martineastwood/penaltyblog ? reverse-engineered prompt

Build me a Python package for football analytics that helps me pull in match and event data, clean and reshape it, and then use it to predict results and betting markets.

I want it to handle common soccer data sources, let me work with nested JSON without a lot of manual cleanup, and make it easy to calculate things like match probabilities, over under chances, implied odds, and team ratings such as Elo, Massey, Colley, and Pi. It should also include a few strong match models like Poisson and Dixon Coles, plus some Bayesian style models if that fits naturally. If possible, add simple ways to scrape public football data and create a few nice visualizations for analysis and presentations.

Make it feel like a polished, production ready library with good docs, examples, and tests. Use current best practices, and look up any API details online if you need to.