Mo Knows Model Update: 9/9/2014

I accidentally blew my model up over the weekend and had to put in a few hours, late at night while watching Colorado State vs. Boise State, to fix it.  When I following got it to produce outputs, I wasn't pleased with the outputs: I want the outputs to quantify my gut and this didn't quite hit it right.  It had BYU in the top 10 and Ohio State a few spots ahead of Virginia Tech, just outside the top 25.  I still think that's probably too high for BYU, although that might be a reasonable ceiling given that BYU has a good chance at 11 wins, which would probably allow them finish in the top 10, depending on when that loss is, to whom, and by how much.

Here is the current top 25:

1	Auburn
2	Oklahoma
3	Florida State
4	Oregon
5	USC
6	UCLA
7	BYU
8	Alabama
9	Georgia
10	Ole Miss
11	Texas A&M
12	LSU
13	Arizona St
14	Notre Dame
15	Stanford
16	Louisville
17	Washington
18	UTSA
19	South Carolina
20	Missouri
21	Mississippi St
22	NIU
23	Baylor
24	Michigan St
25	Marshall

It's not perfect.  Obviously.  I would take Washington, UTSA, and Marshall out if I could and add Wisconsin, Virginia Tech, and probably Iowa.  I would move BYU down probably around 15 or so.  But it's an objective model so I don't want to tweak it.  I like Louisville and Mississippi State where they are, which is higher than they currently sit in the polls.  I think those two have solid teams not to be overlooked this season.  Baylor and Kansas State are two of the more overrated teams, according to my model.

The model has accurately predicted the winners in 84% of games thus far, which is a little low for this point of the season, given that there are really very few 50-50 matchups being played in the first two weeks of the season.  It has gone exactly 50% against the spread, which is lower than last season, when it hit at a 58% clip against the Vegas Wednesday night spread.  I started comparing my model to the spread on Tuesday this year, which may explain some of the difference...

In Week 1 it accurately predicted a few upsets: UTSA over Houston and Ohio over Kent State (hit the point spread exactly).  Of the 8 upsets in Week 1, it whiffed on 3 completely, predicting a bigger margin of victory than the point spread, hit 2 right on, and had 3 where it called for the underdog to cover.  If you knew which 8 games to pick, it would have been a beautiful weekend!

In Week 2, there were 7 upsets and it accurately predicted 3 of them.  In Week 3, it calls for 5 upsets: UTSA over OK State (as I've stated before, my model loves UTSA for some reason, so take that one with a grain of salt), Rutgers over Penn State, South Carolina over Georgia (has that as a Pick, basically, with a 0.3-point edge to the home team), Tulsa over FAU, and Georgia State over Air Force.  Overall, it likes the frontrunners this week: in 51 FBS vs. FBS games, it likes the favorite 36 times and the underdog only 15, 5 of which it likes the underdog straight up.

After the recalibration including week 2 results, BYU moved up to #7.  Looking ahead, it likes BYU in every game the rest of the season by at least 2 scores.  The Cal game is the only game where the point spread has gotten closer since the preseason predictions; in the rest of the games the point spread has only widened as BYU started 2-0 (unexpectedly according to the model) and most of their opponents have slid down in the rankings.

The biggest movers and shakers in the rankings since the preseason are:
Arizona: up 15 spots
California: up 21 spots
Fresno State: down 16 spots
Iowa State: down 26 spots
Michigan State: down 13 spots
NC State: up 14 spots
New Mexico State: up 15 spots
Northwestern: down 16 spots
Penn State: up 17 spots
Rice: down 13 spots
Rutgers: up 15 spots
South Carolina: down 13 spots
Temple: up 16 spots
Tennessee: up 23 spots
Texas: down 19 spots
Tulane: down 20 spots
UCF: down 21 spots
Louisiana-Lafayette: down 14 spots
Louisiana-Monroe: up 14 spots
Utah State: down 13 spots
Vanderbilt: down 20 spots
Washington State: down 19 spots
Western Kentucky: up 15 spots

As you can see, BYU plays 4 of the biggest movers, but only 1 of them is headed the right direction when it comes to strength of schedule.  In the final analysis, I'd rather see BYU 12-0 with a lower SOS than 11-1 with a stronger SOS.  Cal, Middle Tennessee, and Nevada are the only 3 to have moved up the rankings; the other 8 have dropped.

I'm curious to see how BYU responds this Thursday.  BYU usually waits until the fans start to believe before crushing their hearts.  I think they got enough people on the bandwagon with last weekend's thorough trouncing of Texas that the Houston game would qualify.  The model predicts a 20-point win, Vegas says 18.5...does Mo know something Vegas doesn't?  Based on my performance so far, it's a 50-50 proposition going with my model!