So far, most of our analyses have been focused on the offense and given the semi-consistent changes at the top of the offensive staff, that was warranted. Now, before we turn the page to the 2020 season, we need to look at the defense. So today, we'll take a look at Penn State’s defensive performance as measured by WAR over James Franklin’s tenure. In contrast to the offensive side of the ball, which is welcoming its 4th coordinator since 2014, the defense has enjoyed relative stability. In 2014 and 2015, Bob Shoop, now the safeties coach at Michigan, was the Defensive Coordinator before leaving for the same position at Tennessee. However, his successor, Brent Pry, spent those two years as Shoop's linebacker coach, meaning that although there was a new name at the top, it wasn't an unfamiliar one.
To evaluate the defense, we’re using the same fundamentals of WAR that we used for the offense: Havoc Rate, explosiveness prevented, and ball control allowed. The summary calculations of the stats are here:
- Havoc Rate is the percentage of plays where the defense generates a sack, tackle-for-loss, fumble, or interception as a function of the total plays against. It is very similar to Bill Connelly’s HAVOC stat except that it excludes pass break-ups which are not readily available. The way the stat is calculated has higher values as better for the defense (e.g. 25% is better than 15%).
- Defensive-Time-on-Field (DTOF) is the opponent’s time of possession.
- Defensive Points-per-Minute (PPMD) is the explosiveness metric and is calculated as points allowed per minute of DTOF. For example, if a team allows 30 points in a game where time-on-field was 30 minutes (i.e. each time had equal possession) that would equate to a PPMD of 1.0. The lower, the better.
- Defensive Ball Control (DBC) is the opposite of the OBC stat and is the square root of (Yard * Minute-DTOF per Possession)1/2. As with OBC, we want to include a yardage method based on time and possession to generally indicate a team’s ability to get off the field and give their offense a chance to score. Again, lower is better.
- DWAR is the defensive rating for our WAR system and is calculated from both PPMD and DBC. It may be strength-of-schedule adjusted.
The data is from cfbstats.com, which includes only FBS competition, and is for the 2014 to the 2019 seasons.
Since 2014, there have been five teams (2014 Clemson, 2016 and 2017 Michigan, 2018 Miami (FL), and 2019 Ohio State) that have reached a Havoc Rate of 24%, good for tops in that time span. The top 10% of teams generate Havoc at a rate of 19% or higher (average is 15%) and overall this is a stat in which the Penn State defenses have shined over the last six years.
The Shoop teams had the Lions' worst (17.9% in 2014) and best (22.3% in 2015) Havoc Rates. And even in the “worst” year, Penn State ranked 18th in the country. In 2015, Shoop’s team led the country while generating 46 sacks, 103 TFLs, 35 opponent fumbles, and 10 INTs on 883 plays against.
Pry’s teams have averaged an 18.6% Havoc rate with the best year being 19.6% in 2018 (8th nationally). Under Pry, Penn State has not ranked worse than 17th (2016) and has generated at least 38 sacks, 90 TFLs, 20 fumbles, and 8 INTs every year.
Since 2014 only Clemson (21.5%) and Michigan (19.3%) have a better average performance than Penn State’s 19.1%. Since 2016, under Pry, Penn State ranks 8th nationally and 3rd in the BigTen (behind Michigan and Ohio State).
Defensive Ball Control (DBC) highlights a team’s ability to slow an opposing offense and get off the field. An average team yields 31.2 (yard*min/possession)1/2. And great teams (the top 10%) give up less than 27.6. Out of the top-ten teams of the last six years, eight of them have been from the Big Ten, led by 2017 Wisconsin with 23.4.
It is notable that this metric is partially dependent on offensive schemes and execution. That is, teams that dominate TOP (Army, Wisconsin, Air Force, Navy) have lower DTOF. This is, in effect, using your offense to protect your defense and can be valuable in limiting opponent opportunities (possibly requiring them to modify their scheme) and keeping your own defense fresh. While DBC does not strictly isolate the performance of a single unit, it is still valuable to assess performance. DBC also correlates well to yards-per-possession (R2 = 0.83).
For DBC, Penn State has ranged from great (2014, 25.6 (yard*min/possession)1/2, 5th nationally) to slightly above average (2019, 30.9, 55th nationally). In 2019, the defense was on the field for 32 minutes per game which was similar to 2016 and 2018 (32.3). The last time that the Penn State offense won the time-of-possession battle was in 2017 where the DTOF was 29.6 min/game. So, even though the Penn State defense has been generally disadvantaged in DTOF, they still manage to limit their opponents in yards at a better-than-average rate and maintain their above-average performance in DBC.
We often hear about “bend but don’t break” defenses and if DBC measures the bend portion, the defensive points-per-minute (PPMD) measures the break. An average defense allows about 0.93 PPMD. The worst team in the last six years was the 2018 UConn Huskies who allowed 1.76 PPMD which means that for every four minutes their defense was on the field, their opponents averaged a touchdown. They gave up over 50 points-per-game. Don't be like the 2018 Huskies.
On the other side of the spectrum, you have teams like 2017 Alabama (0.43), 2018 Clemson (0.44), and 2019 Georgia (0.44). At these rates, it takes an opponent a full quarter's worth of possession to score 7 points. An outcome of less than 0.67 PPMD puts a defense in the top 10% of all teams and <0.55 means they’re in the top 2.5% of all teams between 2014 and 2019.
How do Pry’s defenses do? Well, his worst year was 2016 where they gave up 0.79 PPMD which still ranked 28th nationally. Outside of 2016, they went: 2017 (0.56, 4th), 2018 (0.64, 11th), and 2019 (0.53, 5th). So while it may be frustrating to see Tanner Morgan go 18-for-20 and give up yards and possession time, points are the great equalizer. And in this way, Pry’s teams have done as well as we could ask.
Summing it Up – DWAR
The average SOS adjusted DWAR rating is 30.1 and again, it combines PPMD and DBC and lower numbers are better. The three best defenses since 2014 are 2017 Alabama (10.7), 2019 Georgia (11.1), and 2016 Alabama (11.3) while the worst team was 2019 UMass (62.6). A value below 19.1 places a team in the best 10% of defenses and <15 is the top 2.5%. Notably, these values are not yet calibrated against an “average” offense. We will work on that through the never-ending summer of 2020.
In the Big Ten, the 2019 OSU team leads the way at (11.9) and is followed by 2016 Michigan (13.0) and 2015 Wisconsin (13.5). Overall, the Big Ten has the 2nd-best average defensive performances behind only the SEC.
Except for 2016, Pry’s teams have been great in terms of DWAR and these follow the strong performances of Bob Shoop – 17.2 (6th) in 2014 and 21.0 (21st) in 2015. If the 2016 team is the floor (25.3, 38th nationally) Brent Pry deserves even more praise than he's received considering the youth that was getting playing time that year combined with significant injuries that hit the linebacker corp. Since 2016, Penn State has ranked: 6th (2017 – 16.2), 17th (2018 – 19.5), and 10th (2019 – 16.9). So, while Penn State hasn't had the top defense in the country, they're consistently in the top-10 nationally and more importantly, have been no worse than third in the Big Ten since 2016 (7th).
As discussed in an earlier post, the expected improvements that the new offensive scheme should deliver (more ball control, less DTOF) should help push the Penn State defense to new heights by keeping them fresh and limiting extra snaps. With experience at all three levels from P.J. Mustipher and Shaka Toney, to arguably the best linebacker in the country in Micah Parsons, to a solid and hungry defensive backfield that includes Castro-Fields and Wade, this team has the talent and experience to lead the country.