If someone could get this forum post to a member of the playoff committee before rankings come out tomorrow, that'd be great. If not, I suppose I'll just have to live with the burden of knowing the truth while our college football overlords sleep soundly this week on a bed of #HotTakes. I suppose I'll let you share in my suffering at the hands of their little pet project, and release my much anticipated week nine FBS rankings.

Fed up with the constant bias and subjective reasoning that leads to overrated teams and biased feedback loops of inflated rankings **cough*SEC*cough**, I set out to create an objective system of ranking every team on the only statistic that matters at the end of the day. Did you win the damn game?

The premise is simple, but explaining can be hard. I'll do my best, though!

Opponent Value:

Each team on your schedule is given an opponent value. This value is based on your opponent's record over their last twelve games. If a team on your schedule is 10-2 over their last twelve games, beating them will earn you a score of +10 for that week, and losing to them will give you a score of -2.

If you play an FCS team, your win value is divided by two, and your loss value is multiplied by two. Playing a 10-2 FCS team would net you +5 for a win and -4 for a loss.

Scoring:

For each team you play, the value you earn for that win or loss is determined by two variables, that team's value at the time of the win, and that team's value in the current week. Beating a team in week one with a prior record of 10-2 returns a value of +10. In week nine, if that team now has a prior record of 8-4 for an opponent value of +8, those scores will be averaged for a score of +9. So, in week nine, your score will return as +9 for beating that team in week one rather than the +10 it'd been originally. If that team continues to lose, your win score will continue to decay. If that team improves, your score for beating that team will improve.

Your opponent score for each game you've played is then added up and divided by the number of games played for an average total score.

Love it? Hate it? Neat. Here you go...

• LETS GOOO! Ohio State reclaims the #1 spot over TCU, who briefly ended the Buckeye's strangle hold on the top spot since week four.
• Can we finally stop assuming Georgia is the far and away favorite for the #1 spot? Yes, their win over Oregon was great, but outside of that, they've mostly played nobodies. However, a couple of those nobodies have the letters "SEC" stitched to their chests and might've been national contenders in the mid to late 2000's, so hey, that's a quality win right? RIGHT?
• TTUN has gained some ground, but is still seen as the least proven undefeated team of the bunch.
• As much as I love seeing OSU in that #1 spot, it could be short lived. This week we play Northwestern with an opponent value of +1/-11. Beating the Wildcats would hurt our overall average score of +6.75 by throwing a measly +1 into the mix, only one point better than losing to Clemson (the only team with a current opponent score of +12/-0).

For the Nerds

Hey, if you like statistics and football, you probably had an opinion of my scoring system on last weeks post. A subject that was critiqued in last weeks rankings was the impact that last season's games have on this seasons rankings. Some like to argue that last years results have zero bearing on anything to do with this season. I understand the sentiment, but tend to disagree. For starters, there's more rollover than turnover from one season to the next in terms of returning players and coaches. Pretending each season exists in a vacuum is important when it comes to narrative bias, but I'd expect much more fluidity in numbers. Sure, some teams are going to have larger swings than others, but over the large data set that is 131 FBS football teams, I'd have to imagine those things level out.

Now, in an attempt to keep my rankings as purely objective as possible, I was reluctant to make a decision that could alter the impact of last years results on this years rankings. As soon as you start weighting metrics, you're making arbitrary subjective decisions on the value of one metric over another often to fit whatever value gives the results that best align to what you wanted to find in the data. However, I realized that by making a non-decision, I was still making a decision.

So, I made two more ranking systems. The original is weighted 1:1 in terms of averaging a team's opponent value at the time of the game (what we thought we knew) with their value in the current week (what we've learned). I decided to make tables that weight these metrics 1:2 and 1:3 because who am I to say that the value of these variables should be weighted 1:1.

• By weighting these metrics at 1:2 and 1:3 we place greater emphasis on what your opponent has done this season rather than last season.
• 8-0 TTUN was able to jump 7-1 Oregon for the 6th spot.
• Kansas State was originally below Oklahoma State even after demolishing them this week. With a greater emphasis on SoS this season we see these teams flip and pull away from each other quite a bit.
• Mississippi State earns a ton of forgiveness, moving up eight spots.
• I've been using our Notre Dame win as a scoring example, so I'll continue that here. Weighted 1:1, that win currently nets us a +9. Weighted 1:2 and 1:3, that win nets us a +8.67 and +8.5 respectively.

How much do my rankings factor last seasons games into this seasons rankings? I decided to create a table that shows what the impact would be on the opponent score for each game in the final rankings of a twelve game season.

In my 1:1 system, your points earned for game one would be 50% reliant on your opponents games from the previous year. Now, my argument for using these games to determine a week one score is that they impact the perception of your opponent at the time of the game, influencing the psychology of each team. It also provides an anchor point for your score if your opponent suffers injuries or has their season derailed for any reason that doesn't reflect the true state of the your opponent at the time you played.

So, the question I'm asking myself is this; is a 50% impact too much? I'm actually leaning towards yes. Looking at the impact of last seasons games on my rankings over the course of an entire season, the 1:1 model will have 27.08% of my final ranking based on games from last year and 62.92% based on the actual results from this season. That seems a little too extreme, and while it's a subjective call, I think the 1:3 model might be the better ranking system.

To be clear, the initial rankings posted above are still from the 1:1 system to provide continuity from last week. Movement in the rankings doesn't mean much if you're switching up the criteria. However, from this point on we could follow the 1:2 or 1:3 model and use this week as our starting point.