Articles
Getting started
How to install footBayes, the models it provides and the data used in the examples.
- Get started with footBayes
- Installation, the overall workflow, the data format and the two datasets used in the articles.
Models and inference
The goal-based models and how to fit them by maximum likelihood and in a Bayesian framework, with static or dynamic team abilities, and how to estimate historical team strengths.
- Goal-based models in a nutshell
- The models of the package, their parameters, the prior distributions and the inference algorithms.
- Static models
- Maximum likelihood and Bayesian fits with constant team abilities, prior sensitivity and approximate algorithms.
- Dynamic models
- Weekly and seasonal team abilities and the four specifications of the evolution variance.
- Historical strengths: the Bradley-Terry-Davidson model
-
Team strengths from match outcomes with
btd_foot(), and their use as a covariate instan_foot().
Checking, predicting and comparing
How to check the fitted models, predict out-of-sample matches and the final league table, and compare competing models.
- Model checking
-
Posterior predictive checks of the goal differences with
pp_foot()andbayesplot. - Predictions
- Match probabilities, home win probabilities and league table predictions.
- Model comparison
-
Predictive metrics on a test set with
compare_foot(), and information criteria withloo.