Loading the matchup…
Loading the matchup…
Hypothetical matchup, 2026 season



26Projected33If OSU upsetsNeutral site
Notre Dame Fighting Irish
Record 6–1 · #1 in ratings
Neutral site
Ohio State Buckeyes
Record 6–1 · #2 in ratings
| Per game, this season | ND | OSU | FBS avg |
|---|---|---|---|
| Turnovers in their games (lost + forced) | 2.6 | 2.6 | 2.5 |
| Penalty yards | 29 | 51 | 57 |
Teams whose games have more turnovers and penalties than average play less predictable games, so their matchup starts with more luck. Luck never changes who is favored, only how often the underdog pulls it off. Even at "almost none", single drives still go either way.
If Ohio State plays at home54% ▲
Ohio State: no wins over higher-rated teams in 7 FBS games
Notre Dame: 1 loss to a lower-rated team (BYU 24–28)
This season, compared with current ratings.
Upset score (under each projected score): the average final in the simulated games Ohio State won. Injuries aren't included; there's no injury data source.
Projected score: the average of 10,000 simulated games, played drive by drive. Each team's offense and defense ratings blend three things from every game this season, all adjusted for opponents: points scored and allowed, expected points (EPA) per play, and success rate on each play, with garbage time left out. Ratings start from last season's and shift as games are played. The simulation adds league scoring, home field, pace and turnover/penalty luck.
Upset score: the average final in only the simulated games the underdog won.
AI score: an AI prediction from the same season stats plus expected points per play, success rates and turnovers. It also weighs each head coach's record and the key players' season stats (passer, rusher, receiver and top defender, including transfers and freshmen).
Coaches and players in the projected score: counted through what each team has actually done on the field (its points and how well it moves the ball play by play), since that's what the players produced.
Accuracy: tested on every FBS game since 2022, each predicted only from games played before it, the model picked about 71% of winners and missed the final margin by about 12.5 points on average.
Not included: injuries, recruiting, rest days, travel or weather.
Out of 10,000 simulated games
10,000 simulated games at a neutral site, luck set to normal.
For entertainment purposes only. These are simulated outcomes, not predictions to bet on.
This season's leaders for each team. The AI score weighs these; the projected score counts them through the points each team has scored and allowed. No injury data is available.
Roster: 111 players · 7 transfers in
Roster: 119 players · 17 transfers in
Each bar is a 3-point range. Bars on the right are Notre Dame wins, bars on the left are Ohio State wins. Hover a bar for its share of games.
| Stat | Notre Dame | Ohio State | Note |
|---|---|---|---|
| Expected points per play, offense | 0.36 | 0.27 | Higher is better |
| Expected points per play allowed | -0.08 | -0.05 | Lower is better |
| Offensive success rate | 50.9% | 50.5% | Higher is better |
| Success rate allowed | 34.5% | 35.2% | Lower is better |
| Points per game | 41.1 | 39.3 | Raw, not adjusted |
| Points allowed per game | 13.4 | 13.4 | Raw, not adjusted |
Every team gets an offense and defense rating that blends points with play-by-play efficiency (expected points per play and success rate, garbage time removed), all adjusted for who they played: moving the ball on a top defense counts for more than doing it against a weak one. Blowouts count as at most a 42-point win. Each rating starts from last season's and gives way to this season's games as they're played.
Each simulated game gives both teams about 10 possessions. Every possession ends in a touchdown, a field goal or nothing, with odds set by how the offense and defense match up, plus a little game-day randomness. Ties go to overtime.
Running it 10,000 times and counting the wins gives the win probability. Ratings use 2,528 completed games and update every morning. Tested on every FBS game from 2022 on, each predicted only from the games before it, the model picked about 71% of winners and missed the final margin by about 12.5 points on average.