Let’s face it: special teams is like flossing. If you do it well consistently, your gums will be beet red and you’ll steal some points, really put the pressure on your opponent, and more often than not flip the script of the game. If you do it really poorly, your opponent will unwind your game plan and make you look foolish, and your dentist will prick your gums more than he needs to, smugly enjoying your pain. If you remember to floss a couple times a week, its really won’t have any huge impacts on your life. See where I’m going with this? This offseason, I’m going to spend some time breaking down the TCU special teams game, poking and prodding to find some more context and understanding for how TCU’s season went awry.
Many fans perceive special teams as the diminutive third facet of the game, ranked behind offense and defense. Special teams certainly holds less weight that offense and defense, but in practice, it can flip the narrative of a game in the course of seconds.
Take, for instance, two examples from TCU’s early season. First, against SMU, after a long rain delay, the Frogs got off to a less-than-stellar start: a punt, a safety, and a missed field goal in three drives provided the Mustangs a 9-0 lead, and the upset was cooking. TCU needed a spark, and that was provided in the form of a classic [player name redacted] 78 yard punt return touchdown. From that point, down 9-7, TCU went on a 35-3 run, routing SMU and leaving a shaky first quarter in the dust.
Second, against Ohio State, TCU held their own, taking a 14-13 lead into the half. With 6 left in the third quarter, TCU had extended that lead, keeping pace with the Buckeyes to make it a 21-19 game. A sophomoric mistake from Shawn Robinson, understandable against a vaunted Ohio State defense, gave the lead to the Buckeyes, and TCU had the ball with a shot to tie things up. The drive stalled, one of the few times it did that game, and a punt would flip the field on Ohio State, giving TCU’s defense more opportunity to play their advantage. Instead, a botched punt sealed the deal for Ohio State. TCU pulled within 5 points, but couldn’t dig out of the hole the interception and the blocked punt built.
Flossing matters, at the extremes.
- Return Coverage was great.
- Distance was low and variance was high.
- Situational punting doesn’t factor in.
The Horned Frogs ranked 115th in Special teams S&P+, 112th in Net Punting. Punt efficiency ranked 86th in the nation, 55.6%. Overall, although TCU special teams had their moments, the special teams just fell flat.
Today, let’s look at how punting specifically limited the Horned Frogs this season, hamstringing drives, giving points to the other team, and losing control of games.
Between Adam Nunez and Andrew David, TCU averaged 34.82 net yards per punt and a measly 38.3 yards per punt, not accounting for returns. The return game, actually, was the highlight of the punting game: TCU allowed 68 punt return yards all season, 4.48 per, which was good for 21st in the nation. The problem with TCU’s punting game comes from two sources: first, poor punting, and second, poor decisions on when to punt.
First, the punting. TCU punted 65 times last year: two punts were blocked, one resulting in a touchdown drive for Ohio State - the other scoring opportunity was squandered by Baylor’s offensive ineptitude. Of the 63 punts, 9 went for less than 30 yards, including an 18 yard punt in the fourth quarter of the Texas Tech game that handed the Raiders a victory and a neat 11 yard punt against Kansas State that helped keep that game closer. The variance of TCU punts was 82.923. Now, it’s hard to say what variance means in a vacuum, but in the context of punts, higher variance implies a more erratic spread of punts, a more widely (and uncertainly) distributed outcome. Looking at the distribution, ideally, of course, the mean should be higher (top 50 teams averaged more than 37 yards of net punting - almost identical to TCU’s unadjusted distance), but also you want the spread to be more tight; that is, you want your punter to give you more reliable outcomes more consistently. If your punter is going to average 35 yards a punt, ideally, you’d want most of his punts to be around 35 yards, so you can game-plan your coverage and make decisions based on that, instead of the wide uncertainty that TCU has. The less like a random draw and the more like a facto of life punting is, the better for your defense and the better for your coverage.
Now, of course, sometimes punts are bounded by field position and situations, which brings us to the punting decision. I created a variable - “deduced yard line” - to measure the situation. Here’s a scatter plot of punt distance vs. DYL to see if situational punting explains TCU’s low yardage or if it’s something else.
If an aberration were to arise in TCU’s punting distance due to situation, we would see a mass of punts on the right side of the graph, mostly towards the bottom half. Specifically, the lower right quadrant of the graph. While there are a few punts there, situational constraints do not appear to bind for the most part. In fact, 4 of TCU’s 5 worst punts came on their side of the field, specifically within their own 40 yard line. The size of the lower left quadrant of the graph, or rather, the number of dots therein, give us a pretty good measure of punting performance. Those punts all take place on TCU’s side of the field, and are therefore unconstrained as to distance. 21 of TCU’s 63 punts (33%) land in that quadrant, and 21 of the 44 (47.7%) punts on the Frog end of the field. Clearly, situational punting wasn’t the issue with TCU’s punting performance, but more so, it was the low total yardage and high variance which made punting an unreliable weapon for TCU’s 2018 game plan.
How Much Punting Affected Outcomes
Let’s go further than surface level diagnostics. How much did the punting game factor into TCU’s winning ability?
I’m going to run two regressions to attempt to capture the explanatory power of TCU’s punting game, with a couple of specifications each. I’ll keep them simple.
- Linear Regression of Point Margin on Punting. (margin = b_o + b_1*punting + other factors).
- Probit Regression of Win Probability on Punting. (margin = f(b_1*punting + other factors).
The other factors I’ll control for I won’t report on, but they will be standard controls for team quality, location of the game, and whether the team won last week or not.
- One additional punt per game is associated with a 1.059 point margin decrease for TCU this season, all else constant.
- 10 additional yards increase in mean punt per game is associated with a 12.3 point margin increase for TCU this season, all else constant.
- One additional punt per game is associated with a .95 percentage point decrease in win probability for TCU this season, all else constant.
- 10 additional yards increase in mean punt per game is associated with a 5 percentage point increase in win probability for TCU this season, all else constant.
Professional disclaimer: none of these are very solid estimates. I wouldn’t trust these numbers further than I can throw them, as it were. Instead, these estimates frame our understanding of the impact on punting for TCU this year, without accounting for the subjective context of each individual punt.
What do we learn? Well, punting performance on its own, in a vacuum, wasn’t exactly the main problem for TCU this year, but it was a significant. The additional yardage in mean punts would drastically alter the structure of a game, which is why you see the 12 point margin swing. Without even parsing out the context, we can see that TCU’s punting played a foundational roll in determining outcomes: TCU lost three games by <20 points, meaning a 12 point swing could’ve radically altered those games, and thus the season.
Next week, I’ll dive into kickoffs and returns and attempt to explain just a little bit more of the mystery and disappointment of the TCU 2018 campaign.