Previewing the 2019 Season: Offense

It’s almost time for another year of Razorback football. The product on the field will probably be better, but our coverage here at Arkansas Fight is definitely poised for a breakout season.

The Hogs labored through the worst season in modern school history last year, going 2-10 with a winless SEC record and a blowout loss to North Texas. Significant contributors on both sides of the ball return and a top-25 recruiting class filled a bunch of immediate needs, but several units on this team are still a few years away from being up to par.

One thing working in Arkansas’ favor is the schedule: while it would be disappointing if the Hogs were good, a four-game non-conference slate of Portland State, Colorado State, San Jose State, and Western Kentucky – all at home – is welcome. The Hogs should go 4-0 here and need just two SEC wins to go bowling. Missouri, Mississippi State, and Auburn come to Fayetteville, while Kentucky and Ole Miss are winnable road games.

This post is meant to be a general preview of Arkansas’ offense. The next post will cover defense and special teams.

Meet the Stats

I’m entering my sixth season doing advanced stats here at Arkansas Fight, and it’s exciting to see how far we’ve come in the stats department. As teased above, our capabilities have increased significantly for this year.

For years, all SB Nation sites that do stats have used Bill Connelly’s S&P+ system to do previews and rankings. It’s a good system: it’s accurate, easy to understand, and informative. Plus, Bill and the team at Football Study Hall made all the data available to anyone. In 2015, I started using EV for the first time after reading this MGoBlog post describing the concept. An identical concept, Expected Points (EP), is described here. I actually prefer EV: it’s harder on my end to calculate, but I think it’s easier for readers to understand because it’s intuitive. But there’s a problem with EV: there’s no Football Study Hall equivalent for it. No open source data. Anything I want to use I have to calculate myself. I’ve developed all the formulas and all the calculation methods. And because my knowledge of database software and statistical programming was very limited, I did most of the work by hand in Excel, so I could pretty much only calculate EV for a single game that Arkansas played in. So since 2015, game previews have still used S&P+, while EV has been only used in recaps.

Now, after spending much of July building a database and writing code, we’re ready to move to an all-EV system. The timing is fortunate: Bill Connelly announced a move to ESPN, taking with him the rights to the S&P+ system he created. So you won’t be seeing much S&P+, but now our EV knowledge is not only vast, it’s also completely unique.

In case you’ve forgotten, here’s a (re-)introduction to EV:

• Expected Value (EV) — or Expected Points — is our basic measure of the quality of each play. It represents the change in the expected number of points brought about by each play. Let’s say you have 1st and 10 at your own 20. Your expected value is about 1.60 points; that is, historical data (for us, every FBS play from 2017 and 2018) shows that all drives in which a team faced 1st and 10 at its own 20 ended up with an average of 1.60 points. Now let’s say you complete a 5-yard pass. That makes 2nd and 5 at the 25, which has an EV of 1.73 points. So that play “created” 0.13 points for the offense (1.60 to 1.73). The end goal, of course, is a touchdown, which is worth the difference between what you had and 6.95 points: six points, plus the opportunity to kick the PAT, which is good 95% of the time. If the kicker makes the PAT, he gets the other 0.05 to get us to 7 (and if he misses, he loses 0.95 to take us back to 6).
• EV can be broken down by play call (rush/pass), by player, and even special teams (placekickers, kick returners, and punt returners can all generate EV). So I can tell you that Alabama averaged 0.52 EV per pass attempt in 2018 (really good) and that Arkansas averaged -0.08 (really bad). I can tell you that when Eastern Illinois committed a block in the back on a kickoff return last season, forcing them to start at their own 15, that penalty cost them 0.49 points, since that’s the difference between where they started (own 15) and where they would have started if they had let the kickoff go through the endzone for a touchback (own 25). Placekicking EV is accumulated based on extra points (0.05 per make, -0.95 per miss) and field goals. Field Goal EV is assigned based on the length of a field goal. For shorter kicks, the offense gets more credit and the kicker gets less, while for longer kicks, the kicker gets more credit. Against Eastern Illinois, Connor Limpert’s 35-yard field goal ended up with 2.5 points retained by the offense and 0.5 added by Limpert, since 2.5 is the average number of points scored by an FBS team attempting a field goal at that distance.
• Added Value, similar to the MGoBlog’s concept of “PAN”, is used to compare EV to some benchmark. There are actually two ways to use it:
• First, you can compare the numbers of a single game against that opponent’s season average to find just how much your over- or underperformance affected the game. Here’s a real-life example: Alabama’s defense surrenders -0.11 EV per rush attempt, which is really good. But Arkansas actually posted 0.12 EV per rush in their game against Alabama, their second-best raw performance of the season. So the Hogs were +0.23 EV per rush above Alabama’s average, and they did it over 29 rushing attempts, which comes to +6.7 points added by the run game, compared to what Alabama normally gives up.
• The second way to use Added Value is to compare every team in a group – the SEC, for instance – to the average of that group. Here is every SEC team on offense last year:

Yikes. You can see that Arkansas’ offense cost it about 8.2 points per game compared to the average SEC offense. Ranking 13th in the SEC on offense is how you go 2-10.

It’s also worth noting that Arkansas wasn’t the worst overall SEC team: that honor would belong to Tennessee, who finished 14th on offense and 11th on defense but managed to go 5-7. How did Arkansas finish in last place without being the worst team? To find out, we can look back at the first way of calculating value-added contribution: by seeing how Arkansas’ performance against each of its opponents stacked up against that team’s season average:

Welp. To call the 2018 Razorbacks a schizophrenic team would be an understatement. It was basically a different team each week. The Hogs never quite put it together: their three best offensive performances came in games where they gave up 65, 52, and 45 points, and their three best defensive performances came in games where they scored 17, 23, and 0 points.

What was the best performance of the season? The answer, of course, is North Texas. Wait… what? Yep, the Hogs went +13 against the Mean Green in a game they lost 44-17. How’s that possible? Well, most of the disasters were on special teams, not offense or defense, so that’s part of it. (During the season, we’ll use a more complete version of PAN that incorporates special teams.) The other part is that North Texas plays in the Conference-USA, so their season averages are really good against bad competition, giving the Hogs a low bar to clear. The same is true for Tulsa (+6.5), the second-best game. Against SEC teams, the Hogs’ best game was against Texas A&M (+5.5), with Alabama (+2.5) being the only other positive game. The worst games of the season were against Ole Miss (-17) and Vanderbilt (-13.5), the two most winnable conference games. That’s how you go 2-10 while Tennessee, a statistically worse team, goes 5-7.

Now for some more detailed run game stats:

• Stuff Rate (StuffRt) is the percentage of a running back’s carries that go for 1 yard or fewer. This stat is designed to tell you a lot about your offensive line, although high-efficiency backs will typically have low stuff rates.
• Line Yards per Carry (LY/A) is another run game stat that measures the offensive line. It’s the average yards per carry if all runs are capped at six yards. So a four-yard run counts as four yards, but a 20-yard run counts as six. The theory is that line blocking is “responsible” for the first six yards, and after that, extra yards are based on downfield blocking and the skill of the running back. This stat can show you when an elite running back is helping to cover for a bad offensive line.
• Explosiveness Rate (Expl%) is the percentage of rushing attempts that gain at least six yards. This figure should be in the neighborhood of 33-35%. It’s arguably the most important of these run game sub-stats – for most backs, about 70-80% of their total rushing yards come on explosive runs – and it is used to help us understand the next one.
• Bonus Yards per Explosive Run (BY/Expl) measures the average number of “bonus yards” (yards beyond six) gained on an explosive run. Bill Connelly called these “highlight yards” because they are not about the offensive line but instead about the running back’s ability to make some highlights in the open field. For example, a 16-yard run counts as explosive and 10 bonus yards are recorded.

And our pass game stats:

• Adjusted Net Yards per Attempt (ANY/A) is probably the biggest passing stat used by football analysts in both college and the NFL. The formula is quite simple: it’s just yards per pass attempt (sacks count as passing attempts), but with every touchdown counting for 20 yards and every interception counting for -45. Those numbers are not arbitrary: they’re the result of fairly extensive calculation (LINK) on the yardage value of a touchdown and an interception.
• Sack Rate (SackRt) is the percentage of dropbacks ending in a sack. This one should be fairly obvious. You may see something called “adjusted sack rate”, which is the sack rate adjusted for how good the opposing pass rush is (or vice versa).
• Adjusted Net Yards per Target (ANY/T) is the same as ANY/A, but for receivers, with targets being the baseline. This number is much higher than ANY/A because receivers cannot be charged with interceptions, and many incomplete passes aren’t charged to a receiver either (for instance, a throwaway).

I’ll use some other stats like passer rating as we go along.

Also keep in mind that all EV plus run and pass game stats can be done by quarter, down, game situation… anything, really. Possibilities are endless and victory comes in finding significant patterns.

Now for some team stats:

• Adjusted Third Down Conversion Rate (Adj3rd) is the actual third down conversion rate compared to the expected third down conversion rate. The expected is calculated using a logarithmic equation based on the yards to go. Shorter distances have a higher expectation of success: 3rd and 3 has a 51.8% chance of conversion, while 3rd and 9 has a 30.7% chance. This number is incredibly useful, as it helps us view third downs in a vacuum. Let’s say two teams both go 3-of-10 on third downs in a game. But Team A had all 3rd-and-12s while Team B had all 3rd-and-1s. Who did better on third down? I’d say Team A, since their low actual third down rate (30%) is really due to failures on first and second down that kept getting them into 3rd-and-12s, and converting 30% of those is actually pretty impressive.
• Points per Scoring Opportunity (PpSO) is the average number of points scored on every drive that gets inside the opponent’s 40-yard line. Drives that get that far should end in points just about every time and touchdowns a good percent of the time. Touchdowns scored from outside 40 yards (like a long pass) aren’t included.

Now it’s time to recap the numbers and look ahead to 2019.

Is 4.4 adjusted net yards per attempt bad? Let me put it this way: it’s more than a full yard per attempt worse than the next-worst team, Kentucky (5.5). It’s only a little more than one-third of the SEC leader, Alabama (12.2). So yeah, it’s bad. Historically bad. Absolutely pitiful.

Quarterbacks

• Key Returnees: None
• Key Losses: Ty Storey, Cole Kelley
• Key Additions: Ben Hicks, Nick Starkel
• 2018 Grade: D
• 2019 Projection: C+

It’s probably best for all involved that Arkansas bid farewell to its two main quarterbacks from 2018. In their place, the Hogs welcome grad transfers Hicks (via SMU) and Starkel (via Texas A&M).

Hicks seems to have the upper hand. Not only did he start for two years for Chad Morris at SMU, he was also on campus for spring practice. Hicks was very productive for Morris, especially in 2017, but his production dipped for new coach Sonny Dykes last year. His ANY/A of 6.9 is on the lower end of what you’d want for a guy stepping up from the American to the SEC.

Starkel is an interesting prospect. He has two years to play, while Hicks only has one, so the safe option is to have Hicks start the season and replace him with Starkel if things go awry. A former four-star recruit, Starkel is probably best remembered in Aggieland for his 499-yard passing performance in the 2017 Belk Bowl. Those numbers are not actually reflected in the table above, as my database collects only regular-season stats. You pretty much always want to take a quarterback who posted 8.1 ANY/A and a 130.8 passer rating at an SEC West school. A warning on the Starkel hype train, though: note the low sack numbers he had at Texas A&M. He won’t have nearly that much time to throw in Fayetteville.

Receivers and Tight Ends

• Key Returnees: C.J. O’Grady, Michael Woods, Deon Stewart, Jordan Jones
• Key Losses: La’Michael Pettway
• Key Additions: Hudson Henry, Treylon Burks, Trey Knox, Shamar Nash, TQ Jackson
• 2018 Grade: C+
• 2019 Projection: B

Nashville (Ark.) native La’Michael Pettway had almost no production during his first three years on campus. Then he led the Hogs in receiving in 2018, starting losing snaps to Mike Woods late in the year, fought with some fans on Twitter, and then transferred. His strange one-hit wonder career is surpassed only by Jonathan Nance, who did the same thing in 2017, but tried to stay for an encore. Nance also had some social media drama before leaving the program in the middle of the season due to lost playing time.

I have no idea what the mood in the locker room actually was, but the consensus seems to be that the bad eggs and the non-buy-ins are gone from a unit that had its ups and downs in 2018. Tight end C.J. O’Grady missed the first four games for personal reasons, then returned and went on a six-game tear in the middle of the season, averaging 60 yards and a touchdown per game during that time. He really needed a wingman to take the pressure off, but now Arkansas has so significantly upgraded in the pass-catching department that he may end up being the wingman. Henry, the nation’s top tight end, arrives on campus to significant hype. Chad Morris may run a spread offense, but he’s a fool if he doesn’t feature the tight end position in 2019.

Arkansas’ wideouts were mediocre at best at getting open. Woods proved to be fairly reliable, while Stewart and Jones have good speed. It’s hard to judge the returning talents because the dumpster fire under center not just in 2018, but in 2017 as well. The Hogs should come out of the gates with improved quarterback play in 2019, allowing both the returnees and highly-touted newcomers Burks, Knox, Nash, and Jefferson to show us their stuff immediately.

These four statistics are arranged this way intentionally. If you think about it, you’ll notice that the left-most stat, stuff rate, is heavily dependent on the offensive line. Each stat to the right is less dependent on the offensive line and more dependent on the back himself. The right-most stat, bonus yards per explosive run, has little to do with the offensive line and is all about the running back’s skill in the open field.

That means the SEC rankings from left to right tell the story: 13th, 13th, 12th, 6th. The less the Hogs’ rushing attack had to depend on the offensive line, the better. Arkansas actually has decent running backs, but blocking for them was mostly a disaster last year.

Running Backs

• Key Returnees: Rakeem Boyd, Devwah Whaley, Chase Hayden
• Key Losses: None
• Key Additions: None
• 2018 Grade: B-
• 2019 Projection: B

The top three options are back for the Hogs. The stylistic differences between them are evident in this table: Boyd is the explosive option. His stuff rate is high and line yards are low, but in exchange, the best one-third of his runs average 16 yards each. When your passing game is as historically awful as Arkansas’ was last year, a back that can create his own fireworks is a necessity. He became the first Hog since Darren McFadden to rush for 100 yards against Alabama, and then had 108 more in the first quarter the following week against Ole Miss before leaving with an injury. Injuries dogged him throughout the year, but if he stays healthy (a big if), he’s a lock to hit 1,000 yards, and maybe even push for 1,500.

The more conservative option is the veteran Whaley. He’s the least-explosive back, but he has the lowest stuff rate and the highest line yards. Basically, he keeps things moving without generating a ton of firepower. If Arkansas’ passing game improves, Whaley’s value goes up significantly.

To drive home how important explosive runs are, note that 85% of Boyd’s rushing yards came on explosive runs. That means 85% of his yards came on the best 33% of his carries. His 81 non-explosive runs gained just 112 yards (1.38 per carry). Even Whaley, the model of an efficient back, recorded 69% of his yards on explosive runs, and his other 57 runs gained just 118 yards (2.07 per carry)

Hayden is a mix of Boyd and Whaley, although he needs more consistency. He gets stuffed too often to be a high-efficiency back, which is probably what he should try to be. He is a good option as a receiver out of the backfield, and that may be his best way to rack up yards if the guys in front of him stay healthy.

Arkansas’ backs are already decent, so any improvement in the overall run game is now dependent on the offensive line.

Offensive Line

• Key Returnees: LT Colton Jackson, C Ty Clary
• Key Losses: LG Hjalte Froholdt, RT Brian Wallace, RG Johnny Gibson Jr.
• Key Additions: OT Myron Cunningham
• 2018 Grade: F
• 2019 Projection: D

All excitement about recruiting and development across the quarterback, running back, receiver, and tight end positions has to be tempered based on the offensive line. It all starts up front. It remains baffling and disappointing how Bret Bielema’s crown jewel ended up like this, but now the Hogs are dealing with the worst (or second-worst… Tennessee) offensive line in the conference. Three starters are gone, including the best player, Froholdt. Despite recruiting success at other positions, the Hogs missed on big targets for the offensive line. That means that just about all improvement will have to come from in-house. I do expect some improvement based on Year 2 in the system and coaching continuity, but the Hogs’ offense is a long way from competing in the big time because of struggles up front.