MLB Game Predictions & Simulations

What the prediction engine does

For every game on the MLB slate, FullCountProps simulates the full matchup 2,000 times — projected lineup against probable starter, bullpen behind him, in that day's ballpark. Out of those runs come a win probability for each side, a most-likely final score, and a projected total. No gut picks, no narrative adjustments; if the number surprises you, the inputs explain why.

How a simulation works

Each simulated game is played out batter by batter: every plate appearance samples from that hitter's matchup-adjusted rates against that pitcher's tendencies, with relievers taking over as the starter's outing ends. Repeat the whole game 2,000 times and the distribution of outcomes becomes the prediction. The inputs and their sources are documented on the methodology page.

Want to poke at it? The what-if simulator lets you swap lineups or starters and watch the win probability move.

Kept honest

Model outputs are captured before first pitch and graded after the final out — results live on the accuracy page, wins and misses alike.