Oh, yeah - there was some pretty good college football, too!
App State almost App State-d Penn State (wait until we get to 2007!), forcing the Nittany Lions to complete a TD drive in the last 2 minutes to force OT and then get the win.
Army almost App State-d Oklahoma. Army's D did a serious number on Kyler Murray and the mighty OU offense and the Sooners needed OT to put Army away.
All Big12, all the time:
OU 59 / WV 56
TX 48 / OU 45
OU 48 / OkSt 47
OSU 52 / MD 51 (honorary Big12)
WV 42 / TX 41
And of course, the most Big12 game of the year:
Texas A&M 74
LSU 72
7OT
That was my second favorite game of 2018 after the OSU beatdown of UM. I watched every minute. It was absolutely amazing.
But enough about me. How about a little Ray's Retro Rankings? Here are the rules:
- I stop all data at the conference championships. No bowl games are included, as the idea is to provide us with a top 25 that we would have selected playoff teams from.
- All calculations are exactly the same as the current Ray's Rankings calculations.
- I will introduce the top 25 in reverse order
- I will compare RR/CFP/AP rankings
- I will provide 3 different playoff scenarios
- 4 team seeded
- 8 team with conference champions and 2 at large, seeded
- 16 team seeded
- We can all have a good laugh at how different things would have been. Or we can all stare slack-jawed in amazement at how closely Ray's Rankings matches human polls.
And here...we...go...
Bottom 5 had pretty pedestrian seasons with maybe the exception of Syracuse, who had the distinction of being 2018's North Carolina - they almost beat Clemson.
Kentucky. Who'd have thunk it? This is also the season that really set Northwestern's expectations so high for 2019: B1G West champs, played not poorly in the B1G championship game, beat Utah in their bowl game. Not really sure what happened to 2019. And who can forget the 7OT thriller between A&M and LSU? I loved every minute of that game!
WV had an interesting year. They put up a lot of points and almost beat OU. UCF was riding a long winning streak dating back to 2017. Florida, Texas, and LSU all played almost exactly to their levels, though I thought for a while that Florida was going to break through.
Chris Petersen took over a UW team in 2014 that had lost its identity. Don James was long gone, and the carousel of coaches had failed to right the ship. Enter the Boise State magician, and almost over night UW became relevant again. 10-3 may not be spectacular, but the Huskies still won the Pac12
So how about differences between the CFP and Ray's Rankings? Take a look below: the Diff column is the variance between CFP and RR; a positive number means the CFP ranked the team higher, and a negative number means they ranked the team lower than RR.
Some were close, some were not. I think it's pretty clear that analytics only carries so much weight with the committee. The ever ephemeral "eye test" plays a huge role, which obviously I hate. There are teams whose body of work (as determined by the algorithm, not some gut feeling) get slighted while others who may be less deserving ride reputation and bias to their advantage.
AP?
Again, some close, some not so much. Voters are a fickle bunch.
For the CFP, we have a variance of 9.4. Variance is the square of the deviation from the mean, averaged. Why square the deviation? Some deviations are positive and some are negative and it gets hard to figure true variance using both. When we square the deviation, we get all positive numbers. Why do we care? Well, we then take the square root to get our Standard Deviation (bet you weren't planning on a lesson in stats, were you?). SD is nothing more than the measure of spread around the mean (average). The SD here is 3.07.
What does it all mean (sorry)?
The average difference between RR and CFP is +1, so on average the CFP ranks teams 1 spot higher than RR. But that doesn't tell anything like the whole story. Around 68% of the differences will be found 3 places higher or 3 places lower than +1. Really that's not too bad. For the AP, the variance was 31.9(!) and the SD was 5.65 on a mean of 2 places higher. Because the voters "just know," right?
So the question is: where are these variances. If you look right below, you can see that the CFP ranked the SEC schools an average of 2 places higher than the computer, and ranked the B1G schools an average of 3 places lower than the computer. ACC was ranked about 1 spot higher while the Pac12 and Big12 were ranked on a level with RR.
CFP
AP
Interesting to me is how the B1G is still ranked so much lower than the computer, but the SEC is closer than it was with the CFP. I would guess that that means the voters are better at assessing SEC teams than B1G teams.
If we had used a real, objective ranking system for the 4 team playoff in 2018, we would have had:
Not bad games, swap out OU for OSU and off we go. I don't think anyone beats Clemson last year based on the numbers.
We're looking at the Avg column now. That is the average rank of the team in 1st Level, 2nd Level, and SRS scores. Clemson is head and shoulders above the rest.
What about an 8 team playoff? Keep in mind that my version of the 8 team is: 5 Power 5 conference champs, the top Group of 5 team, and 2 at large (highest ranked). All teams are seeded:
I'm no artist, I copy and paste parts of my spreadsheet. Sorry. Anyway, here are some great games. UCF gets their shot, and we get more football! Home games for the higher seed (UGA at ND in Dec?!?) in the first round. Also, this would probably force ND into a conference.
16 teams? No problem. This is a straight seeded 16 team bracket:
I did move some teams to avoid in-conference matchups in the first round. Are there any bad games here? I mean, Clemson probably still rolls, but this is way more fun than waiting around on Tuesday nights or the final Sunday for a group of people with an agenda (not the 4 best, but the 4 most popular who can be justified) to pick their favorites.
2018 was a pretty good year for college football. But it could have been better.
My final rankings for 2018. YOU decide if my system works:
No comments:
Post a Comment