Analytics, Basic Stats, and Recent Historic Context: The ABC’s of Penn State Football – Michigan Recap

Introduction

This feels different. Today we don’t write with a feeling of dread but rather mild contentment. I’m still not sure that the Penn State football team is “good” but maybe, at least, they’re recovering from the disaster that was the first five weeks of the season and beginning to figure things out.  

Basic Statistics – Summary

Finally, we can turn the first row green for the Lions who won the most important category of any game – the score. Not only that, but they pretty much dominated the rest of the contest too, racking up 131 more yards, controlling the ball for more than 36 minutes, and not turning the ball over (not even once!). Even the one category that the Wolverines won – rush yards per attempt – is misleading because Penn State still had 80 more total rushing yards.

Look, it wasn’t pretty and there’s still A LOT to be desired, especially in the passing game, but this was an objectively better performance than what we all witnessed in the first five weeks. There was a feeling of competence and decisiveness to the offense. There seemed to be variety in the playcalling. The defense, while still not being able to contain an outside running game, appeared to generally be more aggressive than in weeks gone-by. I want to believe that this is something a team can build upon. There’s an adage in the market: “Stocks take the staircase up and the elevator down, If you consider the 2019 Cotton Bowl as peak Penn State football, the program took a broken, freefalling elevator down through the first five games of 2020. Maybe this is the first stair-step back up towards that peak we used to know, and perhaps took for granted.

Semi-Advanced Stats and VERY Recent Context

Going into Saturday, I had intended to do a deep-dive retrospective on all of the failings of the team’s performance since I fully expected them to be 0-6 by late Saturday afternoon (mea culpa to Coach Franklin and Co.). But don’t worry haters, there will be plenty of time for that in January. Today, instead, we will focus on what improved this week. 

We’ll start on the defense and the graph below which shows the relationship between the yards a defense allows versus the points the opposing offense scores. These data points are for every Big Ten team from 2016-2020 and do not include games from this weekend (i.e. only through the first 5 weeks of the season) and are only against FBS competition. There’s a pretty clear trend that the more yards a team allows, the more points they allow. This is intuitive and obvious. The trendline is shown and indicates expected performance considering all the data. There are a few obvious outliers, including one that sticks out like a sore thumb – 2020 Penn State. Look at that 2020 Nittany Lion logo all by itself at 36 points and 360-ish yards. This is bad. For the number of yards that the PSU defense allowed through five games, they “should have” only given up about 24 points per game. The difference between the actual data point and the trendline is the “residual” and the 2020 Penn State defense had a residual of 11.7 points. That is the worst residual on the graph by nearly 4 points. Interestingly, the 2019 Penn State defense had the best residual, allowing 7.8 points fewer than the trend would predict – for the same number of yards/game allowed as this team. 

The 2019 performance was unlikely to be repeated and regression may have been predictable, but what’s occurring this year is much more than “regression to the mean.”

Now, these point values include the pick-sixes and scoop-and-scores that the offense gave up to Iowa, Nebraska, and Maryland and those aren’t on the defense. But even if those were removed, this defense would still have the worst residual in the Big Ten since 2016 at 7.02 points above expectation.

On the flip side is the same data from the offensive perspective shown below. In this case, above the line is good and below the line is less good. The 2020 Penn State team is marginally below the line, meaning they’re not generating as many points one would expect (~32 PPG) from the number of yards they’re accumulating (418 YPG). 

Interestingly, look at how much better Ohio State has been than everybody else for the last 5 years. They weren’t the outliers (2016 Michigan and 2017 Penn State) producing at a higher than expected rate; they’ve just been good. They’ve averaged >500 YPG every year except 2016. At the other end of the spectrum is Rutgers, a program literally at the bottom of the league every year except 2020. This year, the Scarlet Knight have shown marked improvement (29.6 points & 360 yards per game).

So what we have through 5 games is a defense that is far underperforming where they should be and an offense that’s not as bad as we think but still not great. Let’s explore a bit more.

Red Zone Opportunities

Since 2016, Penn State has had very good red zone offenses and defenses. In the graph below, the y-axis is the number of red zone opportunities for the offense (orange) and opponent (blue) per game. The labels are the average number of red zone points per game that either the offense scores or the defense allows. Pre-2020, a few things jump out. First, the offense generated many more red zone attempts per game with the maximum coming in 2017 with a differential of about 2 opportunities each game (over the 13 games this means the offense had 26 more red zone opportunities than the defense allowed). What’s more is that the offense tended to turn these opportunities into points through generation of a minimum 20.08 PPG in the red zone and the max of 26.38 (note that in 2017, the team had fewer average red-zone opportunities per game than 2016 but was better at scoring, thus the higher number label). 

Through five games in 2020, the offense has averaged 4 red zone trips per game and the defense has allowed 4 per game (through both typical opponent drives and the offensive turnovers in the opponents’ territory). This year, for the first time in 5 years, the defense is allowing 20.4 RZ-PPG and the offense is scoring only 15 RZ-PPG (net -5.4). The red zone scoring rate in 2020 is 65% — well below the average of 89% in the preceding four years. Conversely, the “bend don’t break” Penn State defenses of 2016-2019 allowed an average of 82.5% opponent RZ scoring and this year, they’re allowing a score rate of 90% (60% TDs and 30% FGs).

Against Michigan, the defense allowed only 2 red zone opportunities, half the seasonal average and gacv up 14 RZ points – not great but an improvement. The offense took advantage of its four red zone trips with 2 field goals and 2 touchdowns (20 points; 100% scoring rate). You’d still like to see the FGs turned into TDs but again – little steps of improvement…

Opponent 3rd Down Conversion Percentage

From 2017–2019, I vaguely remember screaming at my TV saying, “Why can’t they stop [Insert mediocre QB here] on 3rd down!?!?”  But upon further review, the Nittany Lions D actually ranged between “okay” and “pretty good” in that category – even though it didn’t seem like it. Not this year, though. Notably though, the league average is 40.1% 3rd down conversions against, which is up more than three percent compared to the 2016-2019 success rate. 

Penn State had its best performance of the year against Michigan, only allowing a 33% conversion rate on 3rd down (and 1 of 2 on 4th down). Very coincidentally (or not) Penn State duplicated its 33% conversion rate vs. Indiana and Nebraska – two games it had chances to win. Compare that to blowout loses against Ohio State (10/18), Maryland (9/16), and Iowa (7/15). 

Havoc Avoidance Rate (HAR)

We’ll end this week with what may have been the biggest improvement for the offense and that was performance in avoiding bad things – Havoc Avoidance Rate (HAR). I’ve written about HAR before but a reminder, it’s the percentage of plays an offense has that doesn’t end in something bad (sack, TFL, fumbles, interception). Against Michigan, Penn State showed significant improvement compared to the previous five games by allowing only three Havoc plays out of 79 total offensive snaps. Here is the HAR summary game-by-game. Those first five blocks are ugly and littered with turnovers, too many sacks, and loads of TFLs. Against Michigan, there are no orange or red bars (turnovers) and barely any negative yardage plays. The Michigan defense probably stinks, but this is progress.

In the graph below, we see the number of plays run (x-axis) against the HAR (y-axis) with the label as the number of non-havoc plays. I’m not suggesting that this is the most important metric, but it is undoubtedly important. In the three games where the offense posted its highest HAR (Michigan, Indiana, Nebraska) they’ve been in contention to win. When HAR dips significantly, Penn State loses by more than two scores. In context, if the PSU runs 75 offensive the plays, a HAR of >90% means that something bad happens on less than 8 plays. When you see a performance like Maryland, Iowa, or Ohio State, for those same 75 plays, you expect negativity on 16 to 20 plays. Protect the ball and don’t let bad stuff happen!

Summary

So, after a 10-point win at a pretty bad Michigan, is Penn State Back™? It’s likely far too early to predict a CFP or Rose Bowl run in 2021, but a win is a win. Penn State didn’t dig an early hole and found the ability to close. So that’s progress. Are there still problems? Lord yes. The offense still lacks explosion, the passing game isn’t fluid, tackling still looked suspect, and Penn State’s defense got pushed around on the ground. But, the possibility exists that this is a turning point and both the players and the staff have figured some things out. Going back to the first graphs – the relationship of yards gained/allowed to scoring – the Penn State defense allowed 286 yards against the Wolverines which should translate to allowing 16 points and they gave up 17. The 417 yards gained by the offense should yield 32 points and they scored 27. So, while there’s still a considerable distance from the 2019 defense or 2017 offense, there was clear improvements in fundamental areas of the game (havoc avoidance, red-zone scoring, limiting 3rd down conversions) No matter how messy Michigan is, credit needs to be given to the Lions for pushing onward and upward and finding a way.