Track team performance across seasons and events.
Our ratings are computed with an Elo-style model that updates teams based on how they performed relative to expectations at each tournament.
For each tournament, we estimate an expected score from teams’ current ratings, compare it to the actual scores, and adjust ratings accordingly. Bigger upsets and stronger-than-expected performances lead to larger changes, while expected outcomes move ratings less.
Model parameters are calibrated by minimizing a pairwise ranking loss on historical national tournament results. This loss penalizes incorrect ordering more heavily near the top of the standings, rewards confident correct predictions, and is optimized over many simulations to produce ratings that best predict Nationals outcomes.
💡 Tip: Click on data points to view detailed tournament results.
No chart data to display