XGBoost-powered prop predictions, real-time Vegas odds, and Kelly criterion sizing — all wired together in an open-source Python pipeline.
BetLab
NBA BetLab isn't a simple tip sheet. It's a full data pipeline built to identify and act on market inefficiencies before lines close.
XGBoost models for both player props (points, rebounds, assists) and game-level outcomes — spreads and totals.
Connects to The Odds API to pull real-time Vegas lines so your analysis is always based on current market data.
Scrapes ESPN for depth chart updates and applies a "Next Man Up" model to adjust predictions for injuries automatically.
EV-based filtering and adaptive Kelly criterion sizing optimize bet selection and bankroll management in every session.
Capture closing lines and measure the true quality of each bet against the sharpest market data available.
Relational schema for high-volume metric storage, enabling backtesting and trend analysis across full seasons of data.
The full pipeline from raw data to a placed bet — automated, optimized, and repeatable every game day.
Pulls the latest games, odds, and player statuses from the NBA API and The Odds API to feed the prediction models with live inputs.
XGBoost models generate predictions and the trading engine filters for high-confidence, positive expected-value opportunities.
Review daily picks with full confidence breakdowns and Kelly-sized recommendations, then bet with a genuine data-driven edge.