v1.04

Moved Stan model code into separate files as the previous method of loading from a string via a temporary file was causing issues for some users on Windows

v1.0.3

• Fixed bug in how the Bayesian models indexed teams in the predict function
• Goals model now only predict individual team names rather than iterables of team names as was causing compatibility issues between different sequence objects.

v1.0.2

• Updated how the Bayesian models handle the Stan files to prevent access denied issues on Windows

v1.0.1

• updated install_stan function to install the C++ toolchain on Windows if required

v1.0.0

• Removed pymc as a dependency
• Updated all other dependency versions
• Added support for Python 3.13
• Rewrote BayesianHierarchicalGoalModel model into Stan instead of pymc and updated prediction method to integrate over the posterior rather than just sampling the mid-point
• Rewrote BayesianRandomInterceptGoalModel into Stan instead of pymc, updated model to use a more accurate random intercept, and updated prediction method to integrate over the posterior rather than just sampling the mid-point
• Rewrote BayesianBivariateGoalModel into Stan instead of pymc, improved model so converges better, and updated prediction method to integrate over the posterior rather than just sampling the mid-point
• Added BayesianSkellamGoalModel model for predicting outcomes of football (soccer) matches based on the Skellam distribution
• Removed obsolete sofifa and espn scrapers
• Optimised RPS calculation
• Optimised ELO code
• Optimised Kelly Criterion code
• Updated FootballProbabilityGrid to store its internal matrix as a numpy array
• Updated all example notebooks
• Increased unit test coverage
• Added CI/CD
• Removed Poetry from build step
• Updated documentation
• Added type hinting to Colley class
• Added type hinting to Massey class