Saturday's CFB Report Card: 27-for-50 on Winners, Value Bets Finish +11 Units
Saturday was a rough one for the raw picks, and we're not going to dress it up. Across 50 FBS games, the sims picked 27 winners — 54%. Against the spread: 47%. On totals: 45%. That's below the bar we hold ourselves to, full stop.
But here's the part that matters more than the scoreboard: the value bets still finished up 11 units.
The difference between picks and edges
A straight pick just asks "who wins." A value bet asks a harder question: where does our simulated win probability disagree with the book's odds by enough to matter? Saturday, the first question had a bad day. The second one paid.
That's not spin, it's the whole thesis. Anybody can pick winners at a coin flip. The value bet finder exists because 47% against the spread loses money at -110 (you need 52.4% just to break even), while a disciplined edge filter can finish up double-digit units on the same slate. The picks are the conversation. The edges are the business.
What the sims got right
The 1-vs-4 coin flip in Bloomington: the model ran Ohio State at Indiana 1,040 times and came out Indiana 31.5, Ohio State 31.3. The actual final was Ohio State 19, Indiana 15 — a four-point game decided on the last play, on a foot out of bounds. Wrong winner, laughably high total, but the shape of the game — a rock fight to the final minute — was exactly what the numbers promised. Some days the model sees the shape perfectly and still can't name the winner. We'll take the shape and own the rest.
The sims saw BYU coming from a mile away. 1,040 runs had the Cougars beating #8 Notre Dame 35.5 to 22.1, a two-touchdown call in a top-10 primetime matchup, no hedging. The final was 28-24, closer than the model drew it up, but the direction was never in doubt: BYU at home, BYU wins. When the sims say a home team takes a top-10 game by double digits, believe the arrow even if you shave the margin.
TCU was the quieter masterpiece. The model had the #14 Horned Frogs at Baylor 26.7 to 21.8, TCU by about five. The actual final was 17-6, TCU by eleven, and here's the fun part: the sims actually underestimated how thoroughly TCU's defense would smother Baylor. Six points. The model liked the Frogs' defense; reality liked it more. Two upset specials, two outright winners, both called before kickoff.
Where we improve
The full analysis is already running, spread by spread, total by total, to find exactly where the model leaked on Saturday. When we know, you'll read it here first. That's the deal: when the sims miss, the report card goes up, and the work to fix it starts the same day.
