# Semi-Objective CFP Ranking System: 2018 Week 9

Submitted by The Maizer on October 25th, 2018 at 12:54 PM

The Preface

This is is the week 9 college football team rankings produced by a semi-objective model as described previously HERE. The model was inspired by Seth's post that proposed a point system to determine bowl eligibility. These rankings aim to be a relatively objective starting point from which to apply considerations such as the eye test, margin of victory, and head-to-head results.

The Disclaimer

In the past, I have had to edit this post a bunch of times to get the table to look right, so now that I can't edit the OP, this is probably going to be a disaster. I don't even know how to insert a table now. Anyone know how without manual html?

The Rules

• +3 points for a conference championship.
• +4 points for a win over a top 10 team.
• +3 points for a win over a top 25 team (not in top 10).
• +2 points for a win over a winning P5 team (not in the top 25).
• +1 point for a win over a winning G5 or a losing P5 team.
• +0 points for a win over a losing G5 or any FCS team.
• -1 point for a loss to a top 10 team.
• -2 points for a loss to a top 25 team (not in top 10).
• -3 points for a loss to a winning P5 team (not in top 25).
• -4 points for a loss to a losing P5 or any G5/FCS team.

Top 10 and top 25 status are determined by this ranking system and the model is solved iteratively until it converges. In scenarios where oscillatory states lead to a failure in convergence, the average points for oscillatory states are used.

The Results

This week there were two oscillatory states, thus the 0.5 point values. Since I don't know how to do tables now, I'm only listing the teams that had non-negative point values and a few other teams of interest.

1. Notre Dame (12.5 points)
2. LSU (12)
3. Clemson (11.5)
4. Alabama (11.5)
5. Michigan (9.5)
6. Kentucky (7.5)
7. Florida (7)
8. Texas (6)
9. Georgia (6)
10. Texas A&M (6)
11. Oklahoma (6)
12. Iowa (5.5)
13. Ohio State (5)
14. NC State (5)
15. Stanford (5)
16. Washington State (3)
17. Washington (3)
18. Utah (3)
19. UCF (2)
20. West Virginia (2)
21. Virginia (2)
22. USF (2)
23. Duke (1.5)
24. Wisconsin (1.5)
25. SDSU (1)
26. Miami (1)
27. Oregon (1)
28. USC (1)
29. Virginia Tech (1)
30. South Carolina (0.5)
31. Buffalo (0)
32. Georgia Southern (0
33. Texas Tech (0)
34. Penn State (0)
35. Mississippi State (0)
36. Maryland (-0.5)

41. Northwestern (-1)

52. Western Michigan (-3)

55. Michigan State (-3)

60. Indiana (-4.5)

87. SMU (-11)

122. Rutgers (-23)

I think this makes sense. It doesn't take into account point differentials, but I'm okay with that, as the goal in these games is binary: win or lose. How do you handle a tie? I would likely use, first, any head-to-head, then some sort of fancystats ranking.

This is key..because current rankings would probably drop LSU.  The biggest problem with using rankings early in the season is that teams are rarely ranked properly. Auburn and Miami (and Georgia) were top ten teams purely based on last season's performance. LSU beat all 3, and get credit for 3 top 5 wins although none of those teams at any point showed they deserved to be in the top 5.

It is based on the current ranking (not when they played) and it is the current ranking as given by this model (not AP or any other ranking).

What's Rutgers? Is there more than one now?

Seems weird to assign a +1 for beating a losing P5 team but -1 for losing to a top 10 team, especially with no accounting for road vs home or margin of victory. This model tells me it's better to beat Rutgers at home by 1 than lose to OSU on the road by 1 in 6 OT.

Well, it makes sense only in that this system seems to emulate the AP poll reasonably well.

If a team is ranked #15 and beats Rutgers at home by 1, they generally stay right about where they are.  If a team is ranked #15 and loses to Ohio State by 1, they will certainly drop 5 or more places.  No, it doesn't make sense.  Yes, that's how voters behave (and what fans have come to expect from the voters).

META-response:

The short answer is "no tables".

The GUI wipes out any table-formatting tags (and any other tags AFAIK), which drove me nuts when I went to post my first Big Expectations of this season. I ended up having to post images of the tables I had. Ugh.

Boo. Thanks for the response. I guess I could do images of chunks of my table as a slightly more palatable solution, but not stoked about that.

This overvalues polls and benefits teams that play over-ranked teams at the beginning. Devalues wins against programs that are not properly ranked.

Can you explain how? This model completely ignores polls and at-the-time rankings. It only uses its own rankings for determination of top 25 and top 10 and starts fresh every week.

are you able to determine possible points remaining for the top 15 squads?

The tricky part is that as top 10 and top 25 teams move into and out of those positions, it changes things a lot.

I like the Notre Dame penalty.

No Conference Championship? You're going down.

I think this model needs to do a better job of accounting for home vs. road wins/losses. It is much tougher to win on the road in college than is being shown in these rankings. I would propose a +1/-1 to the point level for road vs. home. If you beat a top-10 team on the road, then +5; if you lose to top-10 team on the road, +0. Middle of the road power-5 conference teams who are between 25-40 are still tough to beat on the road, especially when the difference between a team ranked between 15-25 and 30-40 is not all that much.

Overall, I think these rating system makes sense and does a good job showing how much bias there is in the polls and how we value the play of certain teams.

I'm not opposed to the idea of including a factor for home/away, but if you apply your +1/-1 strategy across all categories, mathematically you're just going to get every team's points modified by +(# of away games - # of home games).

That would make it negligible and relatively pointless. I guess the main thing I think we're missing in this model is the stuff like PSU beating OSU at home, at night in 2016. If that game had been at OSU, I think we all agree that OSU most likely wins it (and probably by multiple scores).

An away game like @Indiana should be a relatively easy win for a top-15 team, but playing a team like OSU or PSU or ND at home vs. on the road makes a big difference and I'm not sure that this systems gives enough credit to a road win vs. a top-25 team.

Probably a better way to address it is +4 for a road top 25 win and +5 for a road top 10 win with no change to loss penalties. That gives teams with huge road wins a boost without shaking up the rest of the formula.

Maizer, did you ever see, back in the past decade, the College Football News rating system?

The College Football News website ranged from awful to kind of good, but their rating system was a good first pass at figuring out who had the best season, and they used it for interesting things like "best team of the decade" ratings.

The system was pretty similar to yours, but with a few added twists; I think I have it copied down somewhere on my computer--I couldn't find it on the internet.