Skip to content

JR-One-Yard/ralph-footy

Repository files navigation

Ralph -- NRL Footy Forecaster

Ralph is an NRL tipping tool that uses betting market odds and LLM-powered sentiment analysis to generate weekly tips -- and, more importantly, to teach your colleagues about AI, probability, and how markets work.

"I don't know everything about footy, but I learned a few things from the quant desk."

How it works

  • Market consensus: Ralph pulls H2H odds from multiple bookmakers, converts them to implied probabilities (removing the overround), and averages across sources to establish a baseline view.
  • LLM sentiment layer: Claude analyses team news, injuries, suspensions, and recent form to apply small adjustments (+/- 1-5%) to the market consensus via lightweight Bayesian updating.
  • Tip generation: Each game gets a final pick, a confidence level (Lock / Lean / Coin Flip), and a plain-English rationale explaining why.
  • Teaching moments: Every round includes a rotating "Did You Know?" segment covering probability, market efficiency, machine learning, or behavioural science -- footy is the hook, education is the payload.

Quick start

# Install in development mode
pip install -e ".[dev]"

# Run Ralph
ralph

# Generate tips for the current round (coming soon)
ralph tip

Project vision

See GENESIS.md for the full vision, architecture, teaching topics library, and iterative build plan.

About

🏉 Ralph — an NRL Footy Forecaster that tips games and teaches probability, AI, and statistics along the way

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages