What is Happening on Third Downs this Year?

Normally I would be doing my ACC Check In but this week but since the conference races are basically already locked up I decided to take a little liberty with this week's post.

One of the things I have noticed in doing the weekly Advanced Box Scores is that we have struggled mightily this year at converting third downs. I wanted to do some digging on what might be causing those issues.

Quantifying Third Down Success

The most basic measure of third down success is how many first downs you convert out of all your third down attempts, aka third down conversion percentage. Last year Georgia Tech converted 51.4% of non-penalty third downs last year (I think that's why this number is different than what you may find at cfbstats.com or ESPN). That was good enough for number one in the country. This year Georgia Tech has converted only 28.3% of third downs, good for, and this is almost comical if it wasn't so sad, 122nd in the country.

But not all third downs are created equal. Converting a lot of 3rd-and-2's is way easier than 3rd-and-10's (about twice as easy). So if a team just happens to see a lot of third and shorts verses third and longs their third down conversion percentage will naturally increase. We can quantify this relationship using something called a basis-spline. This B-spline, as it's called, is just a way to fit a smooth non-linear function to data, and it looks like this:

This basically manifests itself as a teams' average distance to go. Although it is not quite a linear trend we can still use average distance to first down as a proxy for right now. How does Tech do in this category? Last year our average distance to a first down was 6.5 yards (good for 25th overall) and this year it is 7.2 yards to go (75th overall). So the change in third down conversion rate isn't only to do with us facing harder third downs.

How drastic is our change though? To go from 1st overall to 122nd in third down conversion seems pretty extreme. I looked at the change from 2014 to the 2015 season in both metrics for all teams. I've highlighted Georgia Tech and Virginia Tech's changes by their respective colors:

Ours is one of, if not the, longest arrow on the chart. But we are certainly following a general trend where an increase in average distance to first makes it harder to convert third downs.

A Better Measure (though still not great)

What we might should be looking at is given a team's average distance to go, how good are they at converting first downs? To do this we can build a model to predict the likelihood of converting a third down based on the distance to go to a first down (like the graph before) and then measure the conversions over expected for each third down a team faces. This should measure the ability of team's 3rd down conversions regardless of the distance they face. However, there are still some problems with this;

• It doesn't adjust for opponent. You'd expect a team that plays an easier slate of defenses to be better at converting third downs, all else equal.
• It doesn't adjust for the quality of a team's offense. A team could be "better" at converting third downs because they are a good offense to begin with.
• I'm not running any statistical tests on the validity of this measure. How does it hold up year to year? How correlated is a team's 3rd down performance in the first half of the season to the second half? I don't know.

So What?

You know, I'm not really sure. No matter which way you look at it we suck at converting third downs and our fall from our incredible performance last year is largest of any team. But *why* are we worse on third downs? I'm not sure that's so easy. Sure we are gaining less yards on third downs this year. Are we running different plays than last year? Are we just not hitting big plays on third down this year? Are we facing better defenses on the whole? I think those may have to wait till the offseason to really answer.

Questions, Comments, Concerns? Put them in the comments.