Thanks a lot to everyone for the feedback and commentary on last week's article regarding FSU's relative value to the ACC. There were some pretty interesting side conversations that emerged in the comments section, and on a few other sites within the ACC/SB Nation world. Over the next couple of weeks I'll try to address some of them as time permits.
One of the comments that came up a few times was readers' desire to see more information about each team's variance relative to expected TV viewership. Many fans believe that this may be another good way to try to estimate a team's relative value in the ACC since it should show the 'bump' that the most popular teams provide to less popular teams when the two play.
I learned how to use 'The Jump' this time, so click below to read more and see the outcomes...
What I mean by a team's variance is simple but here's an example to clarify:
- Maryland, on average, had 958k TV viewers per game for its ACC regular season games in 2011 based on estimates I developed with the regressions/analysis explained in the last post
- Similarly, Miami had 2.3m average TV viewers per game for the ACC regular season
- Maryland's first game of 2011 was against Miami (which should be a larger than average draw based on Miami's traditional strength)
- The total estimated TV viewership for the Miami vs. Maryland game was 4.4m people based on the fact that it was a night game shown nationally on ESPN as part of the season kickoff over labor day weekend
- For Maryland, the ‘bump' they got from playing Miami could be seen as 4.4m - 958k = 3.4m additional viewers, or roughly 360% of their average season game
- For Miami fans, on the other hand, the game was big but not all that amazing compared to their ‘average' game. The bump they got from playing Maryland on prime national TV could be seen as 4.4m - 2.3m = 2.1m additional viewers, or roughly 91% more than their average ACC game in 2011.
Hopefully that approach makes sense, but feel free to ask questions or provide feedback in the comments section. I basically replicated this analysis for each of the teams in the ACC in order to estimate the average ‘bump' that they provided to other teams in terms of TV viewers. A few caveats about the data, again:
- This only accounts for men's regular season ACC football games
- This is based on one year of data (2011) since that's all the ratings information I've been able to find
- Non-conference games were removed because it would be much harder to estimate the variance since I don't have data sets for any other conferences
- I used actual TV ratings where available, but for most of the games I had to use the estimated TV viewership based on the regression that I developed and explained in the last post
On to the fun stuff.... Here's how the data shook out:
The X (horizontal) axis represents each team's total TV viewership for ACC games in 2011. For example, GT had roughly 20m total viewers for all of its ACC games in 2011 and Boston College had a little over 5m viewers throughout the course of their season.
The Y axis represents the additional ‘bump' that teams provided to their opponent's fan bases when the teams played. It can be viewed as a proxy for popularity. For example, when a team played Florida State, roughly twice as many fans as average (110% increase) would tune in relative to expectations. On the other hand, when a team played UVA it was viewed by the other team's fans as a very average game - a few more fans might tune in than normal, but it didn't really move the needle. It goes without saying that teams lower down the Y axis (Duke, Wake Forest, etc.) actually discourage fans from tuning in to watch TV games.
Here's a table with the same data, in case you prefer seeing it this way:
For the most part, the results from the variance analysis confirm the findings from the estimated valuation analysis, but here are some key points I took away:
- I now understand why ESPN pushes so hard for tough non-conference games for the ACC. Using just the ACC vs. ACC games you can see that there is a consolidated tier at the top with FSU, Miami, Clemson, GT, and VT. I would bet that over time neither team really distinguishes itself from the rest in ACC vs. ACC play. The big differences start to emerge through non-conference games and the national appeal that FSU, Clemson, and Miami bring to the table.
- There is a huge gap between the top tier (FSU, Miami, Clemson, VT, and GT) and everyone else in terms of football impact on TV ratings.
- I was surprised to see that GT managed to have a positive pull on so many games. This may have something to do with the larger Atlanta TV market, or the timing of when they played their games (i.e. better match ups later in the season so weren't competing with MLB playoffs, etc.). Or, they may just be a bigger TV draw for ACC football games than I expected.
- Teams like NC State, UVA, UNC, and Maryland probably tread water each year, but the rest of the league is essentially relying on the top tier teams to bring in ACC viewership.
- FSU played a very boring ACC schedule in 2011 (Wake, BC, Duke, Maryland, UVA, etc) and therefore had a lower overall draw than some of the other ACC teams. Again, this highlights the importance of non-conference games for FSU to retain its national television draw. Note: this doesn't mean that under the current BCS format I'd advocate playing a tough non-conference schedule, but it does show that if FSU plays a bunch of cupcakes every year it is likely that its TV viewership will stay lower than its regional competitors (UGA, UF, Bama, Auburn).
What about you guys? What did you take away from this? Any other questions or comments you'd like to discuss? As always, I'm open to any feedback you have. Thanks.
Note: I have additional data on each of the teams, but based on some of the earlier comments I'm going to adjust it a bit and re-post it later. Check back in another day or so.