G’s Exploration or Welcome to the GCR

If you are reading this, then you must be as crazy about college football as I am. I hope this site and my analytical meanderings will provide insight, drive questions, and stir conversation both within its “four walls” and elsewhere. Before we get into what you will see in this site, a little about me and how the GCR came into being. I grew up in South Carolina, for the most part, and have always lived in the south. I can’t remember a time I didn’t like football, but I was too small to play competitively. As I got older, I followed not only the human vote rankings and was always bothered when they weren’t the same. Now with almost all of the major games on some TV channel, the voters could, perhaps with enough playing at the same time, watch all of them – not likely. In fact, this leads to what is sometimes called “the west coast bias” – Pac-12 games are often played late (Eastern time) and the voters don’t even try to follow the game live.

Imagine how difficult it must have been when there were only 3 network stations – or reliance on radio. Add to that the ridiculous notion of preseason polls (yes, I went bold there and will explain in a second), which are mostly a “how did the teams do last year combined with new head coach/quarterback mental math” situations. The reason I really don’t like preseason polls of any kind are the natural bias it sets up for the rest of the year. Here’s an example: Iowa State just had back to back 8-5 seasons and have only been ranked at the end of a season twice (1976, 2000). They’ve only played in 14 bowls (never a January one) and played 3 years in a row only once (2000-2002). They are ranked 37 to start this season (you have to read really deep in the “others receiving votes” section. What would have to happen for them to be #1? Let’s say Clemson wins the rest of their games by less than 10 (I know, the hurt they put on Georgia Tech shows that it’s not likely, but go with me here) AND Iowa State runs the table averaging 20 point wins. Is that good enough? Not likely – history has shown only a few instances where a top-ranked team loses that spot after a win – even a bad win. For the Cyclones to make the playoff? A bunch of teams would need to lose. This will be a bold statement hard to prove unless it actually happens, but if they go 12-1 they don’t make the playoff if the SEC, ACC, and Big 10 have a 0- or 1- loss team. The presumption of goodness inherent in the preseason guess poll sets up elitism. Ohio State fans may disagree, but they and Alabama and Clemson get some forgiveness that most teams don’t get (of course forgiveness can only go so far with a 3-score loss to an unranked team).

Next came the computer rankings and I was in love. Theoretically, the bias is gone (okay, in some cases that’s true), the preseason crud is gone (okay, in some of them), and fairness prevails (okay, most of the time). While there are dozens out there, a handful were chosen to be part of the BCS experiment. What could be better? They even had rules: the biggest was point differential had to be ignored (I’ll explain how I incorporate that in a later blog). The controversies were over!! No more dual championships (re 1990)! Such clarity! Oh wait, 2 SEC teams. SEC wins every year! Oooh, clearly the system must be wrong – drop the computers – they are the culprits. AARGH!

Around the time all of this was going on, a colleague of mine questioned the actual strength of schedule for the SEC (he’s an ACC fan or at least mostly). I wanted to supply wisdom on the inanity of such a question, but I had no facts. Surely, it was obvious. Wait, maybe it isn’t. How could I calculate who played the most difficult schedules and who had it easy? This led me to the first part of the GCR created – SOS (or at least the first part of it). My next blog will be about assumptions and factors in the model, but suffice it to say now that if team A plays a team (pick a conference) that had 10 wins during the season and team B plays a team (same conference) that had 4 wins, team A had a tougher schedule that week. There are other factors too (future blog), but that was my logic and it showed Kentucky had the most difficult schedule. Their adjusted average was roughly a 10-game winner every week – ridiculous. They were 2-10 that season. So I calculated, in hindsight, that the Wildcats ran a tougher gauntlet than anyone else. So what? After the season is over, who really cares? So, I had to figure out real time as well as other adjustments (opponent’s opponent’s SOS for example). But just knowing a team’s SOS tells us very little. I had to figure out a way to determine how they fared with the games they had.

Finally (and just so you know, it took about 6 months to write the first working model – I’m no genius here), I had something I could use. The first year, I kept it to myself (very limited beta group of one). Then I started emailing to a small group, which has grown, but is still pretty limited. Now, 6-7 years later, I’m finally blogging!! Each year, I add something new (like the blog), usually because of questions my limited audience. I’m hoping opening this site will provide a) a larger audience which leads to b) additional ideas for stats or analysis which leads to c) a much more fun and interactive experience for all of us.

This weekend, I’ll post two things (assuming I can figure out how) – first, the annual explanation of calculations (without providing formulas – sorry, figure out your own and let’s compare) and assumptions and second, the only preseason thing I’ve ever done (see above for how much I hate preseason anything) which is, based on last year’s win totals, the top 25 most difficult schedules going into 2019.

see you soon,

Robert