Baseball Outs-per-Inning Simulator (v2)

Link to Medium article: 3 Outs? Try 5. The Data Says Baseball Would Be Shorter, Faster, and More Exciting

Calibrated Monte Carlo simulator exploring baseball with arbitrary outs per inning, keeping baserunners after 3 outs (bases clear only when the inning ends). Includes realistic base advancement, simple steals, and a full figure suite for articles: heatmaps, histograms, variance vs mean, and EV curves.

Install

python -m venv .venv && source .venv/bin/activate  # Windows: .venv\Scripts\activate
pip install -r requirements.txt

Quickstart

# Metrics for 9 innings × 3 outs
python src/cli.py --mode metrics --innings 9 --outs 3

# EV curve (outs 1..9)
python src/cli.py --mode curve --curve-sims 4000

# Heatmap of average runs per TEAM (1..9 outs × 1..9 innings)
python src/cli.py --mode grid --sims 800

# Run distribution histograms for 9×3 vs 6×4 vs 5×5
python src/cli.py --mode hist --sims 2000

# Variance vs mean for 9-inning games (vary outs)
python src/cli.py --mode varmean --sims 1500

All figures saved in outputs/.

Config

Edit configs/default.yaml to change lineup stats, hit mix, steal rates, or walk/HBP bump. Defaults are lightly calibrated to place 9×3 ≈ 4.1–4.3 runs per TEAM.

Modeling Notes

• Event probs: Walk/HBP ≈ OBP − BA + bump; Hit ≈ BA; within hits, use hit_distribution.
• Advancement: singles often score runner from 2B; doubles sometimes score runner from 1B; HR clears bases.
• Steals: low attempt rates; occasional outs add variance.
• Bases persist through any number of outs — inning ends only at outs limit.

License

MIT