860–670–5 · 1,535 selections · 7.3% ROI
Week 4 / Game totals
Week 4 totals
recommendations.
The current card has 39 plays across the 58 FBS games that currently have both a spread and total. The model reads each scoring matchup in context, using team form, opponent strength, expected pace, game conditions and the market.
The model
A full-game scoring forecast.
The projection weighs how both teams create and prevent points, then accounts for likely game flow, venue, weather and the betting market. It estimates the final combined score and passes when the available edge is too small.
In the expanded 2017–2025 replay, average error falls from 12.81 to 12.78 points across 6,439 games, with improvement in 8 of nine seasons. That is a modest improvement in accuracy. The model learns from 2015–2025 games; this season’s completed games update team form.
Recommendations use three fixed layers. The point model acts first. On its passes, the probability and point directions must disagree before the arbitration can add a play. A probability that clears the −110 gate wins that disagreement; otherwise the point side needs at least a 0.1-point edge. The over policy then fills its cumulative 17.5% quota from the strongest remaining over edges.
Sequential replay · 2021–2025
How these selection rules performed.
Each historical season was forecast from earlier seasons. The figures below use the same model and all three selection layers on this card, risking one unit per bet at −110. They are historical research after model development, not a live betting record.
436–329–3 · 768 selections where core passed
140–102–4 · 246 selections where production passed
1436–1101–12 · 2,549 selections · 8.0% ROI
Season-by-season results and the earlier-data comparison
The promoted policy includes core picks, disagreement-arbitration additions, and the frozen over-additions quota. Results use historical consensus totals; the displayed books on today’s card are a separate price snapshot. Pushes return the stake.
| Season | Selections | W–L–P | Units |
|---|---|---|---|
| 2021 | 510 | 299–211–0 | +60.8 |
| 2022 | 534 | 295–234–5 | +34.2 |
| 2023 | 471 | 262–208–1 | +30.2 |
| 2024 | 479 | 279–199–1 | +54.6 |
| 2025 | 555 | 301–249–5 | +24.6 |
Both total models were also trained on the same expanded history for the comparison below. The 2015–2016 training rows use market spreads where saved spread-model forecasts are unavailable. Point forecasts improve in each 2017–2020 season; early betting returns are mixed.
| Season | Games | Previous model, retrained | Updated model |
|---|---|---|---|
| 2017 | 735 | 13.773 | 13.756 |
| 2018 | 733 | 12.897 | 12.881 |
| 2019 | 734 | 13.080 | 12.949 |
| 2020 | 508 | 13.775 | 13.749 |
| 2021 | 732 | 12.394 | 12.377 |
| 2022 | 734 | 11.886 | 11.891 |
| 2023 | 750 | 12.774 | 12.763 |
| 2024 | 751 | 12.685 | 12.651 |
| 2025 | 762 | 12.314 | 12.309 |
The current card
39 recommendations across 58 games.
Central Michigan at Miami
- Final-total model
- 55.6
- Consensus total
- 54.0
- Model edge
- +1.6
- Regulation model
- 55.9
- Probability cross-check
- Supports · P(over) 55.5%
- Expected possessions
- 11.24
- Weather adjustment
- 0.0
South Alabama at Kentucky
- Final-total model
- 53.2
- Consensus total
- 54.5
- Model edge
- −1.3
- Regulation model
- 52.1
- Probability cross-check
- Supports · P(under) 53.6%
- Expected possessions
- 11.30
- Weather adjustment
- 0.0
Virginia Tech at Boston College
- Final-total model
- 43.3
- Consensus total
- 47.5
- Model edge
- −4.2
- Regulation model
- 43.1
- Probability cross-check
- Supports · P(under) 61.0%
- Expected possessions
- 11.43
- Weather adjustment
- −3.5
Colorado State at UTSA
- Final-total model
- 55.8
- Consensus total
- 58.5
- Model edge
- −2.7
- Regulation model
- 55.7
- Probability cross-check
- Supports · P(under) 57.1%
- Expected possessions
- 11.69
- Weather adjustment
- 0.0
App State at NC State
- Final-total model
- 54.2
- Consensus total
- 55.3
- Model edge
- −1.0
- Regulation model
- 53.3
- Probability cross-check
- Supports · P(under) 53.4%
- Expected possessions
- 11.48
- Weather adjustment
- 0.0
South Carolina at Alabama
- Final-total model
- 52.7
- Consensus total
- 53.8
- Model edge
- −1.1
- Regulation model
- 51.8
- Probability cross-check
- Supports · P(under) 53.5%
- Expected possessions
- 11.46
- Weather adjustment
- 0.0
Hawai'i at Wyoming
- Final-total model
- 45.4
- Consensus total
- 43.8
- Model edge
- +1.7
- Regulation model
- 44.9
- Probability cross-check
- Supports · P(over) 55.0%
- Expected possessions
- 11.47
- Weather adjustment
- 0.0
Navy at UAB
- Final-total model
- 48.7
- Consensus total
- 51.3
- Model edge
- −2.6
- Regulation model
- 48.8
- Probability cross-check
- Supports · P(under) 57.4%
- Expected possessions
- 11.01
- Weather adjustment
- 0.0
Ole Miss at Florida
- Final-total model
- 57.2
- Consensus total
- 59.3
- Model edge
- −2.0
- Regulation model
- 56.9
- Probability cross-check
- Supports · P(under) 56.1%
- Expected possessions
- 11.33
- Weather adjustment
- 0.0
Notre Dame at Purdue
- Final-total model
- 56.7
- Consensus total
- 58.3
- Model edge
- −1.5
- Regulation model
- 57.9
- Probability cross-check
- Supports · P(under) 54.8%
- Expected possessions
- 11.41
- Weather adjustment
- 0.0
Illinois at Ohio State
- Final-total model
- 52.4
- Consensus total
- 53.8
- Model edge
- −1.3
- Regulation model
- 53.7
- Probability cross-check
- Supports · P(under) 54.3%
- Expected possessions
- 10.76
- Weather adjustment
- 0.0
Army at Temple
- Final-total model
- 46.0
- Consensus total
- 48.3
- Model edge
- −2.2
- Regulation model
- 46.1
- Probability cross-check
- Supports · P(under) 56.6%
- Expected possessions
- 10.42
- Weather adjustment
- −1.4
Oregon at USC
- Final-total model
- 59.8
- Consensus total
- 62.3
- Model edge
- −2.4
- Regulation model
- 59.5
- Probability cross-check
- Supports · P(under) 57.0%
- Expected possessions
- 10.95
- Weather adjustment
- 0.0
UCLA at Maryland
- Final-total model
- 55.3
- Consensus total
- 56.8
- Model edge
- −1.5
- Regulation model
- 54.6
- Probability cross-check
- Supports · P(under) 54.6%
- Expected possessions
- 11.12
- Weather adjustment
- −0.4
Wake Forest at Louisville
- Final-total model
- 54.7
- Consensus total
- 56.8
- Model edge
- −2.1
- Regulation model
- 55.8
- Probability cross-check
- Supports · P(under) 56.3%
- Expected possessions
- 11.84
- Weather adjustment
- 0.0
Nebraska at Michigan State
- Final-total model
- 49.9
- Consensus total
- 48.8
- Model edge
- +1.1
- Regulation model
- 49.3
- Probability cross-check
- Supports · P(over) 53.6%
- Expected possessions
- 11.23
- Weather adjustment
- 0.0
Clemson at California
- Final-total model
- 52.2
- Consensus total
- 50.8
- Model edge
- +1.4
- Regulation model
- 50.7
- Probability cross-check
- Supports · P(over) 54.4%
- Expected possessions
- 11.77
- Weather adjustment
- 0.0
Liberty at Coastal Carolina
- Final-total model
- 47.8
- Consensus total
- 49.8
- Model edge
- −1.9
- Regulation model
- 48.3
- Probability cross-check
- Supports · P(under) 55.8%
- Expected possessions
- 11.55
- Weather adjustment
- −1.5
Oklahoma State at West Virginia
- Final-total model
- 59.6
- Consensus total
- 60.5
- Model edge
- −0.9
- Regulation model
- 60.6
- Probability cross-check
- Supports · P(under) 52.3%
- Expected possessions
- 11.64
- Weather adjustment
- 0.0
Missouri at Mississippi State
- Final-total model
- 56.8
- Consensus total
- 58.5
- Model edge
- −1.7
- Regulation model
- 57.2
- Probability cross-check
- Supports · P(under) 54.6%
- Expected possessions
- 11.57
- Weather adjustment
- 0.0
Massachusetts at Sacramento State
- Final-total model
- 47.5
- Consensus total
- 48.5
- Model edge
- −1.0
- Regulation model
- 47.6
- Probability cross-check
- Supports · P(under) 52.8%
- Expected possessions
- 11.38
- Weather adjustment
- 0.0
Air Force at Nevada
- Final-total model
- 48.8
- Consensus total
- 49.5
- Model edge
- −0.8
- Regulation model
- 48.5
- Probability cross-check
- Supports · P(under) 52.0%
- Expected possessions
- 10.66
- Weather adjustment
- 0.0
UConn at Miami (OH)
- Final-total model
- 50.2
- Consensus total
- 51.5
- Model edge
- −1.3
- Regulation model
- 50.6
- Probability cross-check
- Supports · P(under) 53.6%
- Expected possessions
- 11.46
- Weather adjustment
- 0.0
Vanderbilt at Auburn
- Final-total model
- 53.6
- Consensus total
- 55.3
- Model edge
- −1.6
- Regulation model
- 55.1
- Probability cross-check
- Supports · P(under) 55.0%
- Expected possessions
- 11.15
- Weather adjustment
- 0.0
Sam Houston at Texas Tech
- Final-total model
- 56.4
- Consensus total
- 55.8
- Model edge
- +0.7
- Regulation model
- 57.4
- Arbitration check
- Probability wins · P(over) 52.4%
- Expected possessions
- 12.04
- Weather adjustment
- −0.1
San Diego State at Toledo
- Final-total model
- 50.2
- Consensus total
- 51.5
- Model edge
- −1.3
- Regulation model
- 50.7
- Arbitration check
- Point fallback · P(under) 53.4%
- Expected possessions
- 11.65
- Weather adjustment
- 0.0
Arizona at Washington State
- Final-total model
- 50.0
- Consensus total
- 48.5
- Model edge
- +1.5
- Regulation model
- 48.4
- Arbitration check
- Point fallback · P(over) 54.0%
- Expected possessions
- 11.40
- Weather adjustment
- 0.0
South Florida at Bowling Green
- Final-total model
- 50.8
- Consensus total
- 49.3
- Model edge
- +1.5
- Regulation model
- 49.3
- Arbitration check
- Point fallback · P(over) 54.8%
- Expected possessions
- 11.82
- Weather adjustment
- 0.0
UNLV at Akron
- Final-total model
- 54.0
- Consensus total
- 52.8
- Model edge
- +1.3
- Regulation model
- 52.2
- Arbitration check
- Point fallback · P(over) 54.1%
- Expected possessions
- 11.67
- Weather adjustment
- 0.0
Middle Tennessee at Jacksonville State
- Final-total model
- 50.4
- Consensus total
- 51.5
- Model edge
- −1.1
- Regulation model
- 50.4
- Arbitration check
- Point fallback · P(under) 52.8%
- Expected possessions
- 11.56
- Weather adjustment
- 0.0
Rice at Fresno State
- Final-total model
- 45.8
- Consensus total
- 44.3
- Model edge
- +1.6
- Regulation model
- 44.0
- Arbitration check
- Point fallback · P(over) 54.9%
- Expected possessions
- 11.52
- Weather adjustment
- 0.0
Tulsa at Arkansas
- Final-total model
- 52.1
- Consensus total
- 50.8
- Model edge
- +1.4
- Regulation model
- 50.8
- Probability cross-check
- Supports · P(over) 54.4%
- Expected possessions
- 11.72
- Weather adjustment
- 0.0
Georgia Tech at Stanford
- Final-total model
- 48.5
- Consensus total
- 47.8
- Model edge
- +0.7
- Regulation model
- 47.1
- Probability cross-check
- Supports · P(over) 52.5%
- Expected possessions
- 11.24
- Weather adjustment
- 0.0
Kennesaw State at Arkansas State
- Final-total model
- 51.8
- Consensus total
- 52.8
- Model edge
- −1.0
- Regulation model
- 53.1
- Arbitration check
- Point fallback · P(under) 53.3%
- Expected possessions
- 11.70
- Weather adjustment
- 0.0
Troy at Utah State
- Final-total model
- 48.6
- Consensus total
- 47.3
- Model edge
- +1.4
- Regulation model
- 47.9
- Probability cross-check
- Supports · P(over) 54.3%
- Expected possessions
- 11.89
- Weather adjustment
- 0.0
Colorado at Baylor
- Final-total model
- 55.0
- Consensus total
- 55.5
- Model edge
- −0.6
- Regulation model
- 55.5
- Arbitration check
- Point fallback · P(under) 51.5%
- Expected possessions
- 11.63
- Weather adjustment
- 0.0
Texas at Tennessee
- Final-total model
- 54.6
- Consensus total
- 55.5
- Model edge
- −0.9
- Regulation model
- 55.7
- Arbitration check
- Point fallback · P(under) 52.3%
- Expected possessions
- 11.57
- Weather adjustment
- 0.0
Kansas State at Cincinnati
- Final-total model
- 54.9
- Consensus total
- 55.5
- Model edge
- −0.6
- Regulation model
- 54.7
- Arbitration check
- Point fallback · P(under) 51.5%
- Expected possessions
- 11.23
- Weather adjustment
- 0.0
Boise State at Western Michigan
- Final-total model
- 49.4
- Consensus total
- 49.8
- Model edge
- −0.4
- Regulation model
- 49.4
- Arbitration check
- Point fallback · P(under) 51.7%
- Expected possessions
- 11.47
- Weather adjustment
- 0.0
Every prediction
The 58-game board.
Passing is a prediction too. The table retains every lined FBS–FBS game, including the 19 games where all three layers pass.
Read before betting
A prediction is not a price guarantee.
Totals can move quickly, and a half-point can change both the edge and the recommendation. This page reflects a saved market snapshot, assumes standard −110 pricing, and does not account for limits or availability. Available CFBD forecasts are treated as the realized game conditions and translated through the historical weather response. Forecasts can change, so recheck the named book, price and weather before acting. Over additions use the explicitly displayed saved consensus total.
See the current spread board