MLB Win Probability Explained

The Pitcher Is Half the Game

No factor in any sport dominates a single game the way a baseball starting pitcher does. The same two rosters can be a 58% favorite or a 58% underdog depending only on who is on the mound. An ace facing a fifth starter shifts the whole matchup before the first pitch is thrown.

That is why an MLB probability barely means anything until both starters are confirmed. A late scratch, a bullpen game, or a spot start can move a fair price 10 points or more. Our model does not use starting-pitcher stats yet (see below), which is one reason our MLB numbers stay close to 50% and why a market price that knows the starter can disagree with us.

The Variance Problem

Baseball is low-scoring and played nearly every day, which makes any single game close to a coin flip. Over a full 162-game season a great team wins about 60% of the time and a terrible team still wins about 40% — the gap between the best and worst teams in the sport is narrower, game to game, than in any other league.

So you will rarely see us give an MLB team above 65%. When we post 62% on a clear favorite, that is already a strong lean by baseball standards. Don't read a number near 55% as indecision; that is often just what a real baseball edge looks like.

Beyond the Starter: The Bullpen

Starters rarely finish games anymore, so the bullpen decides a large share of outcomes. A dominant back end can protect a one-run lead; a gassed or thin pen gives it back, and a team whose top relievers threw the last two nights is more vulnerable late than its season ERA suggests. Bullpen workload is not a model input yet.

Park Factors

Where a game is played matters more in baseball than the home-field edge matters anywhere else. Coors Field in Denver inflates offense dramatically — thin air lets the ball carry — while pitcher-friendly parks suppress runs. A lineup built on power gains in a bandbox and loses in a cavern.

The same pitcher projects differently in Denver than in a park that swallows fly balls. Our model does not adjust for the park yet; home and road records capture only part of that effect.

Weather

Baseball is played outdoors all summer, and the wind is the factor people underrate most. A 15 mph wind blowing out turns warning-track outs into home runs; the same wind blowing in erases them. Temperature matters too — the ball carries farther in the heat. Our model uses temperature and wind speed at open-air parks (not wind direction yet); the full picture of how weather moves games is in our MLB weather breakdowns.

Lineups and Handedness

The batting order that actually takes the field can swing a game. A star getting a scheduled day off, or a lineup stacked with left-handed bats against a lefty starter, changes the expected runs. Platoon splits (how hitters do against same- vs opposite-handed pitching) are a real effect in baseball. Our model does not use lineups or platoon splits.

What Our MLB Model Uses Today

Nine factors: Elo rating, each team's home and road record this season, runs scored and allowed per game (the two heaviest weights), recent form, head-to-head history, rest days, injuries weighted by position (a starting pitcher on the injury list counts most), and weather at open-air parks. Starting-pitcher stats, bullpen workload, park factors and platoon splits are not inputs yet. Every game is re-scored each morning and every 15 minutes in the three hours before first pitch. The full method is on our methodology page.

Home Field Advantage

MLB home teams win about 54% of the time — real, but the smallest home edge among the major sports. There is no last-change or crowd-noise effect the way there is in hockey or football; the biggest piece of “home advantage” in baseball is simply batting last and knowing your own park. We treat it as the modest factor it is.

Reading Our Breakdowns

Every game shows the factors favoring each side. When pitching shows HIGH impact, that is usually a large gap between the two starters. When a park or weather factor is flagged, check the venue and the wind — those are the levers that turn a pitcher's duel into a slugfest or back again.

Baseball probabilities cluster toward 50/50 more than other sports. That is not the model hedging — it is the sport telling you the truth about how often the underdog wins. For more on what these percentages do and don't promise, see how to read a win probability.

And in a sport this noisy, the only real test is whether the numbers hold up over a long season. We publish how our MLB calls have actually landed on our accuracy page, and how each factor is weighted on our methodology page.