College Football Playoff · Monte Carlo projection

CFP Playoff Odds

— teams make the 12-team field in at least 95% of simulated seasons. The last at-large spots are closest among —.

Based on — simulated seasons, starting from the —, with ESPN data pulled —.

Simulated seasons — Each one plays out the rest of the season and picks a field
Remaining FBS games — Plus all 10 conference title games
Near-locks (95%+ to make it) — Of 12 playoff spots
Most common exact field — of seasons, across — different fields

Projected bracket

The 12 teams most likely to make the field, seeded by their average seed. Percentages are each team's chance of making the playoff at all.

First round (campus sites)Quarterfinal (bye)

Who wins it all?

The full playoff, played out in every simulated season: first round at the higher seed's stadium, then neutral-site quarterfinals, semifinals and final, using ESPN FPI win probabilities.

— is the most likely champion at —. The four favorites combine for — of simulated titles.

Team Conf CFP Make playoff Quarterfinal Semifinal Final Win title ESPN Kalshi

Quarterfinal means reaching it: a bye, or a first-round win. ESPN's title odds come from its own FPI season simulation, so they share this model's team ratings and aren't an independent check. Kalshi prices are the midpoint of the bid and ask; across all teams they add up to well over 100% because long shots trade at a floor of a cent or two. No rust penalty is applied to bye teams here.

Final regular-season record odds

For each Top 25 team and each possible final regular-season win total: how often the team finishes there, and, when it does, how often it wins its conference and makes the playoff.

Faded italic numbers come from records reached in under 1% of simulations, so they're noisy.

Records are regular season only and exclude conference title games. Conference title and playoff chances in a cell are conditional on finishing with that record. The two right-hand columns are each team's overall odds.

Chance to make the playoff

Every team that made the field in at least 1% of simulations. The darker part of each bar is the chance of a top-4 seed and a first-round bye.

Model vs ESPN and the betting market

Each team's chance to make the playoff from this model, from ESPN's own FPI season simulation, and from the Kalshi "make the playoff" market price. The committee part of the model is tuned so its odds line up with the average of the two benchmarks.

Average gap to the benchmarks (RMSE): — after calibration, down from — with hand-set weights.

Where teams get seeded

Chance of landing each seed, for teams with at least a 5% chance of making the field. Seeds 1–4 get byes. A blank cell is under 1%.

Team Make it 1234 5678 9101112 Avg seed

Is a first-round bye worth it?

A bye skips a first-round game the team could lose, but bye teams are 1-7 in their quarterfinals since the 12-team format began. This section measures the bye's value in the model and asks how unusual 1-7 really is.

In the model, the same team's title chance is — higher with a bye than as a 5–8 seed, and its semifinal chance is — higher. Given the betting lines for those eight quarterfinals, a record of 1-7 or worse would happen by chance — of the time.

Title odds a bye adds (vs seeds 5–8) — Same team, bye seed vs seed 5–8, no rust
Chance of 1-7 or worse by luck — Given the closing lines. — if bye teams were as strong as the model's typical bye team
Rust it takes to cancel the bye — Points off a bye team's rating in its quarterfinal
Chance rust is that big — Skeptical view (rust usually a few points). — with a looser prior

What a bye is worth for the same team

Gain in each outcome from holding a bye seed (1–4) instead of the seed shown. Each row averages over all four bye seeds and both teams in the swap, assuming no rust.

Instead of Semifinal Final Title

The #5 seed gains least from a bye because its first-round opponent is usually the weakest team in the field, the lowest-ranked conference champion.

Odds by seed

Chance to reach the semifinal and to win the title, by seed. Seeds 1–4 skip the first round.

How much rust would it take?

A bye team's gain from the bye, if it played its quarterfinal this many points worse than its rating (rust; negative means a rest advantage). The shaded band is the range of rust the eight real games are consistent with (95%). The best single fit is — points, beyond the chart. A skeptical estimate is — points.

The eight quarterfinals bye teams have played

Line = points the bye team was favored by at kickoff (negative: underdog). Pre-game chance converts the line to a win probability.

Season Seed Bye team Opponent Bowl Line Pre-game chance Result Score

The lines expected — wins. In 2024 they expected only —, because byes went to the top four conference champions, so Boise State and Arizona State got byes as heavy underdogs. The 2025 bye teams were favored in three of four games and expected — wins, but won one.

How to read the 1-7 record
  • Partly explained. Half of the losses came in 2024, when seeding rules handed byes to conference champions who were underdogs. That rule is gone, so that part shouldn't repeat.
  • Still unusual. Even after accounting for the lines, 1-7 or worse happens only a few percent of the time by chance. Something like rust, or opponents arriving with momentum, could be real.
  • But only eight games. The range of rust they're consistent with runs from about 3 points to well over 30, and the idea that byes hurt was formed after seeing the streak, which makes any streak look more significant than it is.
  • What it means for the bye. The bye is worth about as much as a 3.5-point rust penalty takes away. With a skeptical view of how large rust can be, the bye is still more likely than not to help. Taking the eight games at face value, it doesn't.
  • Method. The model plays out every simulated bracket: first round at the higher seed's stadium, then neutral sites, with ESPN FPI win probabilities and the same seeding as the rest of the page. Lines convert to win chances with the same normal spread model as the game simulation; a looser conversion gives — instead.

Top 25 at a glance

Current CFP Top 25 with projected regular-season wins and odds at each step to the playoff.

Rk Team Conf Now Left Proj W Likeliest Its odds Win out Title game Conf title Playoff Bye Avg seed Proj seed

Conference title races

The five highest-ranked conference champions get automatic bids. For each team: the chance to reach the title game (light bar) and to win it (dark bar and number).

How the simulation works
  • Remaining games. Every remaining FBS regular-season game is simulated using ESPN's Matchup Predictor win probability for that game.
  • Conference title games. Each conference's standings are rebuilt every simulation. Ties for a title-game spot are broken with that conference's own published procedure, including its multi-team rules: head-to-head counts only when every tied team played the others or one team swept them, and once a team is separated the rest start over. The top two meet in the title game, or the two division winners in the Sun Belt. ESPN has no odds for those games yet, so they use ESPN's FPI ratings, calibrated to match ESPN's own game odds.
  • Tiebreaker order by conference.
    • SEC: head-to-head, common opponents, best-placed common opponent, opponents' conference win %, capped scoring margin, random draw.
    • Big Ten: head-to-head, common opponents, best-placed common opponent (two-team ties only), opponents' win %, SportSource rating, random draw.
    • Big 12: head-to-head, common opponents, best-placed common opponent, opponents' win %, total wins, SportSource rating, coin toss.
    • ACC (2026 rules): head-to-head, SportSource rating, draw. Teams with 8 and 9 conference games tie if they match the leader's wins or losses.
    • American: head-to-head, CFP-ranked team that wins its final game (else computer composite), common opponents, overall win %, coin toss.
    • Pac-12: head-to-head, best-placed common opponent, common opponents, CFP ranking, SportSource rating, coin toss (multi-team ties use a slightly different order).
    • Mountain West: CFP-ranked team immunity, head-to-head, computer composite, overall win %, best-placed opponent, common opponents. The 2026 document wasn't available, so this order is partly reconstructed.
    • Sun Belt (division titles): head-to-head, division win %, best-placed division opponent, common non-division opponents, CFP ranking or composite, FBS win %, coin toss.
    • Conference USA: head-to-head, common opponents, best-placed common opponent, computer composite, opponents' win %, coin toss.
    • MAC: head-to-head, common opponents, best-placed common opponent, opponents' win %, SportSource rating, draw.
    SportSource ratings, computer composites and the SEC's scoring margin aren't public, so they're estimated from FPI and simulated results. CFP-ranking steps use the current CFP ranking.
  • Committee ranking. Each team starts from its current CFP spot. Each new loss costs ranking points. An upset loss costs about twice what a toss-up loss does, and each loss costs more than the one before, so a second or third loss hurts far more than a first. Wins over strong teams earn points. Losing a conference title game costs less than a regular-season loss. A team-strength term rewards teams ESPN's FPI rates higher than their current ranking. Random noise per team reflects committee judgment.
  • Calibration. The committee weights are not hand-picked. They're fitted so the model's playoff odds match the average of ESPN's FPI playoff odds and the Kalshi market, then checked on a fresh set of simulated seasons. The fit gives no extra bonus for winning a conference title beyond the automatic bid itself.
  • Field and seeding. The five highest-ranked conference champions get automatic bids and the next seven teams get at-large bids. All 12 are seeded straight by ranking, and the top four get byes.
  • Limits. Matching overall odds doesn't pin down every what-if. The playoff chance at a given record is the model's own inference, and it can differ from what the market implies. Win chances stay fixed at today's numbers, so late-season uncertainty is somewhat understated.