G’s Exploration or How the GCR Works

Posted on by G Robert

Hello college football fans,

First the basics. There are 2 measures that when crunched together are the Total Score for a given team: Performance Score and SOS. The Performance Score is based on a few factors. 1) the outcome of the game – did the team win or lose, 2) the score differential – a sliding scale that balances giving a team credit for a big win while not rewarding blowouts, 3) level of the opponent – more on that in a second, and 4) location – slight advantage for road and neutral wins, slight penalty for home losses. A definition: level of the opponent is a large, sometimes unfair, categorization in 4 parts. They are, in order, Power 5 (ACC, Big 10, Big 12, Pac-12, SEC, and Notre Dame), the Group of 5 (American, Conference USA, MAC, Mountain West, Sun Belt, and other independents), the FCS (Division I non-bowl eligible conferences such as the Big South or Ivy, that can participate in the FCS playoff) and Other (Division II and NAIA which some FCS play). Playing and/or winning against higher levels provides a bonus/reward while playing and/or losing to a lower level school creates a discount/penalty.

The SOS also has a few factors. 1) the number of wins a team’s opponents’ have (this value changes constantly as the season progresses, but future opponent’s are not included in the calculation), 2) away games are harder than neutral are harder than home games, 3) level of opponent, and 4) opponent’s opponent’s SOS (not all 10 game winners are equal). Combine the Performance Score and the SOS. That creates a given team’s value which is what is ranked.

There are no other assumptions – no carry over from previous seasons – no “conventional wisdom” that team A will be good and team B won’t. Because of that, every team, all 256 of them from Abilene Christian to Youngtown State start the season tied for first (or last), even Auburn (Jason Willard – ha). Because of that and many teams playing below level the first couple weeks of the season (not to mention the Ivy league doesn’t even play a game until September 21), the GCR isn’t published until after week 4 (the 22nd). Otherwise, more than likely, the week 1 report would have Nevada (at least so far) on top because they won a road game against a higher level. Maybe that’s correct, but I do know the first few weeks look really weird. So that’s my choice, I guess, and I’m sticking with it. I will provide analysis each week – the only thing not published will be the actual ranking.

That’s it – please bring on the comments and questions – invite others to the page and let’s have some serious conversations about the greatest of all sports – college football!!!

Robert

3 Replies to “G’s Exploration or How the GCR Works”

  1. Robert,

    I did not know you had data input for level of competition depending on the team’s conference (Power 5, etc.). Wouldn’t the SOS and opponent’s SOS be a truer determination for level of competition? You are almost going against your mantra of not using a team’s reputation as a season starting point (except in this case it is conference reputation). Or is there not enough cross-pollination between the levels to get accurate data within one season?

    Carl

    1. Carl, that is a great question. The weight is very small assuming the higher level beats the lower level – it’s kind of like a very slight adjustment to the win total. If there were no level-related upsets, I could remove that factor and there would be practically no impact on the ranking. The bigger issue is the upset. The categorization of levels helps determine the bonus/penalty which is more impactful. Even with that, if I removed it, there would only be a shift of a few teams a couple of places in the ranking. It’s the lowest weight of any of the factors for both the Performance Score and the SOS.

Comments are closed.