Friday, September 14, 2012

Back from the dead! My Power Ratings through week 2

Hi guys! I'm back from a hibernation over the last few weeks, I've tried starting to write a bunch of posts but it just didn't sound right. Today I'll be talking about my college football power ratings, and how I update them throughout the season.

      First off, I'm going to talk briefly about how I turn preseason power ratings into ones I update during the season. You may recall my preseason ratings that I posted here in late June. My spreadsheet that I update throughout the season can be found here. My first step in translating those ratings into ones I update during the season is to cube their original rating and then divide it by ten thousand (which I did in column M of my sheet). This may seem a little strange, but it separates out the teams more. Now, I get into how to determine how games make a team's rating change.  Heading into each game, both teams put one fourth of their rating into a "pot". The important part is determining what percentage of the pot each team takes out of the game. I should point out a few things about the percentage formula before I reveal it:
1) The winner always gets at least 55% of the pot
2) It's quite possible (and moderately frequent) for the winner of the game to lose points.
3) The sum of the opponents' ratings heading into the game is the same as the sum of their ratings after the game (unless a team is playing an FCS team).
So here's the formula:
(% of points scored)+(Win Bonus) with the limitations.
Win bonus means that you get an extra 5% of the pot for winning by a possession or less, and an extra 15% for winning by more than a possesion. The limitations are as follows:
Guarantee limitation: A team is guaranteed at least twice the amount of points they score in percentage, up to 20% (i.e. a team that loses 24-10 still gets 20% of the pot).
Zero limitation: A team that scores 0 points is guaranteed at least 2.5 times 20 minus the points the winning team scores. This means a team that loses 10-0 still gets 25% of the pot.
You can see this formula at work in the spreadsheet for week 1 in columns M-P (and for week 2 in columns Q-T). If you have any suggestion for it, please let me know. I'm considering adding a margin of victory component.

So, after you figure out how many points a team gets from the pot, you add it to the amount of points they didn't put into the pot, and voila! you have your rating for the next week. I should briefly mention that teams retain about 93% of their rating for beating an FCS team, and about 55% for losing to an FCS team.

Now, I'll share with you guys my top 25 teams in mypower ratings through week 2. Please note that these are not updated through the South Florida/Rutgers game last night. Below them are my power ratings through week 1 for archives' sake. If you would like to see the full power ratings for each week, go to the sheet of my spreadsheet labeled "weekly".

Through week 2:
1. Georgia 131.87
2. Oklahoma 122.73
3. Alabama 116.97
4. Texas 105.19
5. LSU 98.18
6. South Carolina 89.24
7. USC 88.81
8. Tennessee 88.47
9. Clemson 85.81
10. Oregon 85.49
11. Notre Dame
12. Nebraska
13. Florida
14. West Virginia
15. Ohio State
16. BYU
17. Ohio
18. Florida State
19. UCLA
20. Missouri
21. Texas Tech
22. Arkansas
23. Baylor
24. Utah
25. Michigan State

Through week 1:
1. Oklahoma 132.68
2. Georgia 121.77
3. Alabama 106.79
4. Texas 98.03
5. Tennessee 95.64
6. Nebraska 91.31
7. USC 88.63
8. LSU 85.46
9. Notre Dame 84.58
10. South Carolina 84.29
11. Clemson
12. Oregon
13. West Virginia
14. BYU
15. Florida
16. Missouri
17. Arkansas
18. Florida State
19. Utah
20. Ohio
21. Rutgers
22. Virginia Tech
23. Ohio State
24. Oklahoma State
25. Texas Tech

That's all for now folks!