Every MLB game, projected before first pitch.
Win probability, a run-line lean, and a projected total for every matchup on today's slate. Each shown next to the market price, with a confidence tier so the board is sortable by conviction.
Win probability, a run-line lean, and a projected total
- Win probability per side: each team's modeled chance to win the game, shown as a bar next to the market's moneyline price.
- Run-line & total leans: which side of the run line and the over/under the model favors, framed by a projected final-score range.
- Confidence tiers: an Elite / Strong / Good / Moderate label on every lean, so the slate is sortable by conviction rather than raw percentage.
- Final once lineups set: a card updates through the day and is treated as final when the lineups are confirmed, so you know when a number has settled.
The board never names a "lock." It shows the model's number, the market's number, and the edge between them. What you do with that gap is up to you.
Win probability, park & weather, and the fair price
Win probability is not the same as a sportsbook's implied odds. In a two-outcome game the two teams' chances add up to 100%, but a book's posted prices add up to more than that once its margin is included. The model's win probability is an independent estimate of each side's true chance. Which is exactly why it can disagree with the price.
Park and weather quietly move a run total. Hitter-friendly parks and wind blowing out toward the fences tend to push totals up, while a cool, still night in a pitcher's park pulls them down. The model accounts for both automatically, so the projected score already reflects the conditions rather than a neutral-site guess.
Finally, the fair price. Removing the book's margin from both sides of a market gives a de-vigged, or "fair," number. An estimate of the true implied probability with the house edge stripped out. Comparing the model's number to that fair price is how the gap called edge is measured. A bigger tier means a wider gap, not a certainty.
MLB game projections, explained
What is a moneyline pick vs a run line pick in MLB?
A moneyline pick is about which team wins the game outright, at whatever price the market attaches to each side. A run line pick adds a margin (typically 1.5 runs) so the favorite must win by two or more, or the underdog must lose by one or win outright. The board shows a lean on both so you can see which market the model likes best.
How is win probability calculated for an MLB game?
In general terms, win probability estimates each team's chance to win by combining the starting pitchers, the lineups they field, and the conditions of the park and weather, then translating that into a single percentage for each side. It is an independent estimate, expressed as a probability, that can then be compared to the market's price.
What does 'projected score' mean in a baseball prediction model?
A projected score is the model's estimate of how many runs each side is likely to plate, shown as a range rather than a single exact number. It is the basis for the total lean and helps frame how comfortable a win-probability edge is. A projected blowout reads differently than a projected one-run game.
Is a model projection the same as a betting pick?
No. A projection is an estimate of a likely outcome from the underlying inputs. Whether that estimate represents value depends on comparing it to the market price. The board surfaces the gap (what we call edge) but no projection is a guarantee, and any single game can go either way.
How often are MLB game projections updated before first pitch?
Projections refresh through the day as lineups are confirmed, weather updates, and market prices move. A game's numbers are treated as final once the lineups are set, so a card can shift from an early estimate to a confirmed one as first pitch approaches.
What's the difference between the model's number and the sportsbook's odds?
The sportsbook's odds reflect market demand and the margin the book wants to hold, while the model's number is an independent estimate from the inputs above. Removing the book's margin gives a fair price; comparing that to the model's number is how a potential edge is identified.
What do the confidence tiers (Elite, Strong, Good, Moderate) mean?
The tiers are a plain-language label for how big a gap the model sees between its number and the market's. A higher tier reflects a wider gap, not a promise. The same taxonomy is used across every tool on the site, so a Strong lean means the same thing on a game card as it does on a prop board.
See today's full slate
Win probability, a run-line lean, and a projected total for every game. Each next to the market price, updated through the day.