Close Menu
    [OC] My model's predictions for the 2026 Tony Awards, built from precursor-award results and 16 years of data
    Data

    [OC] My model’s predictions for the 2026 Tony Awards, built from precursor-award results and 16 years of data

    Share.
    View 5 Comments

    5 Comments

    1. dsbuddy on

      **Source:** Tony + precursor award results (Drama Desk, Outer Critics Circle, Drama League, NY Drama Critics’ Circle) for 2009-2025, via Wikipedia, plus each show’s pre-ceremony reviews, commercial status, and predicted-odds standing.

      **Tools:** Python for the model (a naive Bayes weighted vote), matplotlib for the charts.

      **Method:** Every input is time-locked to what was known before each ceremony. Accuracy is measured leave-one-year-out (train on every season but one, predict the held-out year): ~62% across all 26 categories vs. ~54% for a “pick whoever won the most precursors” baseline. Bars are colored by confidence tier.

      **Code + data + more charts:** [https://github.com/dsbuddy/tony-awards-predictor](https://github.com/dsbuddy/tony-awards-predictor)

    4. Mrrandom314159 on

      What’d you feed in as the features for the model? And how’d it test on the test sets?