Back to Documentation
Model Documentation
MLB Models
Two complementary baseball models. One for the full nine, one for the first five. Both built on the same scoring-simulation foundation.
Available Models
MLB Game Totals
Full-game total runs predictions built by simulating each team’s scoring distribution from starters, bullpens, lineups, park, and weather.
Key Features
- Per-team run distributions, not just averages
- Starter, bullpen, and lineup-quality blending
- Park, weather, and umpire context layers
- Multi-gate edge filtering with calibrated probabilities
MLB F5 Totals
First-five-innings total runs model focused on the cleanest window of an MLB game — the starter-vs-starter matchup before bullpens and benches blur the picture.
Key Features
- Starter-driven scoring distributions
- Removes bullpen and pinch-hit noise
- Tighter dispersion for clearer probability reads
- Same proprietary calibration and edge gates