We are one week closer to having a composite that doesn't rely on preseason rankings
We inch our way to full conference play with a few more games picked this week (58 FBS vs. FBS matchups in total). That means tighter spreads with fewer guaranteed games on the schedule between Power 5 and Group of 5 teams. But, of course, the SI Composite projects Georgia to beat Vanderbilt by 38 points, so there are still big spreads to be had if that's your angle to picking.
We also inch our way toward rating systems phasing out preseason priors. The Composite does not generate its own numbers soup-to-nuts, it simply averages some of the best predictive analytics systems around, but for an idea of what preseason priors can do, a tweet from ESPN's Bill Connelly is helpful. Connelly runs SP+, a wonderful analytic rating.
In Connelly’s actual rankings that include priors, those teams are, respectively are Michigan (6th), Georgia (2nd), Ole Miss (12th), Texas Tech (48th) and Florida (8th).
So preseason priors matter, and we can't throw them all the way out until we get enough of a sample size from this hopelessly small sample-sized sport. The hope for the SI Composite is that once more games are in and we’re relying on the data from this season, the wins come at a higher rate.
Get fantasy and betting analysis in your inbox by signing up for the Winners Club newsletter: SI.com/newsletters
Before the picks, a brief check-in with UConn (again), in the running to be one of the worst teams ever. The “good” news in the wonky macabre sense of the term we brought up last week is that UConn's rating dropped again to -31.63 despite the fact that they scored against an FBS opponent finally. The bad news is that Clemson’s composite rating actually fell to 19.85 too. That’ll happen when you struggle with Georgia Tech at home in a game the SI Composite got right (we’ve had some hard luck this season, gotta pat ourselves on the back for the wins when we get ‘em).
Last week: 24-28-0 46.15%
Season: 64-81-2 44.22%
Composite Best Bets
DOWNLOAD: PDF or VIEW AS WEBPAGE
The process behind the picks is briefly explained in my Week 1 column. Most of the time, the computer spits out something within one or two points of the real line (and at that point, sharp plays and whale plays are what’s moving the line anyway). But if the computer gives something more than three points, it’s worth a second look.
Favorites
-Minnesota (-31) over Bowling Green
-Boise State (-9) over Utah State
-Georgia (-34.5) over Vanderbilt
-Iowa (-23) over Colorado State
-Louisiana (-13.5) over Georgia Southern
-Buffalo (-13) over Old Dominion
-UNC (-12) over Georgia Tech
-Indiana (-9.5) over Western Kentucky
Dogs
-SMU (+10) over TCU
-Nebraska (+5) over Michigan State
-North Texas (+12) Louisiana
-Colorado (+14.5) over Arizona State
The Big Dogs
-UNLV (+30.5) over Fresno State
-Ohio (+15) over Northwestern
-Georgia State (+27) Auburn
-Rutgers (+18.5) over Michigan
-Tennessee (+20) over Florida
-Southern Miss (+45) over Alabama
-USF (+23) over BYU
More Betting:
• NFL Player Prop Bets: Week 3 Thursday Night Football
• NFL Week 3 Preview: Early Line Movement & Odds Tracking
• College Football Playoff Championship Betting Futures Breakdown