As we noted earlier this week, Ohio State really has to win this game against Penn State to have a good chance of making the playoff.
While there’s a chance that 538’s playoff model (that is used in the linked article) might overweight the value of a conference championship, it is still highly unlikely that the Buckeyes make the playoff and achieve their season’s goals if Penn State gets the upset.
Ohio State is currently favored by 19 and most advanced stats favor the Buckeyes by multiple scores (SP+ likes Ohio State by 15 points, with an 80% win probability). But given Penn State’s high talent level and performance this year, an upset is still well within the realm of possibility.
So what would an upset look like? And how is this game most likely to go? Let’s dig into the analytics.
Before we go much further, it’s worth explaining the most common metric I’ll be using below — expected points added, or EPA. EPA allows us to evaluate the magnitude of success for an individual play, and by extension, a player, side of the ball, or entire team.
While success rate (another common advanced stat) only looks at the binary question of “was this play successful or not” based on predefined goals given a particular down and distance, EPA allows you to not only say that a play was good, but also how good. Expected points themselves are derived from a statistical model for each down, distance, and field position. If a play does better than expected given the play’s down, distance, and field position, then it will have a positive EPA.
EPA success rate then takes the percentage of positive EPA plays as a measure of efficiency. Average EPA takes the mean EPA, so it’s much more susceptible to extremely positive or negative plays, and can capture explosiveness more than just EPA success rate.
To make it easier to understand what a good EPA score is and isn’t, I usually convert EPAs into percentages relative to all other FBS teams. (Note that these are actually the p-values for each EPA metric, so you can interpret the percentage as, “there is a x% chance of a team having a lower EPA than this team did.”)
Alright, let’s start with some overall team comparisons. This first chart looks at all FBS teams, with the x-axis showing offensive EPA per play percentages, and the y-axis showing defensive EPA per play percentages. Essentially, up-and-right is good:
As you can easily see, Ohio State is the most up-and-right team on the EPA map (please also note Clemson, which is coming in fast for Ohio State’s spot). Notably, Penn State is significantly lower in offensive EPA, above almost 67% of teams in offensive EPA.
Let’s also take a look at team talent by position. Cleveland.com’s Doug Lesmerises had an excellent breakdown of each team’s recruiting over the last four years that you should definitely read as well.
This particular chart looks at the top quartile of recruits’ 247 Composite ratings that each team has signed over the last four years. I should also note that I added in Justin Fields to this comparison, but he was the only transfer I added in.
I chose the top quartile as shorthand for each team’s starters (making the obviously not true assumption that the most talented players would get starting spots), although the positional trends are the same if you just look at average player ratings as well.
As you can see, the only position where Penn State has out-recruited Ohio State at the top is at running back, where their run of Saquon Barkley, Miles Sanders, Ricky Slade, Noah Cain, and Devyn Ford rivals any team in the country over the last four years.
Both teams are elite recruiters, but Ohio State has relatively large recruiting advantages in the front seven (both defensive line spots and linebacker), quarterback, safety, and wide receiver.
Nevertheless, Penn State’s top quartile of players were still blue-chippers at every position.
Get ready for some passing
(Instead of a table, I decided to try putting the overall offense vs. defense (and vice-versa) data in chart form this week, so let me know what you think!)
The Penn State defense has been very good this season, with an 82% probability that a random FBS team would have a worse average defensive EPA than them. They have only allowed more than 13 points in three games — Michigan (21), Minnesota (31), and Indiana (27).
The good news for Ohio State is that those teams were Penn State’s highest-ranked by SP+, at 10th, 16th, and 20th. Indiana and Minnesota both had positive average offensive EPAs, with offensive performances rating in the 86th and 72nd percentiles, respectively.
Both teams found success throwing the ball especially. Sack-adjusted, Minnesota averaged 15.7 yards per attempt, creating an explosive pass (20+ yards) on a truly astounding 38% of throws with an overall 67% passing success rate. In terms of passing EPA, Minnesota’s performance was in the top 98.5% of all games this season. Minnesota in particular was able to find success with both RPOs as well as with receivers beating man coverage, allowing for big plays if the corner was beat. (I’d highly encourage you to listen to the episode of Locked on Buckeyes with Ross Fulton, who had an excellent breakdown of Penn State’s offense and defense).
The stats reflect Penn State’s struggles with pass defense, particularly in allowing explosive passes. They rank in the 63rd percentile in average passing EPA, but are in the 89th percentile in passing EPA success rate. Ohio State will likely need to beat man coverage, and might not get the generous cushions that Justin Fields was able to exploit for easy eight-yard gains for so much of the season.
The EPA data suggests that Ohio State will have a tougher time running the ball, however, as Penn State looks like it could have a marginal advantage in both rushing EPA per play and rushing EPA success rate. Ross went into Penn State’s defensive scheme in depth on the podcast linked above, but it’s worth noting here that Ohio State could have some tactics to still run effectively on Penn State, particularly if they can involve Justin Fields in the run game or alternatively constrain the Penn State defense with backside throws. Rutgers was actually solid prep for the Buckeyes run game (even if the Buckeyes went very vanilla with their run game play calling, their run game performance was still only in the 39.5 percentile).
Ohio State’s offense may also be slowed in the red zone and in short-yardage rushing situations. The red zone will be particularly important to watch. Ohio State is within the top 14% in red zone EPA success rate, but Penn State’s defense is strong in the red zone where there is less space for opposing receivers to get open. Ohio State will absolutely need to maximize its scoring opportunities to win this week and next. Look for Ohio State to get Justin Fields involved in the run game in the red zone, or for Ohio State to take a shot for the end zone at the edges of the red zone.
Speaking of taking shots, it’s been interesting to watch Ohio State’s offense take a few more shots downfield in recent weeks, and especially against Rutgers. Ross Fulton pointed out that this may be because Ohio State treated Maryland and Rutgers as scrimmages to prep for Penn State and Michigan.
So, I thought it might be interesting to look at EPA success rate vs. average EPA on only successful plays. Isolating EPA on only successful plays allows you to get a feel for explosiveness even more than when you are just looking at average EPA on its own. As you can see in the chart above, Ohio State is one of the most efficient teams in the country, but is actually fairly low in terms of average EPA per successful play. My sense here is that their average EPA is drug down by the relatively high percentage of efficient-but-not-explosive plays that are their bread-and-butter. Few teams — really only Alabama and Oklahoma — can manage to be both explosive and highly efficient in this metric, so it’s not too surprising that Ohio State might be a little lower here. But it’s worth watching whether the Buckeyes can either maintain the high efficiency level they’ve averaged this season, or whether they can create enough explosive plays to offset a potential hit to their per-play efficiency.
Penn State’s offense is a little volatile, but they can be explosive
Alright, so now let’s look at the Penn State offense. Ohio State has an advantage in each of the EPA categories above except for rushing EPA, where the Nittany Lions are 1% higher.
The most striking thing about the chart above is Penn State’s standard downs (first down, second-and-7 or less, third-and-4 or less, and fourth-and-4 or less). Penn State ranks in the top 15% in standard downs EPA, but only in the bottom 29th percentile in passing downs EPA! Sean Clifford, who has been solid but inconsistent replacing Trace McSorley this season, has struggled in obvious passing situations — and the Buckeyes are likely to really try and exploit that, particularly with Chase Young returning to the defensive line. Penn State ranks 51st in havoc rate allowed this season, allowing havoc plays on 17% of snaps. Against Minnesota, Clifford threw three particularly costly interceptions, because they were mostly in the Minnesota red zone, killing otherwise promising drives.
Overall Clifford ranks 38th in passing success rate and 48th in 20+ yard throws (at a rate of 9.8% of passes, including sacks). I was a little surprised honestly to only see those ratings for the PSU passing game (along with just a 57th percentile for their passing EPA per play), because I came into this preview expecting the Nittany Lions to be more inefficient-but-explosive through the air than they apparently are. We know from last season that KJ Hamler is a huge big-play threat, and he’s far-and-away their top target, although tight end Pat Freiermuth and Jahan Dotson are solid options as well. Apparently Hamler’s status is a little unknown heading into this matchup.
As mentioned above, Ohio State’s defense has a very slight disadvantage if you look at rushing EPA, but rushing EPA success rate projects Ohio State to have a major advantage. This is because Ohio State rarely allows explosive runs (which are very high EPA plays allowed), but the ones they have allowed have skewed their overall average rushing EPA.
Penn State might not have a ton of success exploiting that though, based on their past performance. The chart below shows the percentage of run stuffs on the x-axis, with the percentage of explosive runs on the y-axis:
Penn State manages to avoid stuffed runs better than most teams, but they aren’t quite at the level of playoff contenders as highlighted above. Here’s their rush yardage breakdown:
- Stuffed runs (0 or less): 16%
- 1-3 yards: 36%
- 4+ yards: 48%
- 12+ yards: 7% (97th in the country)
This suggests that explosive runs shouldn’t be that much of an issue.
Finally, Penn State’s offensive performances have been significantly more inconsistent than Ohio State’s this season. The box plots below show both teams’ average EPA percentiles for each game. Ohio State’s worst offensive performance was actually against Florida Atlantic in Week 1, but it was still near Penn State’s median performance:
Overall this projects as a solid Ohio State win, assuming that the Buckeyes State don’t allow explosive runs or a big game from KJ Hamler, and if the offense can take advantage of scoring opportunities and hit some explosive passes of their own to compensate for what could be a tougher game running the ball.