README.md

metadata

title: Paydirt Model Updates
emoji: 🏒
colorFrom: red
colorTo: yellow
sdk: docker
app_port: 8501
pinned: false

🏒 Paydirt NHL Model Updates

A Streamlit application for generating NHL DFS lineup projections and updating model data in real-time.

Features

• NHL Lineup Generation: Automated lineup optimization for DraftKings and FanDuel
• Real-time Updates: Integrates with Google Sheets for live data updates
• Multiple Platforms: Supports both DK and FD contests
• Ownership Projections: ML-powered ownership predictions using XGBoost, LightGBM, and KNN models
• Discord Notifications: Automatic notifications when updates complete
• MongoDB Integration: Stores historical data and projections

Architecture

• Python/Streamlit: Web interface and data processing
• Go: High-performance lineup generation
• MongoDB: Data storage and retrieval
• Google Sheets API: Live data integration
• Docker: Containerized deployment

Setup

Local Development

1. Install Dependencies

pip install -r requirements.txt

2. Configure Secrets

Create .streamlit/secrets.toml:

MONGODB_URI = "your-mongodb-uri"
GOOGLE_PROJECT_ID_1 = "your-project-id"
# ... (see .streamlit/secrets.toml.example for full template)

3. Build Go Programs (if developing locally on Windows)

cd func/dk_nhl_go
go build -o NHL_seed_frames.exe NHL_seed_frames.go

cd ../fd_nhl_go
go build -o NHL_seed_frames.exe NHL_seed_frames.go

4. Run the App

streamlit run src/streamlit_app.py

Hugging Face Deployment

This app is configured for deployment on Hugging Face Spaces using Docker.

1. Create a Space on Hugging Face with Docker SDK
2. Add Secrets in Space settings (Settings → Repository secrets)
   • Add all variables from .streamlit/secrets.toml.example
3. Push Code to your Space repository
4. The Docker build will automatically:
   • Build Go programs for Linux
   • Install Python dependencies
   • Configure the environment

Project Structure

Paydirt_model_updates/
├── src/
│   └── streamlit_app.py      # Main Streamlit application
├── func/
│   ├── NHL_own_regress.py    # Ownership prediction models
│   ├── dk_nhl_go/            # DraftKings lineup generator (Go)
│   └── fd_nhl_go/            # FanDuel lineup generator (Go)
├── Dockerfile                # Docker configuration
├── requirements.txt          # Python dependencies
├── go.mod                    # Go module definition
└── README.md                 # This file

Technologies Used

Python Stack

• Streamlit: Web interface
• pandas: Data manipulation
• numpy: Numerical operations
• xgboost/lightgbm/scikit-learn: Machine learning models
• pymongo: MongoDB client
• gspread: Google Sheets API
• ortools: Linear optimization

Go Stack

• MongoDB Driver: Database connectivity
• High-performance lineup generation algorithms

Environment Variables

Required environment variables (can be set in .streamlit/secrets.toml or as environment variables):

Variable	Description
MONGODB_URI	MongoDB connection string
GOOGLE_PROJECT_ID_1	Google Cloud project ID (primary)
GOOGLE_PRIVATE_KEY_1	Service account private key (primary)
GOOGLE_CLIENT_EMAIL_1	Service account email (primary)
GOOGLE_PROJECT_ID_2	Google Cloud project ID (backup)
GOOGLE_PRIVATE_KEY_2	Service account private key (backup)
GOOGLE_CLIENT_EMAIL_2	Service account email (backup)
DISCORD_WEBHOOK_URL	Discord webhook for notifications
MASTER_SHEET_URL	Google Sheets URL for master data

Usage

1. Open the app in your browser
2. Check connection status in the sidebar
3. Click "🚀 Generate NHL Lineups" to start the process
4. Monitor progress as the app:
   • Builds prop betting tables
   • Generates player-level outcomes
   • Creates stack matrices
   • Runs lineup optimization
   • Updates MongoDB with results
   • Sends Discord notifications

Security Notes

• Never commit .streamlit/secrets.toml to version control
• Use Hugging Face Secrets for production deployment
• Rotate service account keys regularly
• Keep MongoDB credentials secure

License

Proprietary - Paydirt Sports

Support

For issues or questions, contact the development team.