The model turns power ratings into fair value in three stages: it rates every team on one scale, simulates the season and bracket to convert those ratings into a title probability, then reconciles that probability against de-vigged market prices to land on a single fair value. The output is the number the desk treats as the honest price of each title contract. At the top of the current board, that process puts Georgia at 21.8%, Notre Dame at 13.8% and Miami at 9.6%.
What do power ratings actually measure?
A power rating is a single estimate of team strength on a common scale, built so that the gap between any two ratings maps to an expected margin on a neutral field. It is the raw material, not the answer. On its own a rating says nothing about a national title; it only says how one team should fare against another in a single game.
That distinction matters because the title market is not a one-game question. It is the compound result of a full schedule, conference positioning, and a bracket. A high rating is necessary but not sufficient: a strong team on a difficult path can carry a lower title probability than a slightly weaker team with an easier route.
How do ratings become a title probability?
To bridge that gap, the model simulates the remaining season many times. In each run, every game is resolved using the rating gap plus an error term that represents the uncertainty in any single result. The simulation then advances teams through conference standings and the bracket exactly as the real format would.
After enough runs, the model counts how often each team finishes as champion. That frequency is the raw model probability. It already absorbs strength of schedule, the odds of reaching the bracket, and the compounding risk of winning several elimination games in a row, which is why favorites rarely price far above the low twenties this early.
Where do market prices enter the fair value?
Model output is one input, not the verdict. The second input is the market itself, read across Kalshi and Polymarket. Raw contract prices cannot be used directly because the full field sums to more than 100%: that excess is the overround, or vig, baked into every quote.
De-vigging rescales the field so the implied probabilities are internally consistent and sum toward 100%. The de-vigged consensus is then blended with the model's simulated probability to produce fair value. That is why Georgia's fair value of 21.8% sits just under its 23c best price on Kalshi: the cent figure still carries a margin that the de-vig removes.
What does the fair-value board look like now?
The resulting hierarchy is clear at the top and flattens quickly. Georgia stands alone in the twenties, Notre Dame holds the teens, and a cluster of Miami, LSU, Indiana and Texas occupies the mid-to-high single digits. Ohio State and Alabama round out the near-5% tier.
The shape is typical of a title board with the bracket still unresolved: one clear leader, one clear second, then a wide band of contenders separated by small fair-value gaps where a single result can reorder the group.
How to read the gap between fair value and price?
Once fair value exists, the best price across venues shows where a contract trades rich or thin against it. The gap is widest on longshots, where the margin is a larger share of a small number: Ole Miss shows 3.3% fair value against a 5c best price, and Utah 2.6% against 4c. Near the top, the gap is tighter, with Georgia at 21.8% versus 23c and Notre Dame at 13.8% versus 14c.
The practical read is that cheapest venue and fair value are different questions. Kalshi holds the best price on Georgia, Miami and Ohio State, while Polymarket leads on Notre Dame, LSU, Indiana, Texas and Alabama. Traders comparing venues can weigh promo value as well, such as Kalshi's FADE or Polymarket's TGSWC, though those do not change the underlying fair value.
Prices and the model can both be wrong, and fair value is an estimate, not a guarantee. Its value is as a consistent yardstick: a number built from ratings, disciplined by simulation, and checked against a de-vigged market rather than a single raw quote.
