To Run or Not to Run: That is the Question on Third Down
From 2015 through 2017, “Run the Ball” guy was vocal around Lincoln. A Husker legend for running the ball returned in 2018 and things are looking up for the ground game moving forward. However, the analytics guys preach the diminishing value of the ground game. The analytics guys also say that passing should be the only option on third down. I wanted to see for myself how much better Scott Frost’s teams were at passing than running the ball on third. The results were surprising.
On 3rd-and-10, the data say that Nebraska should run the ball if it wants to pick up the first down. If you identify as a “run the ball” guy, you can stop reading this article now, quote my last sentence, use the graph below as proof that running is better on third and ten. For those of you who want more information about why the data say this, the shortcomings of the data, and add more steam to the 2019 Adrian Martinez hype train, keep reading.
This is a logistic regression analysis. What this means is that I’m modeling the probability of a success outcome (picking up the third down conversion) for a given series of predictor variables. One major limitation of the data is the limited scope of information. The data I used only contains two fields useful for modeling third down conversion success – yards to go and play type. With no information about the original play call, its hard to tell if an Adrian Martinez run on 3rd-and-8 was a designed run or a scramble after no receivers were open. The data treat this as a running play so I did as well for my analysis. A good passing threat can make these runs more valuable to the team.
I modeled the third down success for each of Frost’s three Oregon teams where he was the offensive coordinator as well as each of the three teams he was the head coach. For each chart, the solid line represents the probability of converting a third down running the ball whereas the dashed line represents throwing the ball.
Frost’s teams at Oregon looked like what I would expect for this analysis. Running plays having a high probability of success on short yardage situations with passing plays performing better in longer yardage situations. In 2013, Oregon’s breakeven point for running and throwing the ball came at 3rd-and-8. Marcus Mariota’s Heisman campaign saw a clear improvement of passing over running the ball. Oregon in 2015 produced results I would expect. On 3rd-and-5 or shorter running the ball was more effective. In longer situations, passing was more beneficial.
Frost’s head coaching tenure defies logic when it comes to third down success probabilities. In short yardage situations, UCF had a higher probability of converting third down by throwing, but in longer situations rushing the ball was more effective. How did Frost’s team do this? Of the 21 3rd-and-6 or longer situation faced by his Knights in two seasons, QBs picked up the first down fourteen times. Eleven of those fourteen QB conversions were with runs of ten yards or longer.
Nebraska’s plot was the most unexpected. Passing does not produce and better expected outcome than running until 3rd-and-12. Nebraska ran the ball 24 times on 3rd-and-6 or longer, converting just five of the attempts. Of those five, Adrian Martinez converted four. I found the games on YouTube and watched the four Martinez conversion. Three of the four were designed pass plays where Martinez improvised to make the play. The fourth came against Northwestern. On this play, Devine Ozigbo motioned out wide opening the field for a designed QB draw up the middle.
If the snow was whirling around Memorial Stadium and I had to pick up a third down to ice the game would I run the ball based on this analysis? Probably not. All the graphs show the 95% confidence intervals as well to show that there is no clear significance difference between third down strategy. With a small sample size over just one team per season definite conclusion on third down strategy should not be made.
While this exercise didn’t yield any statistically significant results, I think it does emphasis Adrian Martinez’s talent within this system. If a defense tries to drop seven back in an “obvious passing situation”, Martinez has the talent and vision to pick up the first on the ground. As his accuracy continues to improve, Martinez will be able to beat single coverage with elite passing. The only thing that I am certain of after looking through this data is that with #2 behind center, good things can happen in many ways.