College football prediction markets: what makes CFB different

Everything to this point applies to any sports market. This chapter is about why this book keeps using college football: the sport’s structure changes what the market can and cannot know, in ways a trader can actually use.

Too big to watch

The top division alone runs over 130 teams. On a full Saturday, thirty-plus games kick off inside twelve hours. No analyst, no desk, no crowd watches all of it. Attention is the scarce resource, and it pools where the cameras are. The marquee night game gets priced by everyone with an opinion in America. The 11am conference game between two unranked programs gets priced by fewer. Thin attention is where homework still has a chance to matter, which is precisely why chapter 6’s honesty matters here: a chance, at the margins, in places you genuinely know. Not a license.

Talent gaps a draft would never allow

Professional leagues are engineered toward parity. College football is engineered toward dynasties, and the schedule guarantees collisions between programs that recruit different planets. Probabilities that would be freakish in the NFL are routine Saturdays here, which means CFB markets live at the extreme ends of the price scale: exactly where chapter 3’s frequency discipline and chapter 5’s longshot bias do their loudest work. It is the best classroom in sports for learning what 90% actually means, because it shows you 90% forty times a season.

How our desk trades that structure for honesty

The band above is our own college football publication rule, and each edge of it encodes a lesson from this book. The 0.55 floor: below it, our testing can’t distinguish the model from a coin, so we don’t publish. A coin flip sold as conviction is noise at best. The 0.97 ceiling: above it, so few real games exist that the number could never be honestly graded. We also publish our own projected margin and total as context, never picks(only the win probability is graded), and when our number sits far from the market consensus, we usually withhold the call, because our own history says those are the ones we’re most likely getting wrong.

The result is a shorter board than anyone else’s, on purpose: every game we cover carries one number we’re prepared to be graded on, and the grading is public. Your board should be shorter than you want, too. A conference you truly follow, a reference number you trust, and the patience to pass on the other twenty-five games. In a sport this size, knowing three teams deeply beats knowing a hundred and thirty by reputation.