This Positive Expected Value series will ask questions to those involved in sports analytics, sports investing or people who have experience identifying +EV investments.
This week features Bill Connelly, author of Football Study Hall, a fantastic read that is a must for anyone interested in football analytics and how it is applied for teams, coaches and fans. His work is trailblazing for college football and it was a real treat to get to talk to him and learn about how he got his start and how he sees the game of College Football changing and the role analytics plays through it all.
How did you get your start in football analytics and writing?
I started a Missouri blog in 2007 when I was looking for content to look at in the [college football] offseason. I like baseball stats, and being a Pittsburgh fan, went down the road of trying to look five years into the future to see when the Pirates would be good. I used that as a base of knowledge and looked around and found nothing existed for football. Nobody was really doing play-by-play level analysis, so I just started tinkering around and figured out tricks and began to record games and was able to break that down. Once I got a bigger base of knowledge I connected with Football Outsiders in 2008 and started that fall and it slowly picked up and started Football Study Hall within the SBNation network in 2011.
What helps you the most in your process?
I started back in 2007, so then it was learning everything Excel and Access could do and then just getting good at communicating what the numbers are saying. Working with Football Outsiders for three years before doing it full-time, helped me hone the writing side of things by getting reps and figuring out what I was and wasn’t communicating well. Keeping your eye on that feedback of how you are communicating is important, even if it is negative, it can help in the development. One of the reasons I enjoy what [Dr.] Bob does is because he has obviously gotten practice at communicating a lot of information in-depth and it shows.
How has the sport of college football evolved since you began covering it?
More teams are using tempo, so there is now a solid range of teams where you can see, who does and doesn’t do it well. In the late 2000’s especially when you had the Sam Bradford, Graham Harrell, and Chase Daniel, you had teams who were moving quickly because they had an advantage. Now tempo is more of a mindset and some teams do it much better than others. Beyond that, we are further along in the lifecycle of the spread offense, so a base college football defense is more nickel. Teams, in general, are smaller and faster than they were before. I started writing for Football Outsiders in 2008 which was in the middle of that 2006 to 2010 range, which was when the spread really coalesced and took off. We are now in a different part of the cycle where teams have adapted. We have seen how defenses are reacting and now get to see how offenses will react to that. Some teams, like Baylor, have figured out how to go even faster, but it has also opened up the door for teams to go with bigger more plodding offenses to take advantage of what they see on defense.
Alabama’s reaction has been telling, back when Nick Saban asked, “is this what we want college football to be?”, because teams were getting faster and faster and it seemed like the consensus was yes, so Saban said, “OK, I’ll beat you playing that way” and he introduced more tempo into his offense and his defense got simpler was able to adapt on the fly better – by being less exotic. The ones that are good can still find an advantage from tempo, though if you don’t find an advantage then teams can backtrack [to more traditional offenses] to gain an advantage.
Which of your five factors do you find most predictive for future games?
The more I dial into it, the efficiency metrics, which I took from the pro-side of Football Outsiders and tweaked to make more college specific, is the most telling stat. Big plays and the amount you gain on successful plays are semi-random and have less of a correlation than the success rate itself.
Which programs are now at the forefront of analytics?
What they are doing is hard to figure out. Programs can do what they are already doing, only faster. In that way, it is like the introduction of DVD to film sessions. Bob Davies used to talk about his first job in the 70’s splicing film for study and now they have a grad assistant who can do what used to take hours in 10 minutes and it has just sped everything up so they can move on to other things (edit: an anecdote which is detailed in his fantastic book, Football Study Hall). Teams, in theory, can move further toward more analytics and theory because the process has sped up.
However, I have also talked to a college coach recently about this topic and asked, “If you are able to do these processes much quicker how are you able to go above and beyond?”. His answer was telling in that it is tricky because you only get so much time with the players. They are only supposed to get 20 hours a week with them [laughs] , so you can’t make things much more complicated from the player standpoint.
What are the underutilized techniques in football analytics?
When you see how quickly hudl has emerged. That side of talent identification and general evaluation. The recruiting angle is an area that stats will be used more and more, you can now chart tons of football. You can now figure out better ways to evaluate prospects and chart camp data. Only so much can be done from a theory and tendencies standpoint, we can still go further, but we are closer to the limits in those areas.
The acquisition techniques are areas of opportunities for the non-blue blood teams to find an advantage, which is the same reason the spread offense came into existence. The success in team sports comes down to three things: talent acquisition, talent development, and talent deployment. We always think about the deployment side of things, but the acquisition and how you develop the players is just as important. Coaches are starting to figure out the development side with health and tracking, but the recruiting/acquisition side is still limited. These teams which don’t necessarily get the top-10 recruiting classes can make up for the lack of talent through these other areas. Whether it was Virginia Tech for awhile, TCU, Boise State, or more recently Michigan State, Oregon, Louisville. In general, there will always be a list of teams that figure out ways to succeed without top-10 talent. The why will change from time to time, and any variable you introduce is an opportunity to take advantage of an inefficiency not everyone knows about.
What is the biggest misconception about college football?
As we talk to more coaches, I realize within a normal week framework, which seems like a straight-forward process, is vastly different. Teams are recording different [types of data]. Sunday through Friday is vastly different from coach to coach. While the calendar may be uniform for each team, the different ways in which they go about their work and evaluations caught me off guard. For example, I am a Missouri guy and Pinkel was focused on day-to-day process and the daily checklist and they have and the ‘the Process’ from Nick Saban. Some people are set in routines and other are looser in that regard.