Zoltar is my NFL football prediction computer program. It uses a neural network and a form of quasi-reinforcement learning. Here are Zoltar’s predictions for week #1 of the 2026 season (which starts tomorrow, Wednesday, Sept. 9).

These predictions are tentative, because they’re based on early point spread data, and also because that it usually takes Zoltar about four weeks to hit his stride.

Zoltar:    seahawks  by    3  opp =    patriots    | Vegas:    seahawks  by  5.5
Zoltar: fortyniners  by    0  opp =        rams    | Vegas:        rams  by  3.5
Zoltar:       bears  by    0  opp =    panthers    | Vegas:       bears  by  2.5
Zoltar:  buccaneers  by    0  opp =     bengals    | Vegas:     bengals  by  3.5
Zoltar:       colts  by    3  opp =      ravens    | Vegas:      ravens  by  3.5
Zoltar:       lions  by    6  opp =      saints    | Vegas:       lions  by  7.5
Zoltar:      texans  by    3  opp =       bills    | Vegas:       bills  by  1.5
Zoltar:     jaguars  by   10  opp =      browns    | Vegas:     jaguars  by  7.5
Zoltar:      titans  by    3  opp =        jets    | Vegas:      titans  by  2.5
Zoltar:    steelers  by    6  opp =     falcons    | Vegas:    steelers  by  2.5
Zoltar:     vikings  by    2  opp =     packers    | Vegas:     packers  by  1.5
Zoltar:      eagles  by    8  opp =  commanders    | Vegas:      eagles  by  4.5
Zoltar:    dolphins  by    0  opp =     raiders    | Vegas:     raiders  by  3.5
Zoltar:    chargers  by   10  opp =   cardinals    | Vegas:    chargers  by 10.5
Zoltar:     cowboys  by    0  opp =      giants    | Vegas:     cowboys  by  1.5
Zoltar:     broncos  by    4  opp =      chiefs    | Vegas:      chiefs  by  2.5

Zoltar theoretically suggests betting when the Vegas line is “significantly” different from Zoltar’s prediction. In mid-season I usually use 4.5 points difference. For the first few weeks of the season, I usually go more aggressive and use 3.5 points difference as the advice criterion. My strategy is based on empirical results rather than some sort of hypothesis.

At the beginning of the season, because of Zoltar’s initialization algorithm, Zoltar is strongly biased towards Vegas underdogs.

Using a 3.5 point difference threshold (when the difference of opinion between Zoltar and Vegas is strictly greater than 3.5 points), Zoltar has three opinions:

ravens       at        colts: Bet on Vegas underdog colts
bills        at       texans: Bet on Vegas underdog texans
broncos      at       chiefs: Bet on Vegas underdog broncos

A bet on the Vegas underdog Colts will pay off if the Colts win by any score or if the favorite Ravens win but by less than the Vegas point spread of 3.5 points. Notice that because the Vegas point spread ends with a “.5”, a push — if the favored team wins by exactly the point spread — is not possible.

Theoretically, if you must bet $110 to win $100 (typical in Vegas) then you’ll make money if you predict at 53% accuracy or better. But realistically, you need to predict at 60% accuracy or better, to take into account things like overhead.

My system is named after the Zoltar fortune teller machine that you can find in arcades. The arcade Zoltar is named after the Zoltar machine that appeared in the 1988 movie “Big”. I suspect that movie Zoltar was named after the 1960s arcade fortune teller machine Zoltan.