There were 39 bowl games played, not including the championship game. There would have been 40 but the First Responders Bowl was canceled. That's a lot of bowl games.
When the bowl season began, I started tracking the attendance for each bowl. I wanted to see how many bowls were played before mostly empty seats. As the bowl season progressed, I captured more data in an attempt to gauge how 'good' or 'bad' the game was. I didn't watch every game, and I didn't want to dig into every boxscore to come up with a rating, so I went with a simple approach:
- A large stadium mostly empty was 'bad'
- A smaller stadium mostly full was 'good'
- A game with a lot of points scored was 'exciting,' and therefore 'good'
- A game where a lot of points were scored *and* the margin of victory was small was 'more exciting' and therefore 'more good'
- Finally, a high-scoring and close game in a mostly full stadium was a 'good' bowl; a blowout in a mostly empty stadium was a 'bad' bowl.
That methodology has two glaring faults:
- The raw score of a game is not always a good measure of the quality of the game. The MSU vs. Oregon game is the best example: close game, but horrible in every way.
- It assumes the announced attendance figure represents actual attendance, and I doubt very much that's the case for most of the bowl games.
But I went forward anyway because it was fun to track the bowls and play with spreadsheets.
Note: right click and "open image in a new tab" to view in larger format. Images here seem to be scaled to fit without the option to simply click and enlarge. Maybe I'm missing something.
Notes About the Spreadsheet
- The red line represents the dividing line between pre-Christmas and post-Christmas games.
- The "First Responders" bowl is not represented because it was canceled.
- I did not include the championship game.
- Team record shown is after the bowl game.
- Column H divides the total points scored by both teams by the margin of victory. The higher the number, the better: it means a close game with a lot of offense.
- Column K divides the announced "official" attendance by the stadium capacity to yield a "percent full" for the stadium.
- Column L is a game "rating" value. Let's discuss that next ...
The Rating Value
What I was striving for was a gauge of how good a game it was combined with the size of the crowd relative to the stadium capacity. The 'goodness' of the game was indicated by column H (total points scored divided by margin of victory), and the 'atmosphere' of the venue provided by column K (the percent full). My thinking was: a good game in a full stadium is a great experience; a good game in an empty arena is a diminished experience.
So column L is simply column H (game quality) multiplied by column K (attendance percentage). The bigger the number, the better. A large number means a high-scoring game with a close winning margin in a stadium venue that was mostly full. This approach rewards bowls in smaller venues that fill the seats, and it takes away from bowls in huge stadiums that are mostly empty. It's not perfect, but that's what I went with.
Pre-Christmas vs Post Christmas Bowls
When thinking about the bowl games, there seems to be a divide between the "lesser bowls" and the "better bowls," and that divide is Christmas day. The spreadsheet has a line indicating that divide.
Here are the games played before Christmas, sorted by the game 'rating' value:
Note: right click and "open image in a new tab" to view in larger format.
Pre-Christmas 'best' bowl by this rating system: the Bahamas Bowl, where FIU beat Toledo 35-32. A close game in venue that was filled to 90% of capacity. However, it was a small venue to begin with, but maybe that's good: it was a 'lesser' game, so it didn't warrant a big stadium.
Pre-Christmas 'worst' bowl by this rating system: the Frisco Bowl, where Ohio beat up on San Diego State 27-0 in a half-full stadium that wasn't that large a venue to begin with. The game 'quality' math punishes a shutout game: total points (27) divided by margin (27) yields a quality of 1, which is the lowest possible given this math.
Here are the games played after Christmas, sorted by the game 'rating' value:
Note: right click and "open image in a new tab" to view in larger format.
Post-Christmas 'best' bowl by this rating system: the Alamo Bowl, where Washington State nipped Iowa State 28-26 in a stadium that was 84% occupied.
Post-Christmas 'worst' bowl by this rating system: the Belk Bowl, where Virginia's 28-0 spanking of South Carolina yielded a 0.6 game rating. Decent crowd (64% occupied), but the 'quality' math treats a shutout harshly, with the game quality number a meager 1.0.
Numbers Aside: Best and Worst Games
This is subjective, of course, and the comments can be used for feedback on reader feelings about best and worst games.
One game that surely must go in the 'worst' column is the MSU vs. Oregon game. Despite the math of this spreadsheet granting a game rating of 5.7, which was just below average for the post-Christmas games, the game was nearly unwatchable.
Best game? This one is tough because I didn't watch every game. My bias will be towards those I did watch. The Rose Bowl was entertaining. I found the Alabama/Oklahoma game interesting from a "can the Alabama machine be stopped?" perspective.
Column MIN, MAX, AVG, and MED
Here are the minimum, maximum, average, and median values for each column in the spreadsheet:
The games after Christmas were definitely better-attended, as expected. But the total points and margin-of-victory numbers are comparable. The early bowl games can provide entertaining football. Provided you care enough to watch.
Was there a correlation between the "game quality" value (column H) and other factors?
Note: this is really an exercise is creating charts and bitmaps and having fun with pictures. No serious statistical analysis was attempted here. :-)
Game Quality compared to Stadium Capacity
Did the bigger venues result in better games?
- Ohio 28, SDSU 0
- Washington State 28, Iowa State 26
- Baylor 45, Vanderbilt 38
If anything, there's a slight negative correlation.
Game Quality compared to Stadium Utilization
Did attendance as a percent of stadium capacity correlated with the quality rating?
- Marshall v South Florida in the 'Bad Boy Mowers Gasparilla Bowl'
- Ohio State v Washington in the Rose Bowl
A bit, but the correlation is very weak.
Game Quality compared to Latitude of Host City
I'm starting to have a little fun here. Latitude is a measure of how far north a city is, so I checked correlation of game quality against latitude:
A bit of a negative correlation. But I doubt that has anything to do with where the host city is located.
Here's a map of the bowl locations:
The distribution of the bowls location by latitude worked out like this:
- The Honolulu Bowl, which was the farthest south and the farthest west. The farthest east was the Bahamas Bowl.
- New York City, Detroit, and Boise. Boise was the farthest north bowl.
- Draw horizontal lines through New Orleans and Memphis ... 15 bowls were played in that band of latitude.
Cities with the most bowls: Orlando with 3, and the Dallas area with 3; Atlanta and New Orleans tied with 2. Florida is the most represented state for bowls with 8 bowls.
Game Quality compared to Character Length of Full Bowl Name
Now I'm really having some fun. The longest bowl name was the "Military Bowl Presented by Northrop Grumman" (43 characters, including spaces) on December 31st, between Cincinnati and Virginia Tech. It turned out to be a close affair: 35-31 Cincinnati over VT. The shortest name was the "Belk Bowl" (9 characters) on December 29th between Virginia and South Carolina. That was a shocking 28-0 blowout by Virginia. Was there an inverse correlation between the length of the bowl name the quality of the resulting game?
Wasn't really expecting one; didn't see one.
Game Quality compared to Random Number
Because, why not? I generated a random number between 0 and 1 for each bowl game and looked for a correlation:
That correlation line moves a fair amount with each generation of random numbers. I don't think 39 data points is enough to allow the 'random' numbers to yield a flat line every time.
Thoughts About this Exercise
- I freely and fully admit the game 'quality' and 'rating' values are somewhat arbitrary. I think the game 'quality' (total points divided by margin) is an okay measure of game excitement, but it has its flaws (see MSU vs. Oregon); the game rating where the quality rating is multiplied by the attendance percentage is not that useful.
- I wanted to find a simple-to-locate stat in the box score of each game to further refine the game quality value. In particular, I wanted to find something that would explain bad games that I could use to push down the rating of a game that received an otherwise not-bad rating number. I thought about turnover differential, number of punts ... even total penalty yards. But I gave up after trying to find out why Ohio beat SDSU 27-0. See the next section for that exercise.
- I expected to see lower attendance figures in general, but the announced numbers are not horrible. But then again, these are the "reported" attendance values. I suspect actual butts-in-seats was quite a bit less for some bowls.
- Some of the attendance numbers I pulled from Wikipedia right after the game concluded. It's amazing how quickly those pages are created and updated for these bowl games, with data filled in specific to the game that finished just minutes before.
- There's a lot of bowl games. 39 is a big number. The payouts for many of these games are not trivial amounts. I don't know how they make money, or even if they make money. The business case justification for some of these baffles me.
Ohio vs. SDSU Box Score
As mentioned, I was tempted to dig into the box scores and factor in other numbers to weigh the quality of the game. I thought about numbers such as: turnover differential, total yards differential, rushing yard differential, and third-down conversion differential. But I abandoned that when I dug into the 27-0 blowout of SDSU by Ohio in the Frisco Bowl. Based on the score alone, I figured Ohio dominated in some way, or SDSU committed a ton of turnovers or penalties. The box score didn't show anything obvious: there's a bit of an imbalance in total yards, but it's not horrible. The turnover numbers are even, and the third down conversion rate was comparable:
So I'm a bit puzzled why SDSU lost so badly. I didn't see the game, so I can't factor in qualitative things.
My conclusion: if the box score from a 27-0 blowout didn't have flashing red lights about the nature of the game, then other box scores might not either. Sometimes the numbers don't directly reflect what happened on the field. Rather than spend the time digging for a universal "X factor" stat, I went with my simple approach.
The Arizona Bowl
This was played on December 29th in Tucson, which is where I live. It pitted Arkansas State against Nevada. The final score was 16-13 in overtime. The announced attendance was 32,368, but in talking with someone who was there a more realistic number was somewhere around 10,000 ... maybe 15,000.
This gentleman (Mike Lude, a spry 96 years old: he was AD at Auburn and Washington back in the day) said the game featured two very bad football teams, with linemen for both teams very large and very slow. It was a cold day by Tucson standards (in the 40's), so I'm sure a lot of the locals stayed home. I'd be shocked if many Arkansas State fans traveled to Tucson for this. Nevada fans could drive down (8 hours), or fly (Southwest has direct flights), but I don't know if many bothered.
I can't figure out why the organizers of this scheduled it for the same day as the national semi-finals. I can't figure out why the NCAA allowed this game to be scheduled on that day. This bowl is better scheduled before Christmas. Or better still, not scheduled at all. If you want evidence there are too many bowl games, then look at this game. It's been in existence since 2015. Tucson used to host the "Copper Bowl," but that morphed into the "Cheez-It Bowl" and moved to Phoenix.
There are a lot of bowl games; some are well-attended, some are not; some are good, some are not-so-good; they happen year after year, so someone is making money, but how they make money is a mystery.
Hope springs eternal ... when does spring practice begin?