Michigan vs Ohio State 2026 Prediction and Rivalry History
Quick answer: The 2026 model picks
Ohio State over Michigan in the scheduled Michigan–Ohio State
meeting. The favorite’s win probability is 89.1%. The game is listed for
Week 13 on 2026-11-28. This is a statistical
projection, not a guaranteed result or betting advice.
This matchup page combines one dated 2026 model forecast with every completed meeting we
found from the 2016-2025 SportsDataverse schedule files. The forecast is the opponent-adjusted Elo rating alone because no qualifying spread was available.
The historical table is descriptive evidence, not an input that was manually adjusted to make
the rivalry story look closer or more dramatic.
2026 Michigan vs Ohio State model prediction
| Date | Week | Away | Home | Home WP | Away WP | Pick | Source |
|---|---|---|---|---|---|---|---|
| 2026-11-28 | 13 | Michigan | Ohio State | 89.1% | 10.9% | Ohio State | model |
The model treats the listed venue as Ohio State’s home field. Home field is estimated from the
historical FBS-vs-FBS training pool; neutral-site rows receive no home adjustment. A probability
above 50% identifies the pick, but it does not mean the other side has no plausible route to win.
Here, Michigan still owns the complementary probability shown in the table. That is the
most important distinction between a probability forecast and a categorical prediction.
Series record in the 2016-2025 sample
| Metric | Michigan | Ohio State |
|---|---|---|
| Wins | 4 | 5 |
| Total points | 252 | 290 |
| Points per meeting | 28.00 | 32.22 |
| Meetings / ties | 9 meetings; 0 ties | |
Across these 9 completed meetings, Michigan won 4
and Ohio State won 5. The average score contribution was
28.0 points for Michigan and 32.2 for
Ohio State. The average absolute margin was 14.4 points. Those totals help
frame the recent competitive balance, but they do not replace an opponent-adjusted rating. A
team can dominate an old portion of a series while entering 2026 with a very different roster,
coach, schedule strength, and current rating.
Recent trend versus the earlier sample
| Window | Michigan wins | Ohio State wins | Michigan PPG | Ohio State PPG |
|---|---|---|---|---|
| Most recent five | 4 | 1 | 27.8 | 22.2 |
| Earlier in sample | 0 | 4 | 28.2 | 44.8 |
The five-game split is included as a compact recency check, not as a second prediction model.
Small rivalry samples can swing sharply after one upset or one blowout. For that reason, the
published probability comes from the same system used for the full 2026 FBS sheet rather than
from an ad hoc rule such as “winner of the last meeting” or “team with the better decade record.”
That keeps this page comparable with every other published matchup.
Every completed meeting in this sample
| Season | Date | Michigan score | Ohio State score | Winner | Listed home/site |
|---|---|---|---|---|---|
| 2025 | 2025-11-29 | Michigan 9 | Ohio State 27 | Ohio State | Michigan |
| 2024 | 2024-11-30 | Michigan 13 | Ohio State 10 | Michigan | Ohio State |
| 2023 | 2023-11-25 | Michigan 30 | Ohio State 24 | Michigan | Michigan |
| 2022 | 2022-11-26 | Michigan 45 | Ohio State 23 | Michigan | Ohio State |
| 2021 | 2021-11-27 | Michigan 42 | Ohio State 27 | Michigan | Michigan |
| 2019 | 2019-11-30 | Michigan 27 | Ohio State 56 | Ohio State | Michigan |
| 2018 | 2018-11-24 | Michigan 39 | Ohio State 62 | Ohio State | Ohio State |
| 2017 | 2017-11-25 | Michigan 20 | Ohio State 31 | Ohio State | Michigan |
| 2016 | 2016-11-26 | Michigan 27 | Ohio State 30 | Ohio State | Ohio State |
The table preserves the row-level evidence behind the summary. Scores are final results from
the source release; they are not simulated box scores. We do not invent player statistics,
snap counts, injuries, or betting lines for these historical games. If the source marks a game
as neutral, the site column says neutral. Otherwise it shows the listed home team, even when a
rivalry is traditionally played at a recurring special venue.
How the 2026 probability was built
The rating pipeline processes final regular-season FBS-vs-FBS results chronologically. Each
result moves the two teams’ Elo ratings by an amount determined by the pregame expectation and
the final score margin. Beating a strong opponent therefore carries more information than
beating a weak opponent that the model already expected to lose. Ratings use 2024, 2025, and
completed 2026 finals available when the sheet was rebuilt. This is a schedule-strength-aware
approach; raw win percentage alone is especially misleading in college football because teams
play very different conference and nonconference schedules.
Home advantage is estimated from the same historical training pool. The win probability is a
logistic transformation of the rating difference plus that home adjustment. Where the public
SportsDataverse betting snapshot contains a spread, a separately calibrated spread probability
receives 60% weight and Elo receives 40%. When the row says “model,” as this one may, no spread
was available and none was fabricated. The full prediction artifact records which branch was
used for every game.
What this forecast does not know
This is a team-level statistical model, not a real-time injury desk. It does not claim to know
future quarterback availability, weather changes, depth-chart surprises, suspensions, or line
movement after the data snapshot. It also excludes FCS games from the common rating pool. That
choice improves comparability among FBS teams but means the projected-win table elsewhere on
the site is a forecast of FBS-vs-FBS wins rather than each school’s complete overall record.
Rivalry games can also behave differently from ordinary schedule spots because preparation,
emotion, and coaching incentives are unusual. We do not add an unmeasured “rivalry factor” to
force the probability toward 50%. Doing that would make the output more dramatic but less
auditable. The honest reading is straightforward: Ohio State is favored by the
current system, while Michigan retains a real upset probability.
FAQ
Is the pick based only on head-to-head history?
No. The historical series is shown for
context. The pick comes from the common 2026 rating/market pipeline used across FBS games.
Are the historical scores projections?
No. They are completed final scores. Only the
2026 probability row is model output.
Will the probability change?
Yes. New final games update Elo ratings, and a newly
available spread can change a model-only row into an ensemble row on the next rebuild.
Why show Brier scores on the backtest pages?
Accuracy treats a 51% miss and a 90% miss
the same. Brier score penalizes overconfidence, making it harder to hide bad probabilities behind
a favorable win-loss count.
Is this betting advice?
No. It is an informational model snapshot. Readers should not
treat it as a live sportsbook quote or as a substitute for current team news.
Related: all 2026 CFB predictions and projected rankings,
the 2025 Week 1 retrospective backtest,
and the NFL Facts Stats Hub.
Sources
Historical sample: 2016-2025 completed meetings. Model snapshot: 2026 season.
No mock games, scores, or lines.