Python Developer — Sports Prediction Model, Betting API Integration, Automated Pipeline

Project Overview: I am building an automated value bet identification system for the 2026 FIFA World Cup group stage, with subsequent expansion to NFL and NBA seasons. The system uses a Poisson-based expected goals model combined with live odds API data to identify positive expected value betting opportunities across match result and goals markets. What Needs to Be Built: A Poisson football prediction model trained on historical international match data (available as free CSV datasets). Outputs win/draw/loss probabilities and a full scoreline probability matrix for each match. Integration with The Odds API to fetch live bookmaker odds across 10+ bookmakers, de-vig the odds, and calculate expected value versus model probabilities. Integration with the Polymarket Gamma API (public, no authentication) to fetch prediction market prices for the same matches and generate a confirmation/contradiction signal. A backtesting framework using walk-forward validation across the 2010, 2014, 2018, and 2022 World Cup group stages to validate model edge. Must include calibration testing, CLV simulation, and benchmark comparison. A Kelly criterion staking calculator (half-Kelly implementation) that outputs recommended stake sizes per bet based on a defined bankroll. A daily automated pipeline that runs on Mac Mini (macOS), fetches today's fixtures and live odds, runs the model, and sends a formatted Telegram notification with all value bets found that day. The Claude API (Anthropic) is integrated as an interpretation layer — the system sends model outputs and divergence data to Claude API and receives natural language analysis of each bet. I will provide the API key. Technical Requirements: Python 3.11+ Libraries: pandas, numpy, scipy, scikit-learn, XGBoost, requests, anthropic, python-dotenv, schedule, matplotlib All credentials stored in .env file Code must be clean, commented, and modular The system must run autonomously on Mac Mini with minimal daily input from a non-technical operator Data Sources (all provided/free): Historical World Cup data: github.com/jfjelstul/worldcup (CSV) International results: github.com/martj42/international_results (CSV) Live odds: The Odds API (I will provide API key) Team statistics: API-Football (I will provide API key) Polymarket: public REST API, no key required What I Am Looking For: Someone who understands both the technical implementation AND the domain. Specifically: you should understand what expected value means, why walk-forward validation is required (not k-fold), what closing line value is and why it matters, and why the Dixon-Coles correction is relevant for low-scoring football matches. If these terms are unfamiliar, this project is probably not the right fit. Timeline: 3-4 weeks, starting immediately. World Cup begins June 11. Deliverables: Full codebase with documentation Working backtesting report showing ROI across 2010–2022 World Cups Daily pipeline running and sending Telegram notifications 30-minute walkthrough call explaining how to operate the system