Every game on this week’s slate — the side we pick to win and our probability. There is always a pick side, so a coin-flip game shows up here as exactly that: a number just past 50%. Kickoffs run Thursday, October 15 through Monday, October 19.
12
games on the board
12
settled so far
64-34
record, all season
65%
picks correct
Thursday, October 15
Sunday, October 18
- Houston Texans vs Jacksonville JaguarsLost ✗
Jacksonville Jaguars, 64% to win
Projected total 42.2
- Carolina Panthers at Philadelphia EaglesWon ✓
Carolina Panthers, 58% to win
Projected total 44.4
- New Orleans Saints at New York GiantsLost ✗
New York Giants, 58% to win
Projected total 45.2
- New York Jets at New England PatriotsWon ✓
New England Patriots, 80% to win
Projected total 42.1
- Tennessee Titans at Indianapolis ColtsLost ✗
Indianapolis Colts, 75% to win
Projected total 45.8
- Chicago Bears at Atlanta FalconsWon ✓
Chicago Bears, 56% to win
Projected total 46.2
- Pittsburgh Steelers at Tampa Bay BuccaneersQB change
We pulled our call when the starting quarterback changed.
- Baltimore Ravens at Cleveland BrownsQB change
We pulled our call when the starting quarterback changed.
- Arizona Cardinals at Los Angeles RamsLost ✗
Los Angeles Rams, 80% to win
Projected total 51.8
- Buffalo Bills at Las Vegas RaidersWon ✓
Buffalo Bills, 72% to win
Projected total 53.3
- Los Angeles Chargers at Kansas City ChiefsWon ✓
Kansas City Chiefs, 82% to win
Projected total 42
- Dallas Cowboys at Green Bay PackersWon ✓
Green Bay Packers, 53% to win
Projected total 49.8
Monday, October 19
Tomorrow's card, by email
One email each morning with the day's free picks in full, across every sport we cover, published before play and graded afterwards. Our nine-chapter beginner's guide to prediction markets comes with it, free. Unsubscribe anytime.
How to read this board
Seattle Seahawks, 60% to win. Each row is one call: the team we pick to win and how likely we think that is. The probability is the pick — it is what gets graded, and a 63% call that loses is a 63% call that lost, not a bad model.
Every game on the slate gets a call. There is always a pick side, so the closest games publish just past 50% — and our confidence is capped at 90%, the edge of the calibration table our model was validated on. One number per row, and it is the probability: a signed spread sitting beside it read as an against-the-spread call, which is not what this desk publishes. We do project a margin, it is on the game page where there is room to label it, and no sportsbook line appears anywhere on this page.
NFL games can end in a tie — one or two a season do. A tie is void: it shows on the board as what it is, and it never counts for or against the record.