Congratulations to all participating teams! Here are the leader boards for the two tasks.
Please note the following points:
The Prediction Set contained a total of 736 matches. Of these matches, 22 could not be included in the final evaluation because the location was changed (e.g., home/away swap), the game was rescheduled or the game was aborted. Therefore, the final results are based on the predictions of the remaining 714 matches.
The field "Gap%" refers to the difference in percentage to the top-performing model.
The leader boards include four baseline models: Baseline_X_1, Baseline_X_N, Baseline_0_N, and Baseline_0_1. These models use only elementary statistical information and serve as null models. These models will be described in more detail in the special issue.
The leader boards also include the predictions by teams that include a member of the organizing team. These predictions are out-of-competition, of course. Another out-of-competition model is the bookmaker model, which is based on bookmaker odds.
Task 1: Prediction of exact scores.
Task 2: Prediction of probabilities of win, draw, and lose.