🏈 Ken's CFB Model

Dashboard Weekly Picks My Bets Win Totals Surprises

2026 season

Backtested edges this model actually has (walk-forward, 2014–2025): spreads vs early-week lines at edge β‰₯3, weeks 4+ (53.6%, p=0.009); totals weeks 2–3 (week 1: no picks β€” no data yet). Flat 1 unit β€” bigger edges did not win more in backtesting. Break-even is 52.4% at βˆ’110. Run picks early in the week; apply your injury/news read before betting.

This week's picks β†’

Top 25 projected win totals

Click a column header to sort.

How to read this table
β–² / β–Ό flag
The market model's lean vs the Vegas number (see the Surprises page), shown when it's 0.75+ wins: β–² lean over, β–Ό lean under. Hover for the number. Validated against eight years of real Vegas totals: flagged teams missed by 2+ wins in the arrow's direction 28% of the time (average team 19%) and the wrong way only 11%.
Proj rating
Projected team strength on the SP+ scale: points better (+) or worse (βˆ’) than an average FBS team on a neutral field. +27 means a 27-point favorite over an average team; 0 is exactly average; βˆ’20 is a 20-point underdog to an average team. Built from last year's final SP+, returning production, portal production gained/lost, and the 4-year recruiting-class average.
SOS
Strength of schedule: the average Proj rating of this team's scheduled opponents. Same scale as Proj rating, so higher = harder: +14 means a typical opponent is a top-15-caliber team; βˆ’5 means a typical opponent is a bit below average. FCS opponents count at about βˆ’25.
Mean wins
Average regular-season wins across 10,000 simulated seasons of the team's actual schedule (bowls and playoff excluded). Driven by Proj rating vs. each opponent's rating plus home field.
Picked
Record if the model's favorite wins every single game β€” click a team name for the game-by-game picks. Runs higher than Mean wins for good teams and lower for bad ones (a team favored 60/40 five times wins about three, not five). Use Mean wins for the Vegas comparison.
10th–90th
The range of wins the team landed in 80% of the simulations. A 7–11 range means 10% of seasons finished below 7 wins and 10% above 11.
Games
Regular-season games currently on the schedule that the model can simulate (most teams 12; a few show 10–11 where an opponent isn't yet in the model, e.g. a new FBS program or an unannounced game).
Vegas
Sportsbook consensus season win total (over/under), as of the date noted in the project status.
Edge
Mean wins minus Vegas. Positive = model leans OVER, negative = model leans UNDER. Edges beyond about Β±1.5 are worth a look; check what the model can't see (QB situations, coaching changes, injuries) before betting.
#TeamProj rating SOS10th–90th Mean winsPickedVegasEdge
1Texas Tech 23.9 -2.5 10–12 10.9 12-0 10.5 +0.4
2Miami 21.2 -0.2 9–12 10.1 11-1 10.5 -0.4
3Notre Dame 20.1 0.4 9–12 10.1 12-0 11.5 -1.4
4Oregon 25.2 5.4 8–11 9.8 11-1 10.5 -0.7
5Utah 17.5 1.0 8–11 9.6 12-0 8.5 +1.1
6Ohio State 26.4 9.2 8–11 9.4 12-0 9.5 -0.1
7Georgia 22.9 7.6 7–11 9.3 12-0 9.5 -0.2
8Indiana β–Ό 21.1 5.6 7–11 9.3 11-1 10.5 -1.2
9Texas A&M 19.9 6.0 7–10 8.7 11-1 8.5 +0.2
10Penn State 14.7 3.9 7–10 8.6 10-2 8.5 +0.1
11Toledo 4.4 -10.3 7–10 8.4 11-0 7.5 +0.9
12South Florida 3.2 -7.0 6–10 8.4 10-2 8.5 -0.1
13James Madison β–Ό 2.2 -7.7 6–10 8.3 11-1 8.5 -0.2
14BYU 12.4 2.6 6–10 8.1 10-2 8.5 -0.4
15Louisville 11.3 1.6 6–10 8.1 11-1 8.5 -0.4
16SMU 8.8 -0.7 6–10 8.1 9-3 8.5 -0.4
17Virginia 7.3 -1.4 6–10 8.0 10-2 7.5 +0.5
18Ole Miss 18.4 7.7 6–10 8.0 9-3 7.5 +0.5
19Iowa 14.2 4.4 6–10 8.0 9-3 7.5 +0.5
20USC 16.3 7.4 6–10 7.9 8-4 8.5 -0.6
21Oklahoma 18.1 9.3 6–10 7.9 10-2 8.5 -0.6
22North Texas 3.2 -5.4 6–10 7.9 8-4 5.5 +2.4
23Memphis 2.2 -6.3 6–10 7.9 9-3 7.5 +0.4
24Texas β–Ό 19.1 12.1 6–10 7.8 10-2 9.5 -1.7
25East Carolina 1.5 -6.4 6–10 7.8 10-2 7.5 +0.3

All 136 teams β†’