What is an over/under bet?
The total isn't a guess at the final score. It's the number that splits the money — and betting it means ignoring who wins entirely.

Short answer: an over/under — also called the total — is a bet on the combined score of both teams. You're not betting on who wins. You're betting on whether the game produces more points than the posted number, or fewer.
It's the one major market where you can be completely wrong about the result and still cash.
Reading a total
A football game might be posted at 44.5.
- Over 44.5 wins if the two teams combine for 45 or more.
- Under 44.5 wins if they combine for 44 or fewer.
A 30–17 final is 47 points: over. A 20–17 final is 37: under. Which team won never enters into it.
Baseball works the same way with a smaller number — a total of 8.5 means nine or more combined runs goes over, eight or fewer goes under.
The half-point, again
Same reason as the point spread: the .5 makes a tie impossible.
At a whole number the game can land exactly on it. A total of 44 with a 44-point game is a push — stake returned, nobody wins. That's a real outcome, which is why pushes are graded as their own result in our ledger rather than quietly rounded into wins or losses.
The number and the price, separately
As with spreads, two things are being quoted and they move independently.
The total is 44.5. The price is usually around −110, which means you risk $110 to win $100 and need 52.4% to break even.
A book that's taking too much money on the over can move the total from 44.5 to 45, or leave it and shift the price to −120. Those are different adjustments with different implications, and a bettor comparing only the totals across sportsbooks is seeing half the picture.
What actually moves a total
Totals respond to different things than sides do, which is the main reason they're worth pricing separately.
In football: wind above roughly 15 mph is the single most punishing condition for a passing game, and it moves totals more than most injuries do. Beyond that — pace, whether either offense plays fast, injuries to skill players, and whether a game projects as a shootout or a slog.
In baseball: the starting pitchers dominate everything else. Then the park, which varies in measurable and stable ways, and again the wind — blowing out at an open stadium is a genuinely different game from blowing in.
This is why a model prices a total separately from a moneyline rather than deriving one from the other. The same game can be well-priced on the side and badly priced on the total, and vice versa.
What the total is actually claiming
The same thing a spread claims, pointed at a different quantity.
A total is not the sportsbook's forecast of the combined score. It's the number the book believes will draw roughly equal money to the over and the under. It's a price, shaped by where bettors will put their money.
That distinction matters because it's where the opportunity lives. Public money leans over — rooting for points is more fun than rooting against them — and a book that knows this can post a number slightly higher than its true estimate and still split the action. That gap between "the balancing number" and "the honest number" is the only thing worth betting into.
Our model doesn't try to predict the final score. It produces a probability for a specific total, converts it into its own fair price, and compares it against the price on the board. Where the two agree there's nothing to bet, which is most of the time. A pick exists only where the gap clears a minimum edge threshold — and most predictions never become picks.
One habit worth building
Read a total as a range, not a line.
A posted 44.5 doesn't mean the model on the other side expects 44.5 points; it means it expects a distribution of outcomes whose middle sits near there. Games land at 31 and at 58. Betting the over doesn't require the game to be high-scoring — only for it to be slightly more likely to exceed 44.5 than the price implies.
That's a smaller, stranger claim than it first appears, and it's why totals records are best read in units with the price attached rather than as a win percentage. Half the bets losing is the normal condition of a profitable totals bettor.
These are model outputs, not guarantees, and not financial advice. If betting has stopped being entertainment for you, our responsible gambling page lists free confidential help.
Published August 27, 2026 · For information only · 21+
Read the theory, then watch it run.
The model's highest-edge play lands in your inbox at 9 AM tomorrow. Free, every day.