Mirrorball predictors: does leading on judges win?
Judges’ leader after week 5 wins 53% of seasons — weak but best of the week-3/5/7 checkpoints.
Ballroom Data · Artometrics-style report · Nov 3, 2026 (America/Los_Angeles)
Headline
In 30 seasons with an identified Mirrorball winner, the cumulative judges’ leader after week 5 went on to win 53% of the time — better than after week 3 (37%) or week 7 (33%). Judges alone are a weak crystal ball; week 5 is the least-bad checkpoint in this build.
Methods & caveats
Leader = highest cumulative judge_mean sum through week K. Season 31 winner missing from winners extract; S35 excluded as incomplete. Fan votes omitted.
Coverage: seasons 1–35; 3,602 performances; 11,027 individual judge marks in the cleaned Ballroom Data SQLite build (Nov 3, 2026, America/Los_Angeles).
Known gaps: Season 31 Mirrorball winner is not tagged as Winners in the performance table (Charli D’Amelio / Mark Ballas is missing from our winners extract). Seasons 33–35 largely lack Kaggle ages/hometowns. Season 35 celebrity names in the wiki scrape are often first-name-only or garbled. Dance-style strings still include some unclean variants. We never impute missing values.
Sources: howisonlab/dwts_dataset (GitHub); Kaggle brupley/dancing-with-the-stars-dataset (CC BY-SA 4.0); Wikipedia DWTS season pages for seasons 32–35 (CC BY-SA 4.0). Scores are per-judge means on a 0–10 scale so 3-judge and 4-judge weeks are comparable. Marathons/relays/megamixes are excluded from style averages unless noted.
Disclaimer: Unofficial fan analysis. Not affiliated with BBC Worldwide, Disney, or ABC.
Downloads
Quarto source, rendered report, chart exports (PNG/SVG), and the analysis CSV.
