No question, 2014 ended on a high note for Notre Dame football. There was plenty of turbulence both on and off the field, but the thrill of a last second bowl win versus a member of the almighty SEC left many fans feeling more optimistic heading into the offseason. In some alternate universe not too distant from the one we all live in, that might even be the only thing to discuss before next August. Ha. Ha. Ha. Don’t worry, I’m not that delusional to think such a thing could come true. But let’s glance once more at the Music City Bowl result before moving onto other offseason topics.
A fan and analyst axiom oft heard leading up to and following bowl games is: “Hopefully [our team] can use a bowl win to springboard into next season.”
Seems fair. Certainly, it’s always better to win a bowl game than to lose a bowl game. You can reach many logical conclusions about bowl wins: 1) Wins assist recruiting, 2) It’s better for team morale, and 3) Maybe the team figures out a new philosophy that’ll carry through to next season. I have no idea whether any of these things are true, but they definitely sound like something that could be true. They’re also the types of statements that I tend to loathe because they’ve been used as crutch arguments by sports fans since the beginning of time. Seemingly obvious truisms are dangerous to fan and team expectations. Extrapolating large truths from isolated outcomes is something always worth exploring in greater detail.
With that in mind, I wanted to answer (or at least give thought to) a very macro question that arises post-bowl season: Empirically, is a bowl win likely to lead to more success the following season, and if so, how much?
To begin to answer that question, I looked at the 210 teams that participated in the 105 bowl games taking place between 2011 and 2013 and asked that very question: Did those teams win more or less the following year, and did it matter whether a team won or lost their bowl game? Before we get into the results, let me say two things. First, I should reiterate, that given the choice, I’d prefer my team win its bowl game rather than lose it. It gives me the good feels to see my team win, and more good feels = positive. However, there’s a difference between what is emotionally rewarding and what actually matters to results on the field.
Secondly, there are plenty of factors that change season over season. I am sure we could micro-analyze the results for every team that I looked at and provide a more nuanced explanation. So, consider this “study” for what it is. If you know nothing else other than whether a team won or lost their bowl game and the number of wins in that season, can you predict how the team did the following year?
Let’s start off with the generalized data:
TEAM SUB-SET | WIN% IN YEAR | WIN% YEAR+1 | % CHANGE: |
---|---|---|---|
ALL | 67.57% | 62.11% | -5.46% |
WON BOWL | 71.70% | 64.17% | -7.53% |
LOST BOWL | 63.44% | 60.04% | -3.39% |
PRE BOWL -WON | 69.36% | 64.17% | -5.19% |
PRE BOWL -LOST | 68.65% | 60.04% | -8.61% |
ALL TEAM 2011 | 66.49% | 61.85% | -4.64% |
ALL TEAM 2012 | 67.35% | 62.79% | -4.565 |
ALL TEAM 2013 | 68.86% | 61.71% | -7.16% |
*Rounding Errors may cause slight variations in % change.
Across the board there is an obvious pattern: Teams that went to a bowl, on the whole, tended to lose more the following season regardless of bowl game result. There are only so many wins distributed each college football season. The general trend was that teams failing to make a bowl game tended to claim more wins the following season over their bowl participant counter parts. From the three years of data that I looked at, the only sub-set I found that posted a better win percentage the following year were the losers of bowls in 2011 who saw a modest improvement in win percentage of 2.45%.Teams that lost their bowl game had a slightly lower winning percentage the next season, but saw a less dramatic change year over year. Bowl losses appeared to be a harbinger of what a team’s talent level more accurately was. As seen above, teams pre-bowl, win or lose, had very similar win percentages coming into the game (less than one percentage point separated the two). Taking a collective 0 for 105 obviously harms the win percentage of bowl losers for that year, but there did appear to be a bit of a hangover effect.
Were these changes the result of a couple of teams nose-diving the following season? After all, 2011 Hawaii Bowl winner Southern Mississippi went 12-2 that year. In 2012, Southern Mississippi went 0-12. As you might guess, that was the largest drop year over year in the data that I reviewed. Was there a handful of teams with similar declines that skewed everything? Below is a graph showing the distribution of win total change. The Y (vertical) axis represents the number of teams who saw their win total increase or decrease by a certain number. The X (horizontal) axis shows the win total changes. So, for example, the blue line, which represents all bowl teams, shows that exactly 15 teams saw their win total decrease by 4 (-4) the following season.
The distributions for teams that won their bowl and overall bowl performance were quite similar while bowl losers saw a more even distribution. The most common result across all three groups was for a team to finish with one fewer or the same number of wins the year after their bowl performance. It was twice as likely that a team would see their win total drop by 2 or more wins as opposed to increasing by 2 or more wins.If you’re more a numbers guy, here’s the actual distributions:
Win Change | All Bowl Teams | Winning Teams | Losing Teams |
---|---|---|---|
-5 or worse | 18 | 11 | 7 |
-4 | 15 | 7 | 8 |
-3 | 24 | 8 | 16 |
-2 | 21 | 13 | 8 |
-1 | 36 | 23 | 13 |
0 | 33 | 18 | 15 |
1 | 25 | 10 | 15 |
2 | 18 | 7 | 11 |
3 | 9 | 5 | 4 |
4 | 9 | 2 | 7 |
+5 or more | 2 | 1 | 1 |
And as a percentage of results:
Win Change | All Bowl Teams | Winning Teams | Losing Teams |
< or = -2 | 37.14% | 37.14% | 37.14% |
-1 to 1 | 44.76% | 48.57% | 40.95% |
> or = 2 | 18.10% | 14.29% | 21.90% |
Win or lose the bowl, there was a slightly better than one in three chance that the same team would lose at least 2 more games the following season. Teams that lost their bowl game were slightly more likely to see a “turnaround” increasing their win total by 2 or more games, but there was only a one in five chance of doing so.
11 of the 210 teams reviewed (5%) saw their win total improve by 4 or more wins the following season. 7 of those 11 occurred in 2012 following the 2011 season. One of those teams, of course, was the Notre Dame team that went from 8-5 to 12-1. That result, however, was hardly the norm. By contrast, 33 teams saw their win total decrease by 4 or more wins. It was three times as likely that a team would be 4 games worse as it would that a team would get 4 games better. That leads to one more sub-set that Notre Dame fans may find interesting before we discuss what all of this means for the 2015 team. I decided, in honor of Brian Kelly having 3 seasons in which he’s finished 8-5 to look at all teams that won 8 games in a season, regardless of whether they won or lost their bowl, to see how those teams faired the following year:
Total Teams: | 45 |
Bowl Wins: | 20 |
Bowl Losses: | 25 |
Win %: | 61.33% |
Win % year+1: | 60.10% |
Change: | -1.23% |
The 2011 Notre Dame team who went 12-1 in 2012 was the most successful turnaround for any 8 win team in the years reviewed. It was one of just two teams to improve their win total by 4 the following year (2013 Boise State the other). It also saw the greatest improvement in win percentage for any 8-win team. By contrast, five of the 8 win teams saw their win total decrease by 4 or more games.
- What Does All of This Mean, Nerd????
I can’t tell you what all of this means. I try to present the numbers without spin and let you the reader reach your own conclusions. Additionally, no matter what I say, some people will inevitably make conclusions based off of what I’m presenting that is not actually what the data even pretends to say. I look forward to the “see, Brian Kelly sucks” arguments in the comments. That said, a few things I do think this lends credence to:
- Most ND fans knew it, but 2012 was an extraordinary and unusual season. For the many that felt like that season came out of nowhere, this data at minimum lets you know 2012 was a season swimming up stream. I do not subscribe to the theory that 2012 was an aberration, and I think that any argument about the Brian Kelly tenure that dismisses or ignores 2012 is fatally flawed. However, expecting those types of one year spikes to return is wishful thinking.
- Avoiding win declines is a sign of a successful program. Make no mistake, championships are always the goal, but avoiding declining win numbers is an accomplishment. At the moment, Brian Kelly has 2 seasons where his win total has diminished, one season where it’s improved, and one season were it’s remained constant. Just another way in which the 2015 season will be important for the Kelly regime.
- Don’t fall victim to believing a bowl win means anything. The numbers simply don’t support it. Making it to a bowl is the strongest indicator as most teams will at least maintain a record above .500. However, believing that a bowl win means anything more than that is fool’s gold. Seriously, just ask 2014 Oklahoma about big bowl wins. Again (for a third time because I guarantee you someone will take this the wrong way) winning the bowl is always preferred to losing. However, there’s no recent evidence to suggest that bowl wins are actual springboards to year over year success.
- Who the _______ am I Watching? ND’s Depth Chart (Literally) by the Numbers (Part II) - August 29, 2019
- Who the _______ am I Watching? ND’s Depth Chart (Literally) by the Numbers - August 27, 2019
- The People’s Free Guide to ND Football 2019 - August 26, 2019
Ryan Ritter
The biggest thing that I take out of the data is that it is really hard to win at college football these days. That goes to your point about successful programs being a measure of finding ways to not have their win total slide season after season.
Really eye opening data.
Paul Young
Winning at the FBS level is very difficult and is never a given. Never. That said, every year there are teams that for one reason or another, play over their collective heads one and due to rivalries, attrition and such, their fortunes in terms of wins and losses return to earth. I think many of those teams win their bowl games and in my opinion, many of those exceptional bowl teams (possibly those winning their bowl games) graduate or lose their players to the NFL. Let’s face it, that’s why many are there in the first place. The following year then, even though coaches hate the term, becomes a rebuilding year. Upper classmen are a luxury in today’s world of college football at the FBS level. The Irish are very fortunate heading into this season; their losses are minimal when you compare them to other contenders. And not to overhype their recruiting of recents years, but it certainly appears the pieces are in place for another run at a double digit win regular season.
LaughingTulkas
Most of this can simply be explained by regression to the mean. The average teams win an average amount of games. Because of FCS games, this average is greater than 50%. Teams that win above the average amount of games (including bowls) are more likely to win a number closer to the average the next year. Of course an average number of wins is different for all teams because of other factors, but overall the phenomenon described here is simply regression to the mean.
Moons
Thanks for reading. Certainly plays a role. The “mean” win percentage for all FBS teams this past season was 52.79%. The FCS wins don’t tip the scales that severely. Giving 7% back year over year is steep for just regression to the mean particularly because the values discussed are *not* isolated outliers but are already composite numbers. Teams will regress to true value, but that doesn’t necessarily mean 52.79%. Schools like Alabama, Notre Dame, Nebraska, Texas, have over 100 years of data with win percentages above 70%. Bigger point though was just noting that bowl wins (or losses really) don’t have an observable carry over effect. I don’t pretend to have the full answer for why that is. Your explanation is as good as any.
Bill Meehan
Regression to the mean is one way of putting it. I tend to look at it this way:
You have roughly 70 bowl teams per year. The following year, some of those teams aren’t going to make a bowl. There are going to be teams that did not make a bowl game that replace them. That’s pretty much where you get your decreased win % from. You can largely ignore the teams that are bowling again, because the distribution of wins among them is largely going to be a wash.
The other thing to keep note of when talking about the +/- 5 wins following year is the simple hard mathematics of it. You’re just going to have less positive increases as there are less games available to win.
I enjoy reading these win/loss breakdowns, but it’s important to keep in mind the other factors that can skew the raw data. I like to look at coaching changes as a big one. You’re going to have small programs who have a great year, and then have their coach poached. These programs tend to be the programs that “regress to the mean” the most. Next you’ve got the program who has opted not to retain their coach, these introduce a lot of variance. Sometimes the kids play up to send a coach out in style, sometimes they fold. Next you’ve got the program that just hired the new big name. It all depends on the state of the program, but these can definitely programs that experience an up-tick in wins the following year. If they’re coming from a non-bowl season, you can guess who they’re taking the wins from.
I’d be curious to take a look at this, not from win %, but from another angle. Of teams that retained their head coach, how many had a better regular season than last year, the same, and worse, then break that down by teams that didn’t go bowling, lost a bowl game, and won a bowl game.
zuskie
Nice article. 2 main things came to mind for me. 1) Coaching changes 2) QB changes.
I’m not sure there are 2 more important factors for performance impact in sports than coach & QB at the college level. I’d be curious to see how that data lines up with your analysis, though I’m sure you’ve burned out on data research on this topic.
Thanks for the analysis.
tbonesays
Poor teams don’t go to bowls and have the most room for improvement.