CLI Reference

After installing the package (pip install -e ".[all,dev]"), the following commands are available on your PATH.

pitch-generate

Generate synthetic baseball pitch datasets.

pitch-generate [OPTIONS]

Flag	Default	Description
--num-games	3000	Number of games to simulate
--at-bats	35	At-bats per game
--seed	42	Random seed
--output-dir	data	Output directory for CSV files

Example:

# Default: 3000 games, ~384K pitch rows
pitch-generate

# Custom: smaller dataset
pitch-generate --num-games 500 --at-bats 30 --output-dir ./my-data

Output files:

• <output-dir>/baseball_pitch_data.csv — Main pitch dataset
• <output-dir>/synthetic_pitch_sequences.csv — HMM training sequences

---

pitch-train

Train a single model.

pitch-train --model MODEL [OPTIONS]

Flag	Default	Description
--model	(required)	Model name from registry
--config	(auto-detected)	Path to model config YAML

Available models: logistic_regression, random_forest, hmm, autogluon, lstm, cnn1d, transformer

Example:

# Train with default config
pitch-train --model lstm

# Train with custom config
pitch-train --model lstm --config configs/models/lstm.yaml

---

pitch-benchmark

Run the full benchmark suite across all models.

pitch-benchmark [OPTIONS]

Flag	Default	Description
--config	(auto-detected)	Path to benchmark config YAML

Example:

# Run all 7 models through 5-fold CV
pitch-benchmark

# Custom benchmark config
pitch-benchmark --config my_benchmark.yaml

---

pitch-ablation

Run ablation studies on a specific model.

pitch-ablation --type TYPE [OPTIONS]

Flag	Default	Description
--type	(required)	Ablation type
--model	(from config)	Model to ablate

Ablation types: feature, architecture, data, hyperparam

Example:

# Feature importance ablation
pitch-ablation --type feature --model lstm

# Data scaling ablation
pitch-ablation --type data --model transformer

---

Make Targets

The Makefile provides shortcuts for common commands:

Target	Command
make install	pip install -e ".[all,dev]"
make data	pitch-generate
make train MODEL=lstm	pitch-train --model lstm
make benchmark	pitch-benchmark
make ablation TYPE=feature	pitch-ablation --type feature
make mlflow	mlflow ui --backend-store-uri experiments
make test	pytest tests/
make docs	mkdocs build --strict
make docs-serve	mkdocs serve
make clean	Remove build artifacts