Blog Post

APAYP: A New Advanced Punting Stat

Patrick Barlow • Oct 04, 2023

The problem with traditional punting stats

Punting is a unique aspect of football. In a perfect game, it would never happen. A punter ideally shouldn't get any playing time. Punters most often get noticed because of a mistake. When was the last time you heard about a game-winning punt? Can you recall a time a punter was named MVP of a game? And yet, a great punter is a game-changer. The ability to flip the field is an underappreciated aspect of the sport. Perhaps this is why there is a distinct lack of advanced punting stats. While advanced stats may not help how punting is perceived, it certainly can't hurt.
You may ask, "Why are advanced stats needed for punting?" Generally, only 4 punting stats are ever used: Yards per Punt, Net Yards per Punt, Punts inside the 20, and Touchbacks. So what's wrong with these stats? Let me paint you a picture, or rather, a table.
Punts Yards per Punt Net Yards per Punt In 20 Touchbacks
Punter A 19 47.2 44.1 9 2
Punter B 28 49.0 42.2 10 3
Punter C 27 46.2 43.7 15 2
Punter D 23 44.7 41.1 5 1
Which of these punters would you consider to be the best? Punter B has the best Yards per Punt. Punter A has the highest Net Yards per Punt. Punter C has the most inside the 20. Punter D has the fewest touchbacks. They all have valid arguments, but each argument has at least one major flaw.
  • Yards per Punt benefits punters with bad offenses. If your offense can't get past your own 30, you have more opportunities to send punts as far as you can.
  • Net Yards per Punt punishes bad coverage teams. If you put the punt where it needs to be, but your coverage team whiffs their tackles or assignments, why should you be punished?
  • Inside the 20 benefits punters with mediocre offenses. If your offense often stalls around midfield, you have more opportunities to pin your opponent deep.
  • Touchbacks don't say much by themselves. Maybe your offense gets you to midfield, but you routinely kick too far. Maybe your offense can't move the ball, so you don't have to worry about touchbacks.
Each of these statistics tells a small part of how effective a punter is, but even collectively they don't quite paint the whole picture. That's where advanced stats come into play.

The solution

To accurately gauge the effectiveness of a punter, you need more than individual stats or aggregated information. You need to look at the data for every punt. The first stat to look at is Available Yards per Punt.
Available Yards per Punt is simple. If the ball is at the 50-yard line, a 50-yard punt will result in a touchback. This means you have 49 yards available to you. Similarly, if the ball is on your own 1, a 99-yard punt will result in a touchback, so you have 98 yards available. The next step is to look at how many yards were actually punted. As I mentioned earlier, using raw yardage benefits punters with poor offenses. So instead we need to look at this as a percentage.
Percentage of Available Yards Punted is exactly what it says. It's used to compare punts regardless of position on the field. Continuing our examples from above, if the ball is at the 50, and a punter punts it to the 5, they punted 45 out of 49 available yards or ~91.8%. If the ball is on your own 1, and you kick it to the opposite 35, you punted 64 out of 98 available yards or ~65.3%. Is the first punt really that much better than the second, given the situation? If the second punter wanted to punt 91.8% of available yards (like the first punter did), they would need to kick a 90-yard punt, which is absurdly long. These 2 scenarios, therefore, have to be treated differently.
Adjusted Percentage of Available Yards Punted, or APAYP, resolves this inequality. If a punter punting from their own 1 can't be measured against the opposing goal line, what can they be measured on? The only answer is other punters. APAYP uses the average of the 10 longest non-touchback punts of the season as a baseline. Through week 5 of this season, that equals 69.9 yards. For obvious reasons, a punter punting from the 50 can't be measured against that same distance. Instead, every punt is measured against whichever distance is shorter: yards to the goal line, or the 10 longest punt average.
Now that we've established this baseline, we can accurately compare the 2 punts from our example. A 45-yard punt from the 50 is 91.84% of available yards (45/49). A 64-yard punt from the 1 is 91.56% of available yards (64/69.9). So while the 45-yarder is marginally better, when taking the spot on the field into consideration, we see that they are nearly identical.
I've rated each punt throughout the season so far using this baseline. All scores for a player are averaged together to give that player their APAYP rating. Going back to our table from above, here's how each player fared, with their national rank in parentheses.
APAYP (rank) Punts Yards per Punt Net Yards per Punt In 20 Touchbacks
Punter A 71.49% (22) 19 47.2 44.1 9 2
Punter B 75.18% (8) 28 49.0 42.2 10 3
Punter C 76.53% (4) 27 46.2 43.7 15 2
Punter D 77.7% (2) 23 44.7 41.1 5 1
Is that how you would have rated each of these punters? Using APAYP, we can see not only how good a punter is regardless of position on the field, but we can also quantify how much better they are than their competition and an average punter. I'll be tracking this every week going forward, and you can find updated ratings and standings here. While this stat may not change how punters are perceived, maybe it should.

Follow us today!

By Patrick Barlow 14 Nov, 2022
Each week during the college football season, we'll come here to react to the new TRAM Top 25. TRAM, like most models and systems, gets better and more accurate as the season progresses.
By Patrick Barlow 07 Nov, 2022
Each week during the college football season, we'll come here to react to the new TRAM Top 25. TRAM, like most models and systems, gets better and more accurate as the season progresses.
By Patrick Barlow 31 Oct, 2022
Each week during the college football season, we'll come here to react to the new TRAM Top 25. TRAM, like most models and systems, gets better and more accurate as the season progresses.
By Patrick Barlow 28 Oct, 2022
Predictions are part of what makes college football so fun as a fan. The problem is this sport is wildly unpredictable outside of a handful of teams. If a single game is so hard to predict, how can you predict an entire season? Here's a secret: you don't; you let your computer do it for you. That's exactly what I did. Using TRAM as a base, I simulated the remainder of the 2022 regular season of college football, including conference championship games. Each game was simulated several hundred times to give us the most accurate information. You can find the final standings based on this simulation by pressing this link . This simulation was originally run on 10/26/2022.
By Patrick Barlow 24 Oct, 2022
Each week during the college football season, we'll come here to react to the new TRAM Top 25. TRAM, like most models and systems, gets better and more accurate as the season progresses.
By Patrick Barlow 17 Oct, 2022
Each week during the college football season, we'll come here to react to the new TRAM Top 25. TRAM, like most models and systems, gets better and more accurate as the season progresses.
By Patrick Barlow 10 Oct, 2022
Each week during the college football season, we'll come here to react to the new TRAM Top 25. TRAM, like most models and systems, gets better and more accurate as the season progresses.
By Patrick Barlow 08 Oct, 2022
In part 1 of this series , I looked at which teams were upset the most. In the end, the Oregon Ducks were crowned the Most Overrated Team. Now it's time to look at everyone's favorite teams: the underdogs that pulled it off. Once again, I'll show you the data in 2 ways. First, I'll show which teams have pulled off the most upsets in total. Second, I'll show which teams have the best Upset %, which is the % of times they pulled off the upset when in a position to do so. I'll also filter out teams that have been underdogs fewer than 20 times for 1998+ data and 6 times for 2014+ data. In part 1, I defined an upset as a game in which: an unranked opponent beats a ranked opponent, or a ranked opponent beats an opponent ranked more than 2 spots higher than them (for example, #4 over #1 is an upset, but #3 over #1 is not) I will continue to use this definition. I'm also still using only the AP Poll. With that, let's look at the underdogs.
By Patrick Barlow 05 Oct, 2022
Everyone loves watching them. They're part of what makes college sports so great. They're also elusive. After all, there's a reason one team is the favorite and the other isn't. I'm talking about upsets. Here I'll take a look at which college football teams are upset the most. In part 2, I'll look at which teams pull off the most upsets. Before proceeding, I need to point out that there is an inherent flaw in simply looking at the number of times a team has been upset. The more a team has been ranked, the more often their chances of being upset are. Because of this, I'll show how many times a team has been ranked, how many times they've been upset, and their upset % (% of weeks ranked ending in an upset). I'll also filter out teams that have been ranked less than 20% of the time (rounded up). First, what is an upset? I've defined an upset for the purposes of this article as a game in which: an unranked opponent beats a ranked opponent, or a ranked opponent beats an opponent ranked more than 2 spots higher than them (for example, #4 over #1 is an upset, but #3 over #1 is not) For ranking purposes and consistency, I'm only using the AP Poll. Note also, that the "Weeks Ranked" does not include postseason rankings, since teams can't be upset after the season ends. I've gathered data as far back as 1998, the beginning of the BCS/CFP era. I'll break this down a few ways and see if I can't find some trends.
By Patrick Barlow 03 Oct, 2022
Each week during the college football season, we'll come here to react to the new TRAM Top 25. TRAM, like most models and systems, gets better and more accurate as the season progresses.
Show More
Share by: