rate.

Even so, they were not pretty. One came down 21-10 to Kentucky early in the fourth quarter. Franks has happy feet in the pocket and sails his throw inside past an outside-breaking Josh Hammond. The other was late in the first quarter against Georgia, down 10-0. Hammond gets behind his defender on an intermediate out route, but Franks shortarms the throw right to the waiting Bulldog. Franks overthrew a wide open Van Jefferson on a go route to begin that game and it took him more than a quarter to quit underthrowing passes afterward.

The only margin bracket that is well above the expected rate (by 1.4 percentage points) is when Florida was up by a score. His two interceptions there are very different.

The first came against Colorado State. Dan Mullen had called a three-level pattern to the left side of the field with Moral Stephens going on a short out, Hammond going long up the hash as a near receiver, and Tyrie Cleveland doing an intermediate in route from the outside.

Hammond thinks he breaks free and puts his hand out because the guy covering him cuts off short to cover an intermediate zone. However, Cleveland’s defender out wide notices Hammond and tails him. Franks doesn’t notice the switch, and the defender jumps the pass. That very well could’ve been a pass he only attempts because he knows the Colorado State defense is bad — he also had several ill-advised throws against Idaho that weren’t picked only because it was Idaho — but it also could’ve just been an error in reading the defense that Mullen had yet to coach out of him.

The second came in the red zone against LSU. A defensive lineman beats Fred Johnson’s block and comes free at Franks. Instead of eating the ball and taking the sack, he tries to throw. The ball goes high and far because Franks is being hit while he throws, and Grant Delpit gets an easy pick. If he did pull it down and protect the ball, his interception percentage on throws in that situation would fall to half a percentage point below the expected rate.

Franks certainly did learn to take better care of the ball, but he has room to grow. He seemed flustered in those throws against UK and UGA, and he didn’t make a good read on the Colorado State interception. His one throw while down a score was also a bad read; he tries to hit Cleveland on a short slant but doesn’t see a Mississippi State linebacker hanging out in the same area. That backer pops the ball up in the air for a teammate to haul in.

The ideal world is that Franks will get with interceptions where Tim Tebow was. He, by the way, threw six interceptions as a sophomore with a rate of 1.7%, not that much lower than Franks’s 1.9%. They weren’t that far apart in this one, narrow element of running Mullen’s offense.

That’s not what I mean, though. Only two of Tebow’s 16 career picks came while trailing: one against Auburn in 2007 and one against Alabama in 2009. It’s not that Tebow necessarily was a wizard at protecting the ball while behind; he almost never played from behind because Florida was so good during his time. Most of the picks he threw were bound to be with a lead because he usually had the lead.

If the Gators can get better as a team and play from ahead more often, then Franks’s picks will be more like the Colorado State one where he got greedy rather than the Kentucky or Georgia ones where he’s all out of sorts in part because the team is trailing. That would pave the way for more improvement ahead.