The 2011 Irish season was full of a lot of “what if?” questions. What if Jonas Gray holds on to the ball against USF? What if Dayne Crist doesn’t snap the ball for a second time against USC? What if Denard never got a visit from Aladdin’s Genie? (By the way, I looked up the characters from Aladdin and his name was actually “Genie” – you would think Disney could have had a little more creativity) What if Tommy Rees played a little more Madden so he could learn how to read a Cover 3 defense?
I decided to do a statistical analysis to find out if ND’s luck was actually bad (and they really were two fewer fumbles away from being 10-2) or if this team had 8-4 written all over it from the beginning.
The Process
For those of you who haven’t thought about regression analysis since your freshman statistics class, I envy you. For the entire reading audience, I’ll try to make this as short and sweet as possible while still showing you how I came up with this data.
I took a multitude of team statistics from the NCAA database in an attempt to find the 5 that were the most predictive of overall team success in terms of regular season wins. The Final Five were: Offensive Passing Yards, Offensive Rushing Yards, Passing Yards Allowed, Rushing Yards Allowed, and Turnover Margin. All of these were accounted for on an average-per-game basis.
In the end, these Final Five statistics account for 74% of the variation in regular season wins among FBS teams during the 2011 season. In predicting the number of wins you would expect a team to have based on these Final Five statistics, you can use the following equation:
Expected Wins = 6.0 + (0.022 * Offensive Passing Yards) + (0.021 * Offensive Rushing Yards) – (0.026 * Passing Yards Allowed) – (0.017 * Rushing Yards Allowed) + (1.40 * Turnover Margin)
Please remember that all of the stats are on an average-per-game basis. Also please note that giving up passing yards actually hurt teams more than giving up rushing yards…hmm…
The Results
Before diving into the college football world in general, let’s start with the Irish. According to this model, you would have expected Notre Dame to win (drumroll please…) 6.03 games in 2011. Wow – it must be the turnovers; IT MUST be the turnovers, right??? Wrong. Holding all else equal, improving the per game turnover margin from -1.15 to 0 would only change the expected win total to 7.64 which (while clearly not a possible win total) is still just below the actual number of wins. In all honesty, this truly shocked me. I would have guessed that if we simply had turned the ball over 5 times fewer we would have been spending New Year’s in New Orleans or Glendale.
Looking at the college football world more generally, who were the biggest BCS conference overachievers?
*It seems that overachieving for many of the other schools on that list turned out a lot better than overachieving for ND
And the biggest BCS conference underachievers?
Seeing an analysis that says an 8-win regular season was overachieving was certainly not the result I was hoping for (or expecting), and I imagine that most of you would feel the same way. Looking at the other teams that cracked the top 10 overachievers in terms of wins you’ll find that, for the most part, these schools had what we would consider more successful seasons than the Irish did.
This analysis is by no means perfect. There is clearly something to be said about the timing of turnovers (yes–8 turnovers instead of 10 against USF and Michigan could have meant the difference between an 0-2 start and a 2-0 start), and there is also something to be said about quality of opponents and whether or not yards are picked up in critical points during the game or during garbage time. Nonetheless, it is still a useful tool to look at teams that were able to capitalize on the yards they gained, the yards they prevented others from gaining, and the turnovers they caused or avoided.
So where does this leave the Irish for this upcoming season? Well, a top 10 turnover margin (like Kelly has done before in his career) would have put the Irish at an expected win total of about 9 for the regular season last year (still needed to improve that yardage differential). It isn’t fair to expect that kind of turnaround based on last year’s pathetic effort, but a move into the realm of a positive differential could pay enormous dividends.
Outside of Notre Dame, what made the other teams on the list so successful? Why were they able to rack up more wins than you would have expected based on their on the field performance? While numbers are nice, there are other very important elements (that may be difficult to measure) that the best teams are able to display—making big plays at the end of the game, differentiation in field position from special teams, avoiding 99 yard fumble returns that end in 14-point swings (you can tell I’m scarred by these), nailing opportune field goals, getting out to an early lead, etc. Frankly, these are areas we struggled in last season (other than some thrilling exceptions) and along with a less-insanely-awful turnover differential could have led to a season that looked more like Michigan’s.
To our readers: Do these numbers surprise you? Where do you see the most room for improvement this coming season? On a scale of 1-10, how much do you cringe when you hear the term “multiple regression”?
On a side note, I am excited to be part of the HLS team and hope to provide you with some thought provoking and interesting material in the future. If you are a Tweeter/Twitterer, you can follow me @HLS_Twibby.
Proof that, sooner or later, everyone comes around to love the Irish.
After growing up as a Notre Dame hating, misguided youth, it only took one visit to campus during high school for Twibby to realize the error of his ways. From that point on Twibby and the Irish have had what can only be described as a true Hollywood love story. When he's not reminiscing about his time in South Bend and pondering ways to get a 5th year of eligibility as a student, Twibby writes about Notre Dame and the rest of the college football world for HLS. Along with the Irish, he is a diehard fan of the Chicago Bears, Bulls, Blackhawks and Cubs with a strange affection for Northwestern Wildcats football.
Read all posts by Twibby
Follow @HLS_Twibby
- Chicks Dig the Long Ball: Irish Receivers ’13 - October 23, 2013
- Breaking Down the Run/Pass Balance - September 26, 2013
- 3 Questions for MSU - September 19, 2013
How did you arrive at the static numbers you did? The 6.0 multiplier, .022, .021, etc?
The 6.0 multiplier is the average number of games won, so you are basically calculating the variation from the average (it’s actually a few hundredths higher than 6.0 because of games vs. FCS teams). The other numbers are calculated by doing a multiple regression in Excel. Without explaining the math, essentially the multiple regression finds the numbers for each variable that lead to the best explanation of wins across the 120 FBS teams.
This goes to show that nerds can not figure out football. Break down stats all you want, just like you said they’re imperfect. The human factor plays too much of a role here. You can not measure grit, heart, determination, toughness, and having ice in your veins. I would argue that 3 fewer fumbles gets us 3 more wins last year. If Dayne would have handled that snap against usc on the 1 yard line and we punch it in there we’re tied up. Instead, we saw a 14 point swing which happened to be the final margin.
No, that team underachieved. As I have said before, if we can find good, doesn’t necessarily have to be great, qb to lead this team, we’re a top 5 program.
That’s not the way I read the analysis. Don’t get me wrong, I think our expected wins should be higher, but consider for a moment that this team won 8 games with a turnover happy and young QB. Not only that, they did so with a very porous secondary, no punt return game, a punter that couldn’t punt for a few games, oh and all the other stupid mistakes made by people not named Tommy Rees (and there were plenty).
To me, overachieving can point to one of two things: luck or coaching. If you think luck was on our side this season, I want to be drinking what you’re drinking.
As strange as it sounds, these numbers encourage me. With everyone a year wiser, a rectified QB situation, and hopefully a secondary that can hold its own, I think we could be looking at a solid 2012 campaign despite the SOS.
Speaking of SOS, one more thing to keep in mind: the Irish faced a top 25 ranked SOS in 2011 with superb offenses. That hurts them in this analysis as well.
Twibby, I would love to see an SOS adjusted analysis. You could pull in Sagarin’s schedule numbers, which may adjust things a bit.
In general, feels about right though. I would argue that the timing of ND’s turnovers is where the luck part really comes in.
KY, completely disagree. If stats can’t explain things, no one would ever use them. Yet, everyone does. It’s not a perfect analysis as Twibby notes, but it presents an interesting perspective.
Note I said they can’t explain everything due to the human factor. Tex throws in and interesting point, SOS. Stats can show what is likely, but not what actually happens. The old saying, that’s why they play the game, comes to mind.
Biscuit – I was thinking about doing a SOS adjustment but couldn’t quite figure out how to incorporate it. Once I get the kinks worked out, I’ll fire out an update.
Cool. Check Sagarin on USA today.
The thing that jumped out at me was that the overachievers this past year are normally some of the best teams in FB, year in and year out. The underachievers are normally some of the worst teams in FB every year. The main exception is TX with maybe A&M. I agree that the stats you indicated are the major indicators of success orfailure in FB. It does not include timing as others indicated like fumbling at the 1 yd. line. Mostly, I am thinking of giving up a lot of yards and then stopping the O near the goal line or blocking FGs or having short drives on O, etc.
I am interested to see how 3rd down conversion rates, both offensively and defensively, would change this. I feel like that was one of the areas where we struggled last year, and it could account for why this math has us “overachieving”.
There are lies, damn lies and then there are statistics. Or something. We underachieved the hell out of last year: USF, UM and USC were all not only winnable games, they are games we gave away. I read the entire post/argument and I agreed with the entire premise, except for the parts where I was as confused as my dog, Buttercup, when I explain something to me and he just sits and cocks his head to one side or the other and just gives me that look. You know, that look. Anyway, everything made total sense except for the answer. I’m going back to my central premise: if we value the football on offense and competently execute the offense consistently, we win 10 games. I don’t care if we have Andrew, Everett or a two-headed monster behind center. Gunner needs to redshirt, BTW and IMHO. If we protect the ball and execute the offense, we will win 10 games.
I think the analysis says that we overachieved based on our number of turnovers, not overachieved in general. We definitly underachieved based on how close this team was to ten wins, but we OVERachieved based on how many times we turned the ball over.
*to him.
How do you control for the yards differential that was a direct result of turnovers? The stats just show how much an average turnover changes the expectancy of a win. The soul crushing 99 yard turnover returns still suggest that ND was unlucky because our turnovers were anything but average.
Completely agree. Unfortunately controlling for that across all teams would probably take thousands of hours of watching games from last year unless someone knows where you can find that kind of statistic.