College Football Bowls & Playoff Betting Lines

Coverage

16 games in this week's window · 16 with a spread · 16 with a total · 3 with a moneyline · 0 with no line at all.

This board describes what the market posted and what happened. It is not betting advice and no line here is actionable — the prices are historical closing numbers, not quotes.

• The consensus line is the MEDIAN across every book that posted that market, not a single book's number and not an average — a median is unmoved by one outlier book. It is a DISCRETE median, so the number shown is always one a book actually posted; an interpolating median would print prices like -2.75 that nobody offered and that make an exact push impossible. The number of books behind each market is published with it; some games rest on one book.
• Spreads are shown from the home team's perspective in book convention: NEGATIVE means the home team is favored. Each game also carries an explicit favorite so nothing depends on the sign.
• Moneylines are aggregated as implied probabilities and converted back to a price. Medianing American odds directly is not monotone across the -100/+100 boundary. The two sides' implied probabilities sum to more than 1; that excess is the book's hold, which is published and removed to get the win probabilities shown.
• Line movement is measured only over the books that posted BOTH an opening and a closing number, with the open and close both medianed over that same set — so the opening plus the movement always equals the closing number shown in that block. That block's close can differ slightly from the headline consensus, which is medianed over every book including the majority that never posted an opening. Openings are sparse, so movement is published only when at least 2 books reported one, which in practice means 2024 onward.
• Against-the-spread grading uses the consensus closing line. A game landing exactly on the number is a push and is counted separately, never as a win or a loss.
• The model line is this site's own SRS rating, fit on games through the END OF THE PRIOR WEEK — it never sees the result it is being compared against. Only FBS-vs-FBS games get a model column: an FCS opponent, or a team under the four-game minimum, gets none rather than a guessed one. The two error figures are averaged over exactly the same games, which is what makes them comparable.
• A single week is roughly fifty games. Any weekly record here — the market's, the model's, home teams', favorites' — carries a standard error near seven percentage points and should be read as description, not as evidence that anything predicts anything.
• Line coverage is uneven: nearly all FBS-involving games in recent seasons carry a line, older seasons fewer, and games below the FCS level none at all. The coverage block reports what was available.

FACT (posted lines, final scores), DERIVED (consensus, grading, model line, records)