The convergence is finally of a decent enough quality to start making reliable predictions for this season. It was probably good enough last week, but I was way too busy with other stuff last week and missed my window to run the necessary numbers. So here we are. As a reminder on what this all means and how to read it, the instructions are
HERE. Instead of just putting up a few of the matchups I'm going to go ahead and put up all of them so that I can reference them on my phone throughout the week. At the top are some of particular interest. The rest are below.
- Looks like the data suggests Oklahoma will cover the spread at TCU.
- NC St and Clemson will be much closer than Vegas thinks it will be.
- LSU will cover the spread
- Ohio St will stay undefeated
- And my Utes will win at home!
If you have comments or questions please let me know
@NateEMiller.
Away | Home | Vegas | Spread Proj. | Range |
Stanford | Arizona St | 2.5 | 8.2 | 6.9 |
Oklahoma | TCU | 7.5 | 17.4 | 7.0 |
Michigan | Michigan St | 7.0 | 1.2 | 4.7 |
NC St | Clemson | -16.5 | -0.8 | 3.6 |
Colorado | Washington | -15.5 | -13.0 | 1.0 |
Penn St | Indiana | 14.0 | 4.4 | 3.8 |
Miss. St | LSU | -6.5 | -15.1 | 2.5 |
Oregon | Washington St | -1.5 | -2.5 | 0.1 |
Ohio St | Purdue | 14.0 | 17.0 | 0.7 |
USC | Utah | -6.5 | -5.5 | 0.9 |
Away | Home | Vegas | Spread Proj. | Range |
Georgia St | Arkansas St | -14.0 | -12.8 | 1.7 |
Colorado St | Boise St | -23.5 | -18.9 | 2.4 |
Air Force | UNLV | 12.5 | -3.4 | 3.1 |
Tulsa | Arkansas | -6.0 | -5.6 | 2.7 |
Illinois | Wisconsin | -25.0 | -22.4 | 4.2 |
Maryland | Iowa | -11.5 | -10.2 | 3.7 |
Miami (Oh) | Army | -7.5 | -30.9 | 5.9 |
Auburn | Ole Miss | 4.0 | -15.5 | 1.7 |
Northwestern | Rutgers | 20.5 | 13.8 | 7.9 |
Buffalo | Toledo | -1.5 | 15.2 | 5.6 |
Cincinnati | Temple | -3.5 | 11.3 | 1.6 |
North Carolina | Syracuse | -10.0 | -32.2 | 1.6 |
Virginia | Duke | -7.5 | -13.7 | 2.9 |
Bowling Green | Ohio | -16.5 | -17.8 | 3.9 |
FAU | Marshall | 3.0 | -1.8 | 3.7 |
Utah St | Wyoming | 15.5 | 32.1 | 11.7 |
Charlotte | Middle Tenn | -17.0 | -14.9 | 2.8 |
W. Michigan | C.Michigan | 4.5 | 18.4 | 0.7 |
E. Michigan | Ball St | 3.0 | 3.9 | 4.3 |
Minnesota | Nebraska | -3.5 | 13.9 | 4.3 |
Wake Forest | Florida St | -10.5 | -0.2 | 2.1 |
Houston | Navy | 13.0 | 16.0 | 3.5 |
Alabama | Tennessee | 28.5 | 21.8 | 11.0 |
Colorado | Washington | -15.5 | -13.0 | 1.0 |
UTEP | Louisiana Tech | -23.5 | -30.4 | 8.0 |
Kansas | Texas Tech | -20.5 | -37.9 | 9.2 |
Louisiana | App. St | -24.5 | -31.0 | 12.1 |
Akron | Kent St | 4.0 | 6.9 | 0.6 |
CCU | MASS | -3.5 | 1.8 | 4.1 |
SMU | Tulane | -7.0 | -6.4 | 1.4 |
California | Oregon St | 6.5 | 5.3 | 1.8 |
Memphis | Missouri | -9.5 | -16.7 | 6.0 |
Ga.Southern | New Mexico St | 13.0 | 18.5 | 2.3 |
Connecticut | South Florida | -32.5 | -36.0 | 1.6 |
Texas St . | UL-Monroe | -11.0 | -17.2 | 0.8 |
UCF | East Carolina | 21.0 | 25.0 | 4.3 |
UTSA | Southern Miss | -17.0 | -9.0 | 2.1 |
Fresno St | New Mexico | 14.0 | 24.8 | 14.9 |
Rice | FIU | -- | -27.7 | 6.2 |
North Texas | UAB | 1.0 | -4.8 | 1.0 |
Vanderbilt | Kentucky | -11.0 | -23.8 | 10.2 |
Old Dominion | W.Kentucky | -5.0 | -4.4 | 1.5 |
Arizona | UCLA | -7.0 | -3.1 | 0.1 |
Nevada | Hawaii | -3.5 | -21.2 | 2.3 |
Largest Spread of the Week
Kansas | Texas Tech | -20.5 | -37.9 | 9.2 |
Tightest Spread of the Week
Wake Forest | Florida St | -10.5 | -0.2 | 2.1 |
Highest Regression Uncertainty of the Week
San Jose St | San Diego St | -27.5 | -28.2 | 18.3 |
Least Regression Uncertainty of the Week
Arizona | UCLA | -7.0 | -3.1 | 0.1 |
How is this different than IowaHawk's marble model?
ReplyDeleteConsiderably. I am using a set of metrics that I identified and feed into an iterative multi-regression algorithm that I designed to get to a converged set of rankings each week. Every game affects every teams ranking through the connections within the network of regressions. I have been working on building the math and the code since around 2010 but have made very few edits in the past couple years. The predictions here on this post come from taking the metrics and the regressions and attempting to extrapolate them by one week into the next matchups. Some weeks it does an amazing job. Some weeks it sucks royally.
Delete