The Marginbeta
CBB · CWS OddsField

CWS Odds — Who Wins Omaha

50K Monte Carlo · Regional sims · Pitching edge

50,000-sim Monte Carlo through the NCAA tournament: regional survival, super-regional, and national-title probabilities for every team in the projected field. Team ratings powered by PEAR Ratings. For the upstream Field of 64, see the Field of 64 projection.

Field source:
Rating engine:

Projected bracket walk

The model's most-likely path through the field, built from the same 50,000-sim Monte Carlo that drives the odds table. Each round's pick is the modal winner among the sims where every prior round's modal pick also occurred -- so the percentage next to each team reads as "given the walk to date, the share of sims in which this team advances from this round." A 62% conditional pick reads very differently from an 84% one. Bullpen depletion from regional play carries into the super and into Omaha (α = 0.35 between rounds, α = 0.75 before the bo3 final), so a team that grinds through a 6-game regional pays for it.

The modal walk shows the single most-likely sequence of round winners. Because bracket structure matters, the champion in this walk can differ from the team with the highest overall championship probability -- both views are shown.

Loading bracket walk…

Upset Watch

The modal walk above shows the most likely path one round at a time. The 50K Monte Carlo tells a richer story by sampling every game with rating noise. These are the regionals where either the 1-seed is barely a favorite or a lower seed clears the 25% upset threshold. Sorted by upset weight.

Loading Upset Watch…

Pitching Edge

Starter rotation strength for the title contenders. Color-coded ERAs: green ≤ 3.00, red ≥ 4.50. Where the source column reads D1B probable the data is the official D1Baseball weekend probable; Estimated means we inferred from this season's pitcher appearance log when probables weren't yet announced. Bracket sims run on ?rotation_aware=1 use these per-game deltas instead of the team rating average.

Loading rotation data…

Injury Report

Where pitcher health has pulled a team's rating down. Each entry is a rating-point adjustment; in our per-game logit, one point is worth roughly nine percentage points of single-game win probability. Sorted by magnitude.

Loading injury data…
How we built this — model, bracket, field

The model

Per-game logit fit on 72,809 D1-D1 games (2014–2024) plus the current PEAR Ratings. K_DIFF = 0.37 on rating diff (CI [0.364, 0.379]). HFA_LOGIT = 0.23, ~6% home boost (CI [0.217, 0.249]). Per-game Gaussian rating noise σ ≈ 2.0 captures starter-day variance. Fatigue penalty 0.40 rating pts per cumulative game in a regional or super series.

Bracket structure

16 regionals × 4 teams each (double elimination) → 8 super regionals (best-of-3) → 8 teams arrive in Omaha, split into two 4-team double-elims → best-of-3 final. Host bonus has been fit to zero in the current era -- it's fully captured by home-field advantage on neutral-vs-host games.

Field of 64

We pull weekly projections from D1Baseball and Baseball America, plus our own auto-projection from PEAR Ratings, then blend them into a consensus field (3-of-3 teams first, then 2-of-3 by rating, then 1-of-3 to fill 64). Selection day is Monday, May 25 -- after that we replace projections with the actual NCAA bracket.