The BBMI Weekly Review · Sept. 29 to Oct. 4, 2026
Inside a 30-20 Week: A Kneel-Down, a Pair of Failed Two-Point Tries and a Strike-Zone Challenge
Five BBMI picks this week were settled by half a point or half a run, three against us and two for us. Here is every result from the NFL, college football and MLB, and the season record for every product.
Minnesota had first-and-goal at the Miami 2, the game in hand, and the Dolphins had no timeouts left. The Vikings knelt. The U.S. Bank Stadium crowd had just given kicker Will Reichard a standing ovation for going 5-for-5, and then it booed its own offense for kneeling at the goal line. The final was 15-10. Minnesota is 4-0 and did not score a touchdown on Sunday.
Kneeling was the right call for Minnesota and the wrong result for us. BBMI’s premium spread pick was the Vikings minus 9.5 points; our own line had them winning by 12.3. A touchdown on that drive makes it 22-10 and the pick covers. A field goal makes it 18-10 and the pick still loses. The kneel-down left the margin at five.
Vikings minus 9.5: three ways to end the driveThat is what this column is for. BBMI forecasts games in the NFL, college football, college basketball, Major League Baseball and college baseball, and runs a paper-tracked WNBA model. Every pick is logged before kickoff, tipoff or first pitch and graded from the final score, and nothing is removed from the record. Each week this review goes through those results, including the ones that came down to a single play, whichever way it broke.
The week, by product
The live products went 30-20 this week. College football had a profitable Saturday, MLB’s playoff picks won 10 of 13, and NFL totals had a bad Sunday. One week is too small a sample to judge any of these products: NFL totals had seven picks, so a single result moves the weekly record by about 14 points of win rate. The season columns are the better guide.
| Product | This week | Week ROI | Season | Season ROI | Break-even |
|---|---|---|---|---|---|
| NFL spreads | 3-1 | +43.3% | 13-15 | -11.3% | 52.4% |
| NFL totals | 2-5 | -45.4% | 11-12 | -8.7% | 52.4% |
| NCAAF premium ATS | 15-11 | +10.2% | 42-31 | +9.9% | 52.4% |
| MLB F5 moneyline | 4-0 | +77.7% | 135-89-38 | +2.3% | 58.7% |
| MLB F5 run line | 1-0 | +90.9% | 126-100-1 | +6.4% | 52.4% |
| MLB F9 moneyline | 3-2 | +8.8% | 122-72 | +5.9% | 59.4% |
| MLB F9 run line | 2-1 | n/a* | 256-140 | +1.4% | 63.7% |
| NCAAB spreads, 2025-26 | No games | 257-138 | +24.3% | 52.4% | |
| NCAAB totals, 2025-26 | No games | 795-484 | +18.7% | 52.4% | |
| NCAA baseball spreads, 2026 | No games | 622-465 | +9.3% | 52.4% | |
| NCAA baseball totals, 2026 | No games | 108-65 | +19.2% | 52.4% | |
| Total, overall premium | 30-20 | +12.0%† | 2,487-1,611-39 | +12.3% |
How to read this table. Records are wins-losses, with a third number for pushes where a product has any. ROI is profit divided by the amount risked, one unit per pick. NFL, college football and the MLB first-five run line are graded at standard -110 juice; the MLB moneylines and full-game run line are graded at the price posted when the pick was logged. College basketball and college baseball spreads and totals are graded at -110. Break-even is the win rate needed to finish even at those prices; at -110 it is 52.4%. F5 means the first five innings and F9 the full game; the run line is plus or minus 0.5 runs on F5 and 1.5 on F9. *The published log does not print F9 run-line prices, so that product’s weekly ROI is not computed here; its season ROI uses the posted prices (median -184). †The weekly total covers the 47 picks with printed prices, about 5.7 units won, and leaves out the three F9 run-line picks. The season total adds each product’s units won or lost, about 509 units on 4,137 picks risked, and matches the overall premium record on the Model Performance page, which combines the same products. Every product in the table is live: logged before the game and graded after. Seasons run from the first live pick: May 23 for MLB, Sept. 4 for college football, Week 1 for NFL spreads and Week 2 for NFL totals. College basketball and college baseball are between seasons, so they had no picks this week; their rows show the completed 2025-26 college basketball season and the completed 2026 college baseball season. Each row is a product counted in that overall premium record. The college baseball moneyline is tracked on its own page but is not part of the overall record, so it is not in this table.
WNBA picks are a paper experiment, tracked without wagers until the model shows a credible edge, so they are not in the table or the totals. This week the paper spreads went 1-0 and the paper totals 1-2; for the season, spreads are 16-16 and totals 34-29. The WNBA model is judged on CLV, closing line value: the average number of points the market moved toward BBMI’s side between the moment the pick was logged and the closing line. Zero means the line did not move, and positive means the market moved toward BBMI. Season CLV is -0.20 points on spreads and +0.29 on totals, from picks logged since July 19.
Settled by half a point
at
Against us: Titans at Ravens, over 42.5
BBMI projected 49 total points, and this was one of the week’s three top-tier totals picks. Baltimore led 24-7 at halftime and did not score again. Lamar Jackson’s left ankle rolled under a Tennessee defender on a sack in the final seconds of the half, Tyler Huntley played the second half on a drizzly afternoon, and Tennessee turned it over on downs inside the Baltimore 5. Three points from that trip would have cleared the number. Instead, Cam Ward’s 3-yard touchdown pass to Wan’Dale Robinson with 1:12 left and a 2-point pass to Gunnar Helm made it 24-18. That is 42 points against a 42.5 total. Malaki Starks recovered the onside kick, and the over lost by half a point.
at
Against us: Utah State at Boise State, Boise State minus 19.5
BBMI had Boise State winning by 28.6. The Broncos led 31-18 in the fourth quarter when Maddux Madsen hit Ben Ford for an 11-yard touchdown at 8:27. Boise State went for two instead of kicking, and the try failed. Final: 37-18, a 19-point win against a 19.5-point line. An extra point would have covered.
at
Against us: Phillies at Braves, Wild Card Game 1, Phillies plus 1.5
Don Mattingly brought closer Jhoan Duran in during the seventh inning, and Duran allowed a three-run homer. Atlanta won 5-3. A two-run win beat the Phillies’ run and a half by half a run.
at
For us: North Texas at Tulsa, Tulsa plus 1.5
This one we will take, with a caveat. BBMI had Tulsa winning by 15.9, one of the largest edges on the college board. North Texas outgained Tulsa 559-352 and led 45-38 in overtime. Damari Alston scored to make it 45-44, and Tulsa went for two and the win. Oran Singleton Jr. was stopped inches short of the goal line, and the call was upheld on replay. North Texas won by one, and Tulsa plus 1.5 covered by half a point. The pick cashed; the forecast behind it was wrong about what kind of game this would be.
atFor us: Phillies at Braves, Wild Card Game 2, Braves plus 1.5
BBMI saw Game 2 as nearly even, Philadelphia by 0.03 runs, and took Atlanta with the run and a half. Atlanta led 3-1 into the eighth. Kyle Schwarber and Bryce Harper hit back-to-back homers off Dylan Lee to tie it, and Alec Bohm hit a two-out solo homer off Didier Fuentes in the 10th. Atlanta put two runners on in the bottom half, and the game ended on an automated strike-zone challenge, the first postseason game to end that way. Phillies 4-3: a one-run loss, and the Braves covered by half a run.
NFL, Week 4
Spreads went 3-1. Cleveland, taking 3 points at home, beat Pittsburgh outright 27-24. Chicago covered 3.5 in a 23-12 win over the Jets, and the Giants, getting 2.5, beat Arizona 36-24. Minnesota was the loss.
Totals went 2-5. Six of the seven picks were overs. New England and Buffalo combined for 55 points on a 49.5 total, and Arizona and New York for 60 on 44. Several of the over losses came down to a few plays. The Ravens game finished at 42 against 42.5 after Baltimore was shut out in the second half. The Colts and Commanders, playing in London, finished at 43 against 46.5, 3.5 points short. Green Bay and Tampa Bay finished at 31 against 39, and Green Bay had chances to add to it: the Packers wasted two opportunities to extend a three-point lead in the fourth quarter, both ending in turnovers, the second a Jordan Love interception in the red zone. A touchdown on one of those drives and a field goal on the other would have cleared the total. The Jets and Bears finished at 35 against 43.5.
The one under went the other way. BBMI projected 43.1 points for Kansas City at Las Vegas against a 47.5 total, and the Raiders led 13-10 after three quarters, 23 points in all. The fourth quarter produced 34. Kenneth Walker III’s 76-yard run at the end of the third set up Patrick Mahomes’ 8-yard touchdown pass to Tyquan Thornton, the first fourth-quarter points the Raiders’ defense had allowed this season. Matt Gay kicked two field goals to put Las Vegas back in front 19-17, Walker scored from 2 yards out, and a Raiders fumble set up another Kansas City touchdown for a 30-19 lead with 3:02 left. Kirk Cousins’ 20-yard touchdown pass to Brock Bowers and a 2-point conversion made the final 30-27: 57 points.
Chiefs at Raiders: under 47.5Seven picks make a small week. With one more point in Baltimore and two more Green Bay scoring drives, this would have been a 4-3 week instead of 2-5. Even an accurate totals model should expect weeks like this in the NFL, where each week offers a short list of games and a play or two can decide several of them. One week is a poor sample, which is why this review reports each product’s season record alongside the week.
+2 Green Bay scoring drives
For the season, spreads are 13-15 with an ROI of -11.3% and totals are 11-12 at -8.7%. Both are below break-even through the first month of the season. The full log is on the NFL Model Performance page.
College football, Week 5
The premium board went 15-11 on 26 picks, a +10.2% ROI. Wisconsin covered 8.5 in a 31-3 win over Michigan State, Nebraska covered 15.5 in a 48-23 win over Maryland, James Madison beat Marshall 45-17 laying 15.5, and Memphis won 59-8 at Charlotte laying 20.5.
The largest edges did not do the most work. The three biggest gaps between BBMI and the market went 1-2: Miami (OH), favored by 10.5 and by 28.9 in our numbers, lost 24-20 to Bowling Green; Illinois, favored by 9.5 and by 26.3 in our numbers, lost 24-17 to Purdue; and Tulsa covered by half a point. The worst forecast of the week was Eastern Michigan at Massachusetts. BBMI had Massachusetts by 18.6, and Eastern Michigan won 38-14, a miss of 42.6 points. Virginia at Florida State was next: BBMI had Virginia by 8.8, and Florida State won 38-7, a miss of 39.8.
For the season, premium picks are 42-31, 57.5%, with an ROI of +9.9% and 7.2 units won. In the 2025 walk-forward test, which applied the model to each 2025 game using only information available before kickoff, the same premium rule went 273-157, 63.5%, with an ROI of +21.3%. That is a backtest, not a live record. Combined with this season, it makes up the 503-pick, 62.6% record shown on the college football picks page. The full log is on the college football Model Performance page.
MLB, Wild Card round and Division Series Game 1
The first-five moneyline went 4-0: the Yankees in both games against Boston, each 2-0 after five innings; San Diego in Game 2 against the Cubs; and Atlanta in Game 3 at +105, which led 5-0 after five. The full-game moneyline went 3-2. San Diego won both of its Wild Card games, 8-0 and 4-1, and the Braves won Game 3 6-2 on a pick BBMI flagged as a Best Bet.
Yankees · WC Game 12-0 after fiveWYankees · WC Game 22-0 after fiveW
Padres · WC Game 2WBraves · WC Game 3 · +1055-0 after fiveWPadres · WC Game 18-0WPadres · WC Game 24-1WBraves · WC Game 3 · Best Bet6-2W
Astros · WC Game 1White Sox 6-3L
Guardians · ALDS Game 1White Sox 3-0LThe two losses had the same opponent. BBMI favored Houston in Game 1 of its Wild Card series, and the White Sox won 6-3. It favored Cleveland in Game 1 of the Division Series, and the White Sox won 3-0.
All four MLB products are ahead for the season: the first-five moneyline at +2.3%, the first-five run line at +6.4%, the full-game moneyline at +5.9% and the full-game run line at +1.4%. See the MLB Model Performance page, and the Bracket Pulse page for BBMI’s playoff odds.
WNBA, first round (paper)
The WNBA model is paper tracked, so it risks nothing and carries no ROI until it shows a credible edge; it is judged on closing line value. In Game 2 of Golden State against Dallas, BBMI took Dallas getting 4.5 points, and Dallas won by eight to force Game 3. The over in that game cashed; the BBMI log records 208 combined points against a 157.5 total. The overs in Game 3 (150 points against 159.5) and in Indiana at Las Vegas (177 against 181.5) did not. For the season, paper spreads are 16-16 with an average CLV of -0.20 points, and paper totals are 34-29 at +0.29. See the WNBA Model Performance page.
What BBMI does
For every game, BBMI builds its own number, a point spread, a total or a run margin, from team and player ratings and game simulations, then compares it with the betting market. When the gap clears a threshold set in advance, the model logs a pick before the game starts. The thresholds, every logged pick and every result are public on the Model Performance pages.
This week’s picks and forecasts: NFL picks, college football picks, MLB picks and WNBA picks. Rankings: NFL, college football, MLB and WNBA. Playoff odds: NFL Playoff Pulse, MLB Bracket Pulse and WNBA Playoff Pulse. Longer analysis is in Research.
The next review covers the Division Series, NFL Week 5 and college football’s Week 6.
Prediction first. Result second.
Sources and method. Pick records, prices, edges and results were read from the BBMI Model Performance pages on Oct. 4 and 5, 2026: NFL graded through Week 4’s Sunday night game, college football through Oct. 3, MLB through Oct. 3, with no MLB picks logged on Oct. 4, and the WNBA through Oct. 2. Weekly ROI for NFL and college football is the figure each page prints; weekly ROI for the MLB moneylines is computed from the posted prices in the log, one unit risked per pick. Game details come from the linked reports. Edge is the gap between BBMI’s line and the market’s line; premium picks are those that clear each product’s published threshold. All products except the WNBA are live: logged before the game and graded after. The one walk-forward figure in this review, college football’s 2025 test, is labeled where it appears and is not included in any live record.
Related research
Can You Beat Vegas Long-Term? The Scoreboard Through September 7
Since December 29 we have posted a model line on every game we cover, logged every pick before tipoff or first pitch, and graded it in public. Installment 1 of a running series. The hypothesis is not that a model beats Vegas on every game; it is that a cohort of games, defined in advance, can be profitable long-term. One live cohort clears the vig under a rule that never moved, a second is positive under a rule finalized in-season, one wins 63% and has not made a dollar, and two more are positive only in walk-forward replays a live season has not yet confirmed.
NCAAB premium +24.2% (257–138) · College baseball underdog ML +25.0% · MLB run line 62.9% at −1.4% · NCAAF 2026 7–1MethodologyEveryone Is Trying to Predict Sports. Almost Nobody Is Good At It.
The market we're up against is 99.7% calibrated and funded by an industry measured in hundreds of billions. Here's how close our models actually get across four sports, where we fall short, and why we publish every number — including the ones that aren't flattering.
Pinnacle r² 0.997 · MAE vs Vegas across four sportsMLB · Forecast ReviewChecking Our Wild Card Numbers: Three Favorites Advanced. Houston Went Home.
BBMI favored all four hosts. The Braves, Yankees and Padres moved on, while the White Sox swept our co-favorite Astros. Here’s how the round played out.
Favorites 3–1 · expected 2.33 · one Game 3 in four seriesSee it live
The live model behind this piece — the same numbers the article is built on.