The Portland State Preview: A Stats Study

(For an overview of Arkansas Fight’s advanced stats, please check out this article.)

If we count the disastrous 2018 season as a “Year Zero”, then Saturday marks the start of Year 1 of the Chad Morris era. Hope springs eternal. Fans who were bitterly pessimistic at the end of 2018 are suddenly excited.

That hype may not be entirely misplaced. “Year Zero rebuild, followed by massive improvement” has been a theme in recent Arkansas history:

“Overall PAN” here is Rush PAN + Pass PAN + Rush D PAN + Pass D PAN. Penalties and special teams are not included (which is why the 2011 team, which led the FBS in non-offensive touchdowns, appears so low).

If the 2019 team makes a jump similar to the 2009 team (+11) or the 2014 team (+15), then the Hogs may actually field the fourth-best team of this era, 2006-2019. An overall PAN of +5 would be better than eight of Arkansas’ 2019 opponents were last year, including Kentucky, Ole Miss, Texas A&M, and LSU.

The warning is that on-paper improvement and wins improvement aren’t always identical. Petrino’s second team went from -3 to +8 in 2009, but the record only went from 5-7 to 7-5, thanks to several close losses. The 2010 team improved by only two PAN (+8 to +10), and yet the win total jumped by three, to 10-2. Bret Bielema suffered the same fate: his 2014 team that went 6-6 was (on paper) better than the 2015 and 2016 teams that went 7-5. On-paper improvement pretty much always pre-dates wins. So if the Hogs do make a big jump in 2019 and get to +5, the record may only be 6-6, but a big 2020 will be set up.

Defining Efficiency

Efficiency measures the ability of an offense to stay on schedule. Compared to its sister stat, Explosiveness, this one is fairly easy to understand. An incomplete pass or a play for a loss puts the offense “behind schedule”. An offense that spends a bunch of its time “behind schedule” is said to be inefficient.

For years, we have used a statistic to measure efficiency: success rate. Success rate is the percentage of a team’s plays that are “successful”. Here at Arkansas Fight, we’ve defined success thusly: on first down, a play must gain 40% of the yards to go to be successful; on second down, a play must gain 67%, and on third and fourth downs, a play must gain 100%. So a 4-yard gain on 1st-and-10 is successful. A 6-yard gain on 2nd-and-3 is successful. A 2-yard loss on 1st-and-10 is not successful. And so forth and so on. It’s mostly intuitive.

We’re now going to define success slightly differently. This measure is much simpler and ties into our EV system: success is now defined as any play that generates positive EVA. So a play that takes the offense from 2.42 EV to 2.49 EV (for +0.07 EVA) is now successful. I checked this new way of calculating success against the entire 2018 football season and found that it changes… almost nothing. It turns out that the 40%-67%-100% formula for determining success was pretty accurate. But we’ll stick with this new way.

Success rate is very easy to adjust for opponent. All we have to do is measure our raw numbers against the opponent’s average. So let’s say Arkansas records a 44% success rate on offense against Alabama, and the Alabama defense’s season average success rate was 40%. Arkansas is said to have a marginal efficiency of +10%, because Arkansas’ success rate against Alabama was 10% better than what Alabama normally allows. Early in the year, we won’t be able to use much marginal efficiency because, obviously, we don’t have enough data on every team in 2019 to formulate an accurate season average. I’m working on a “weighted-average” system that combines early 2019 results with 2018 numbers, but it’s not finished at this time.

So that’s efficiency. The more elusive figure is efficiency’s evil twin: explosiveness.

Defining Explosiveness

Everything in this section is taken from this wonderful piece in Toward Data Science, written by Blake Atkinson. He notes the key problem with trying to define explosiveness: not all yards are created the same. What’s harder: gaining 1 yard on 4th-and-1, or gaining 1 yard on a run when the running back has already gained 27? Going from 27 to 28 yards on a play is not very hard, but going from 0 to 1 is very hard. That’s one problem.

Another problem is that some plays are going to be touchdowns no matter where they start from. Atkinson links a video of a 90-yard touchdown run by Adrian Peterson, noting that the play gained 90 yards only because it started from Washington’s own 10-yard-line. Had it started from the opponent 40, it would have only gained 40 yards. This is why using isoYPP – or “yards per successful play” – isn’t sufficient: a team that rips off a huge number of 20-yard plays, but no 40-, 60-, and 80-yard plays can’t really be called explosive.

In theory, an explosive team is one that generates a big mix of 20-, 40-, 60-, and 80-yard plays. The most explosive teams should have the largest variability in the number of yards gained on any given play.

That “variability” has another name: inequality. You mostly hear this phrase used in the political and economic spheres to talk about income inequality. Simply, “income inequality” refers to the measure of variability (inequality) in all the incomes of every person in a nation. And there’s a statistic to measure inequality: the Gini coefficient. Gini coefficients are given as a number between 0 and 1. A Gini coefficient of zero means all numbers in the dataset are the same (“equal”). A Gini coefficient of 1 means that the entire sum of all data points belongs to just one point (“unequal”). So a high Gini coefficient means that there is high inequality among the data points.

We can apply a Gini coefficient to a set of plays to determine explosiveness. The easiest way to do this is to combine success rate (x-axis) and the Gini coefficient of successful plays (y-axis).

Boom! That’s pretty easy to read. You can see where some of the best offenses rank. The Hogs have, uh, a long way to go. Keep in mind that I used raw success rate – not marginal efficiency – and if I had used opponent-adjusted numbers the Hogs would have fared (slightly) better in terms of efficiency. But still, last year was rough.

We can do this for just run plays:

This is pretty much what you’d expect. Teams like Wisconsin and Army fall into the “not explosive & efficient” grouping, while Clemson and Memphis (the #1 rushing team in America according to EV/PAN) tended to be more explosive. The Hogs were neither, but had Rakeem Boyd been healthy all year, they probably would have at least made the “explosive & not efficient” quartile.

We’ve discussed Chad Morris’ goal of building a more explosive offense by building a more explosive run game. This is going to take some time, as Morris first had to rebuild the foundation set by Bret Bielema, who was more interested in high-efficiency offenses. Take, for example, the 2015 passing game, which led the FBS in opponent-adjusted ANY/A:

That’s what happens when you emphasize the tight ends in the passing game and focus on staying on schedule.

We’ll keep an eye on the Gini coefficient as the season goes on. It will help us understand how explosive Arkansas’ offense is becoming.

A Peek at Portland State

Head coach Bruce Barnum enters his fifth season leading the Vikings. After a 9-3 first season that included a trip to the FCS Playoffs (the program’s first since 2000), things have not gone smoothly. Portland State is just 7-26 over the last three seasons, including 4-7 last year. Even if they’ve improved (they will be, as most starters return), they probably won’t be a more significant threat than Eastern Illinois was last season, and the Hogs won that game 55-20.

When Arkansas has the ball

Here’s a brief explainer on the numbers:

• PAN and marginal efficiency (the bolded numbers) are opponent-adjusted, and the opponent-adjusted ranking out of 130 FBS teams is provided. Arkansas having -1.1 PAN means that Arkansas’ run game cost it 1.1 points per game compared to the cumulative average of Arkansas’ opponents. (A PAN of zero means that particular unit was playing at a 6-6 level, give or take.) Arkansas having -7.0% marginal rushing efficiency means Arkansas’ rushing success rate was 7% worse than the cumulative average allowed by Arkansas’ opponents.
• The other rushing and passing numbers are given raw, but the ranking is still opponent-adjusted. I did this because it’s easier to read the raw numbers than the opponent-adjusted ones (for example, it’s easier to understand 4.2 ANY/A than -42.8% adjusted ANY/A), but it’s still important to opponent-adjust these numbers for rankings purposes.
• The three numbers at the bottom are just raw, both the value and the ranking.

Arkansas’ main offseason upgrade came in the passing game. The Hogs have a new quarterback and added a four-star tight end and three four-star wide receivers. That should improve the skill positions. Improvement in the offensive line (90th stuff rate, 70th line-yards per carry, 99th sack rate) will have to be internal, as the Hogs missed on some big recruiting targets.

The running backs are already talented, especially when they break into the open field. And it’s worth noting that the offense, even as it was struggling, was pretty disciplined, ranking 15th in net penalty EVA (this includes all penalties committed by either team while Arkansas was on offense).

With Ben Hicks starting, I expect the Hogs to throw a lot, because I think Morris would prefer to get more reps for his quarterbacks and receivers, who need experience more than the running backs right now.

Portland State’s defense isn’t going to offer much resistance. In 2018, the Vikings gave up 72 points to Nevada and 62 to Oregon to open the season. Eastern Washington scored 74, Idaho State scored 48, and Montana State scored 43. So, yeah, this isn’t exactly an SEC defense. They ranked 122nd in the FCS in total defense, which is next-to-last.

Eastern Washington ran the ball 27 times for 378 yards and five touchdowns, including runs of 81, 74, 66, 36, and 33. EWU also completed a 68-yard touchdown pass in that one.

When Portland State has the ball

The Hogs have solid linebackers (at least the starters… depth is an issue), but it will be interesting to see how the re-tooled defensive line and secondary perform against a weak opponent.

The 2018 Hogs were good at limiting big runs (36th in explosive run rate, 53rd in bonus yards), good at getting after the quarterback (34th sack rate), and good on third down (46th). They were bad at pretty much everything else. Often put into bad positions to start with (118th in field position), they struggled to disrupt the run game and left too many receivers open.

Portland State is going to test the secondary. The Vikings love to play wide open, often putting four or even five receivers out. They love big plays are willing to trade completions (53%) for big catches (14.4 yards per completion). The quarterbacks (there are two) don’t appear to have very strong arms and most big pass plays are catch-and-runs. Their run game is not good, but both quarterbacks that they used in 2018 are fairly mobile. A sustained pass rush can force mistakes, but a lack of pressure could allow them to scramble around until someone comes open downfield.

As I mentioned in the special teams season preview, the Hogs actually might be greatly improved on special teams. They have an above-average kicker in Connor Limpert. They will probably once again have an above-average kick coverage unit. If De’Vion Warren stays healthy, kickoff returns will also be above-average. The Hogs have a new punt returner (Treylon Burks) and may have a new punter (Sam Loy, or Reid Bauer). It’s not hard to see improvement.

Keys to the game

1. Be mistake-free in the passing game. Portland State has a truly horrible passing defense, so there’s not going to be much to learn there. Besides, does anyone else remember Ty Storey going 12 of 17 for 261 yards and three touchdowns against EIU last year? The stats aren’t going to tell the story of this game. But Morris says Ben Hicks is the starting quarterback because of his command of the offense, so that’s what Hog fans should be looking for. Everyone on the same page, no delay of games or confused timeouts, no wrong routes, etc.
2. Defensive disruption. Last year’s defense was decent on third down and at limiting big plays, but it was helpless against moderate gains on first and second down. Even EIU was able to churn out 20 points and 357 yards against a generous Razorback defense. A more serious clamp-down of an inferior offense would be nice.
3. Return game dominance. I expect better punting, punt returns, and kickoff returns from the 2019 Razorbacks. Let’s see that in game one.