Quick Reference¶

Command reference for common operations.

Update All Predictions¶

Simple (Shell Script)¶

./scripts/update_all.sh

Detailed (Python Script with Timing)¶

# Run everything
uv run python scripts/run_all_predictions.py

# Run only specific models
uv run python scripts/run_all_predictions.py --models gbm
uv run python scripts/run_all_predictions.py --models nn
uv run python scripts/run_all_predictions.py --models classic
uv run python scripts/run_all_predictions.py --models gbm,nn

# Skip dashboard
uv run python scripts/run_all_predictions.py --skip-dashboard

Run Individual Models¶

GBM Models¶

# Standard GBM
uv run python scripts/run_gbm_predictions.py --variant standard --horizon 1y
uv run python scripts/run_gbm_predictions.py --variant standard --horizon 3y

# Lite (for stocks with limited history)
uv run python scripts/run_gbm_predictions.py --variant lite --horizon 1y
uv run python scripts/run_gbm_predictions.py --variant lite --horizon 3y

# Opportunistic (max price in window)
uv run python scripts/run_gbm_predictions.py --variant opportunistic --horizon 1y
uv run python scripts/run_gbm_predictions.py --variant opportunistic --horizon 3y

Neural Network Models¶

# Multi-horizon NN predictions (writes into valuation_results)
uv run python scripts/run_multi_horizon_predictions.py

Classic Valuations (DCF, RIM, etc.)¶

uv run python scripts/run_classic_valuations.py

Generate Dashboard¶

uv run python scripts/dashboard.py

View Results¶

Dashboard: Open dashboard/valuation_dashboard.html in browser

Database:

sqlite3 data/stock_data.db "SELECT ticker, model_name, upside_pct FROM valuation_results ORDER BY upside_pct DESC LIMIT 10;"

Common Workflows¶

Daily Update¶

./scripts/update_all.sh
# Then open dashboard/valuation_dashboard.html

Test Single Model¶

uv run python scripts/run_gbm_predictions.py --variant standard --horizon 1y

Refresh Only GBM Models¶

uv run python scripts/run_all_predictions.py --models gbm

Database Info¶

Location: data/stock_data.db

Main Tables:

Table	Description
`current_stock_data`	Current snapshot (598 stocks)
`valuation_results`	All model predictions with timestamps
`snapshots`	Historical fundamentals (358 stocks)
`price_history`	Historical prices

Model Names:

gbm_1y, gbm_3y
gbm_lite_1y, gbm_lite_3y
gbm_opportunistic_1y, gbm_opportunistic_3y
nn_1y, nn_3y
dcf, dcf_enhanced, growth_dcf, multi_stage_dcf
rim, simple_ratios

Testing¶

Run Tests¶

# Run all tests
uv run pytest

# Run only fast tests (skip model tests)
uv run pytest -m "not requires_models"

# Run only model tests (requires trained models)
uv run pytest -m "requires_models"

# Run specific test file
uv run pytest tests/test_gbm_models.py

# Verbose output
uv run pytest -v

Test Markers¶

Tests are marked with these categories:

Marker	Description
`requires_models`	Needs trained models (GBM, NN). Auto-skipped if models missing.
`requires_data`	Needs full database with historical data. Auto-skipped if missing.
`slow`	Long-running tests (>5 seconds)
`integration`	Integration tests
`unit`	Fast unit tests

Auto-skip behavior: Tests marked with requires_models or requires_data will automatically be skipped if the resources don't exist. You'll see a helpful message:

SKIPPED [1] tests/test_gbm_models.py:15: Trained models not found. Train models first with scripts in models/neural_network/training/

Help¶

# Script help
uv run python scripts/run_gbm_predictions.py --help
uv run python scripts/run_all_predictions.py --help

# List all test markers
uv run pytest --markers