Thursday, September 1, 2011

College football week 1 computer predictions

One thing I thought would be fun (for me, at least) would be to try to come up with my own statistical model to use to try to predict the outcomes of college football games. A couple people on already do it, so I stole the idea from them and decided to give it a try this year.

Basically I compiled all the stats I felt were relevant, and then organized them and created ratings (relative to the average) for a handful of different categories. Then, using my own ideas of how they relate to each other, made a few formulas that I can use to plug those statistics in and then get a prediction for how those teams will fare against each other. Basically I do two sets of predictions and then average them. One of them looks at things from one team's offensive perspective and pits them against the defense's ratings, and the other looks at that same team's defensive stats and pits those against their opponents offensive ratings. From there, I add in home field advantage and my own top secret rating to account for strength of schedule, and out comes a prediction.

I will try to post these each Thursday in advance of that weekend's games. I'll probably only do around 20 games a week, mainly Big 12 games and other games of note, simply because it can be a little time consuming. And for the first few weeks, I'm forced to use last season's stats, which admittedly is somewhat useless. I am hoping that about 3 or 4 weeks in, each team will have enough current history to maybe make some semi-valid predictions. Until then, I'm using last year's numbers and with all the year-to-year turnover in college football, it renders these predictions pretty useless, but still fun. I expect that some of them will be pretty good, and some of them will be wildly wrong.

As you'll see, a few of these really jump out, and for some of them, there are obvious reasons. The Auburn score, for example, is obviously affected by the fact that Cam Newton was on the team last year. Everyone with a brain this year has rated Auburn to be worse than they were last year, so take this first prediction for what it's worth.

Some big numbers also pop up. Oklahoma State will supposedly score 72 points. That may sound crazy, but at the same time, that actually probably does happen once a week in college football, so it's not much of a stretch to say that OSU could do that this week. I actually kind of like the fact that my predictions include some huge blowouts and high scoring games, because those things do happen in real life, but people rarely predict them.

Anyway, all that being said, here is what came out of my computer for week 1:

Teams/Vegas line/My computer prediction

TCU -6/56       36
Baylor              24

SMU                26
A&M -16/56     36

UNLV               11
Wisc -35/56    36

Miss St -29      50
Memphis          12

Wake Forest      7
Syracuse -6     30

Utah St               16
Auburn -22/56    62

Akron                     6
Ohio St -34/48     47

Miami (OH)               9
Mizzou -17/47.5      31

Kent St                      6
Alabama -37/45.5    30

UCLA                     28
Houston -3/61.5     56

W. Michigan             34
Michigan -14/61       33

Minnesota               26
USC -21/51.5         47

S. Florida                   15
Notre Dame -10/47    19

BYU -3/55                  35
Mississippi                 30

S. Carolina -20/62      62
E. Carolina                 28

Rice                           19
t.u. -24/55                  48

Tulsa                         31
OU -24.5/64.5           58

Boise St -3.5/51         27
Georgia                      17

Oregon -3/55          32
LSU                         22

ULaLa                           18
Ok State -36.5/62.5       72

FAU                              10
Florida -35/46.5            35

No comments:

Post a Comment