Using the Bradley-Terry model, we compare all teams in D1 Women’s Volleyball to assign ratings and eventually rankings.
Updated weekly, you can scroll through this dashboard to view current / historical VolleyDork rankings.
Pairing the VolleyDork Ratings from our B-T model with OpenVolley’s volleysim package for R, we’re able to run simulated matches to predict the winners of upcoming matches.
Future iterations will include a Brier score to help assess the accuracy of our probabilistic predictions.
match predictions by set
An even deeper look into how our predictions breakdown. This dashboard shows the probabilities of all possible outcomes.
how it works
Statistics Professor Dwight Wynne explains how the Bradley-Terry model works, what inputs it requires, and how we use the results to make predictions.
Accuracy of predictions
Using Brier Scores to measure how accurate our predictions are, we’re able to evaluate how well we can predict the future.
And more importantly, are we improving our predictions over time as we gather more information about what’s important.