Redshirt
Analysis

Inside the Model: Power Ratings to CFB Fair Value

How the CFB model turns power ratings into fair value: simulate the season, convert win paths to probabilities, de-vig market prices, then read the edge.

By Redshirt Editorial · 2026-08-03
Analysis
vsvsvs
Key takeaways
  • The model converts power ratings into title probabilities by simulating the full season, then compares that fair value against de-vigged market prices.
  • Notre Dame carries the highest national-title fair value at 11.2%, just ahead of Ohio State at 11.1% and Oregon at 10.3%.
  • De-vigging strips the overround from raw contract prices so a team's fair value reads as a clean probability.
  • Ohio State trades at a best price of 12c on Kalshi against a model fair value of 11.1%, one of the tighter gaps on the board.
  • Fair value is an estimate, not a guarantee; both the model and market prices carry error and revise as information arrives.

The model turns power ratings into fair value in four steps: rate every team on one scale, convert rating gaps into per-game win probabilities, simulate the season to count title outcomes, then compare that probability against the de-vigged market price. The output is a single number per team, a fair value, that can be read straight against a contract quote.

At the top of the current national-title board, Notre Dame's fair value is 11.2 percent, Ohio State's is 11.1 percent, and Oregon's is 10.3 percent. Those are model estimates of title probability, not prices, and they are what every quoted contract gets measured against.

What is a power rating and why start there?

A power rating expresses each team's strength as one number on a shared scale. The value of the number is not the number itself but the gap between two teams: that gap, adjusted for home site and neutral fields, maps to a win probability for a single game.

Starting from ratings keeps the process disciplined. Instead of pricing a title directly, the model prices the thing it can actually estimate, a game, and lets the season structure do the rest. A schedule of favorable matchups compounds into a higher title number; a gauntlet drags it down even for a strong team.

How do game probabilities become a title number?

Per-game probabilities are the inputs; the season simulation is the engine. The model plays the full schedule many times over, sampling each game by its win probability, then advancing simulated results through the conference races and the playoff bracket.

Counting how often each team finishes as champion across all simulations yields its title share. That share is the fair value. Indiana at 8.2 percent and Miami at 6.8 percent sit where they do because their simulated paths to a title clear less often than the teams above them, not because of any single result.

Why de-vig the market before comparing?

Raw contract prices are inflated by an overround: add up every team's price and the field sums to more than 100 percent, because the spread between buy and sell has to live somewhere. Comparing a model probability to a raw price without adjusting is an apples-to-oranges error.

De-vigging scales the field back toward 100 percent so each price reads as a clean implied probability. Only then does a comparison mean anything. The consensus fair value shown for each team is this de-vigged number, blended across venues such as Kalshi and Polymarket.

Where does the model see the tightest gaps?

The interesting cases are where best available price and fair value nearly touch. Ohio State's best price is 12c on Kalshi against a 11.1 percent fair value, one of the closer alignments near the top. Notre Dame's 13c best price on Polymarket sits just over its 11.2 percent fair value, and Texas prices at 11c on Polymarket against 9.6 percent.

Reading the board this way turns a list of numbers into a shortlist. The model does not chase the highest fair value; it flags contracts whose price sits at or below the model line. For traders comparing venues, promo access such as Kalshi FADE or Polymarket TGSWC changes net cost but not the underlying fair value, which is the point of measuring price against the model in the first place.

None of this is a forecast dressed as certainty. Both the simulation and the market price carry error, and both revise as depth charts, results and liquidity move. Fair value is the reference line, not the last word.

National-title fair value, top of board
Notre Dame11.2%
Ohio State11.1%
Oregon10.3%
Texas9.6%
Georgia9.6%
Indiana8.2%
Miami6.8%
TeamsNotre DameOhio StOregonTexasGeorgia
Advertisement
Polymarket ad

Frequently asked questions

What is a power rating in a college football model?

A power rating is a single number estimating a team's strength on a common scale. The model uses the gap between two teams' ratings, adjusted for site, to estimate the probability of each game result.

How does the model turn power ratings into title fair value?

It simulates the season many times using per-game win probabilities, then counts how often each team wins the title. That share becomes the team's fair value, such as Notre Dame at 11.2 percent.

What does de-vigging a market price mean?

De-vigging removes the built-in overround from raw contract prices so the field sums closer to 100 percent. It converts a listed price into a cleaner implied probability for comparison against model fair value.

Where does model fair value sit tightest to price?

Ohio State is close, with a best price of 12c on Kalshi against a 11.1 percent fair value. Notre Dame's best price of 13c on Polymarket sits just above its 11.2 percent fair value.

Does a higher fair value mean a contract is a good trade?

Not on its own. A trade only looks attractive when the best available price sits below the model's fair value, and even then both numbers can be wrong.

About the author
Redshirt Editorial

Redshirt Analytics editors cover college football prediction markets: how contracts price the season, where the value sits, and how the platforms compare.