Gross Yards = Gross, the Puntalytics primer you didn't know you were waiting for

The date is December 30th, 2018. (It might ring a bell -- BDN shuts out Washington in the season finale, and that combined with a Vikings loss sends Philly to the playoffs.) Our story starts with 11:31 left in the 3rd quarter, when a Smallwood run comes up short on third down. 4th & 1 and of course, Doug is going for it, but BDN takes the snap and flags fly...Delay of Game. Now it’s 4th & 6 at the Washington 45, and Cam Johnston, waiting attentively on the sideline, knows what that means - punt time.

Cam executed excellently, dropping a punt at the 9 with no return. See for yourself

Cam’s reward for this beauty is a rise in his percentage of punts inside the 20 (I20%), and a drop in his touchback percentage (TB%). However, through the eyes of the common punting statistic -- Gross Yards/Punt -- this punt was bad! In 2018, every punter averaged a Gross Yards/Punt of at least 40, meaning even the worst punter would see their average drop on account of this gorgeous punt. This situation alone should convince you that Gross Yards is just that: gross.

Our first and main goal was to fundamentally reshape the idea of the "average punt." The single largest confounding factor, as hinted above, is the line of scrimmage at which the punt takes place. When deep in their own territory, punters can boot away, occasionally punting as far as 60, 70, 80 yards. When asked to punt in what we like to call "Pin Deep Territory" -- less than 40 yards from the opposing end zone -- punters have no such opportunity. Born from this feeling that punters are "helped" or "hurt" based on where they are placed on the field, we arrive at the silly name of our grand solution: Scrimmage Help/Hurt Adjusted Real Punting, or SHARP.

What is SHARP?

SHARP is a metric to evaluate punts based on a punt’s gross yardage and line of scrimmage. As mentioned above, the first and most important step is to determine what an "average" punt looks like at each line of scrimmage. Consider the following plot, where every dot is one of the over 2,000 punts from the 2018 regular season. The x-axis shows the line of scrimmage, and the y-axis shows the gross yards of the punt. In red, we’ve included the overall average Gross Yards / punt.

You can see that on the left side of the plot, punts above the red line are generally good, and punts below the red line are generally bad. On the right side of the plot, however, this is less true, and at the extreme, all punts are below the red line!! This reflects the conundrum that Cam faced above.

Compare that plot with the one below:

This plot is the same as above, with an important additional feature - a yellow line that tracks the average punt at each line of scrimmage. (What we’ve done here is a little more complicated than simply averaging all the punts at each line of scrimmage, because that would be extremely noisy, but those details aren’t super important. The shaded region around the yellow line is a Confidence Interval, which we have to include so that data scientists don’t get mad at us.)

With this idea of "average" in hand, we can now go about evaluating punts by how much better or worse than average they are. We do this by simply dividing the gross yards of each punt by value of the yellow line at that LOS (and multiplying by 100, because yuck decimals). A punt that falls directly on the yellow line would score a SHARP of 100.

Let’s consider this metric in the case of Cam’s punt above. The average punt from the opposing 45 (that’s the value of the yellow line when Yards From Own End Zone = 55) is 33.6 yards. Cam’s 36-yarder scores:

SHARP = 36 / 33.6 x 100 = 107.1

meaning that this punt is 7 % better than average. In contrast, consider a punt that went the same 36 yards, but started at the punter’s own 30 yard line. We know just from our general football intuition that this is a terrible punt. The yellow line tells us that the average punt from one’s own 30 travels 47.7 yards, and we can calculate:

SHARP = 35 / 47.7 x 100 = 75.5

meaning that this punt is 24.5 % worse than average.

Who’s the SHARPest?

Here are the top 10 punters of 2018 by SHARP:

Note that Gross isn’t actually that predictive of SHARP! Brett Kern and Sam Koch both logged an average gross lower than that of Andy Lee but leapfrogged him in SHARP. This means that, over the course of the season, they were the best able to consistently outperform that yellow line.

There’s one important disclaimer about SHARP. SHARP was designed to fix the single biggest issue with Gross Yards/Punt: line-of-scrimmage bias. However, there are other problems with Gross Yards/Punt, and SHARP inherits all of those problems! Our future work will be aimed at addressing all of these problems.

SHARPnet, and a new generation of metrics

The second-largest issue with Gross Yards/Punt (and therefore, the largest issue with SHARP) is that it doesn’t consider returns. To address this, we can calculate SHARPnet, which is calculated just like SHARP, except using net yards instead of gross. Here are the top punters from 2018 again, still ranked by SHARP, but also showing their SHARPnet. We’ve extended the list from the top 10 to the top 14, so as to include Thomas Morstead and Rigo Sanchez and their impressive SHARPnet scores.

Hopefully, these two new metrics have you excited about the future of Puntalytics, and specifically about some new metrics that we’ll be rolling out as the season gets underway! In the meantime, be sure to shoot us a follow, @ThePuntRunts, to stay up to date about punting news, highlights, and analytics.

As always, a huge, huge thanks to the folks at @NFLscrapR, who've made it so easy for folks like us to access and analyze NFL data. We'd also be remiss not to mention @SeanFromSeabeck, fellow punting enthusiast and now employee for the Ravens, who has helped us along the way.