A Numbers Nightmare: The Tragedy of the 2021 Nebraska Cornhuskers
A 3-9 season that should have been a whole lot better!
The early part of the college football offseason is devoted to looking back at the recently finished season and pondering what could have been if certain things went in another direction. Football has a good bit of variance baked into it, so perhaps a fumble bouncing a different direction or a tipped interception falling incomplete could change the course of a game or even season for some teams. Maybe you even play prevent defense on the last play of the game so you don’t lose to a FCS team for the first time in program history (Go Noles!). Many teams may have claims to “we should have been better”, but no team this season, or in College Football Playoff history, has a stronger claim than the 2021 Nebraska Corn Huskers.
To look back at the tragedy of the Cornhuskers, we will primarily be using a metric known as “post game win probability” (PGWP). This metric comes from CollegeFootballData.com, and is defined as follows:
“Post game win probability looks at advanced metrics like success rates, PPA, and scoring opportunities and assesses the probability of each team winning should the game be played again with equivalent stats. In other words, if you take all of the plays in a game and shuffle them into a random order, how often would each team be expected to win?”
Things like penalties and turnovers are not included in the metric due to the noise associated with each variable. PGWP is just one way to observe a game and get an understanding of which team played well enough to win, regardless of the scoreboard.
Wins Over Expectation
Adding every PGWP together will get a team’s expected wins throughout an entire season. Once we have that number, we can use it to look at how many wins above or below expectation a team accumulated in a season. The sweet spot for a team to sit in is within the -1 to 1 wins above expectation area. Anything above 1 signals the team most likely had some luck on their side, anything below -1 and the team probably should have another win or two under their belt.
Nebraska was expected to win 7.6 games this season based on PGWP. They only managed to get to 3 wins. At -4.6 Wins Above Expectation, the Cornhuskers sit in a league of their own in the College Football Playoff era. If we break it down game by game we can see exactly which games were games Nebraska should have had a decent chance to win:
Season Overview
With the exception of Ohio State, every one of Nebraska’s losses this year was by less than one score. They won the post game win probability battle in 9 of their 12 games, against teams like Ohio State, Michigan, Michigan State, and Iowa. Iowa needed a safety, a blocked punt returned for a touchdown, and a fumble recovered to complete a comeback on the Huskers, who had multiple opt outs including QB Adrian Martinez. A fumble deep in Nebraska territory set up a game winning field goal attempt for the Michigan Wolverines, who might not have made it into the playoff had they lost to Nebraska.
It should be noted that we are not making the case that luck was the absolute only factor in these Cornhusker losses. There were breakdowns at all three facets of the game that also contributed to the losses. When losses happen, the Quarterback typically get’s some flak for their play, regardless of if its warranted or not. QB Adrian Martinez was no exception this, as many Huskers fans called for a change throughout the season.
However, when you look at his 2021 season stats, he was actually pretty good. He saw career highs in Expected Points Added per Play, CPOE, PFF grade and the QB Composite score. He certainly didn’t play like a QB on a 3 win team. We can go even further by breaking down one of these metrics that can actually serve as a QB post game win probability metric: ESPN’s QBR.
Taken from this ESPN article explaining how QBR works:
“On a game level, however, a QBR of 75 means that holding all other factors constant (defense, offensive teammates, etc.), a quarterback’s team would be expected to win about 75 percent of time, given that level of QB play.”
We can take QBR and create an expected wins metric similar to the method used above. Once we have our number, we can compare it to the teams actual wins to create a QB wins over expectation metric. A QB with a high wins over expectation would indicate lower QBR’s relative to the amount of wins the team had. This could be a sign of a strong supporting cast around a QB, or perhaps some luck was involved in the season. On the other end, if a QB has a low wins over expectation, their supporting cast might not have held up their end of the bargain, or they were unlucky.
Whether it was bad luck or a supporting cast that let him down, Adrian Martinez had one of the lowest wins over expectation in the College Football Playoff era. So did the supporting cast around him let him down?
One way we can look at the team around Martinez is through F+ ratings, which are a combination of Brian Fremeau’s FEI Ratings and Bill Connelly’s SP+ ratings. We can look at the teams around Nebraska to gauge what kind of season they should have had. Looking at the 5 teams directly above and below Nebraska in F+ ratings, you can see a similar theme reoccur:
Advanced Metrics Overview
Nebraska’s season long advanced metrics continue to point to them winning more ballgames. They ranked 30th in offensive F+, 19th in defensive F+, 31st in FEI and 37th in SP+. The teams around them include teams that made it to their respective conference championships (Western Kentucky, Iowa, Wake Forest) or teams that had years they will never forget. The only teams outside of Nebraska that would like to forget this season was Florida, who fired Dan Mullen, and maybe Boise State. As you dive into the season long metrics, its clear Nebraska should have won more games than they did.
Redoing The Season With Simulations
Earlier we learned that post game win probability is meant to show how often a team would win if the game happened again, given the stats of the game. Well, we can simulate those games (5000 times) and see how many wins Nebraska comes out with at the end of the simulation. After 5000 simulations… we arrive at two seasons where the Cornhuskers only get three wins. Most of the simulations hovered around the 6-9 win mark, which lines up with our other expected win totals. A majority of the simulations lead to Nebraska being bowl eligible, while some lead to even greater heights. I can’t imagine the contract extension Scott Frost would have received if they reached 11 wins.
Conclusion
At the end of the day Nebraska can only claim three victories in their 2021 campaign. Their advanced metrics paint a much more optimistic picture about the Cornhuskers season, and its clear that out of any team they hold the title of “we should have won more this season”. Eventually these will have to turn into wins for Scott Frost, or else he may end up getting the rug pulled out from underneath him. The 2021 Nebraska Cornhusker season will have its place in history as one of the greatest 3 win teams, which is good of badge as any to wear for a losing football season.
If you want to dive in to the data like I do, check out @CFB_Data and @cfbfastR on Twitter, where you can learn how to get started in the world of College Football data analysis!
If you want to see more charts and one off analysis, follow my twitter page, @CFBNumbers
Cover Photo: USATSI