Sioux fan tournament prediction bias

One of my favorite parts about NCAA Hockey Bracket Challenge (did anyone else notice how EVERYONE is trying to do their own this year? Imitation and flattery and all that…) is seeing what Sioux fans pick. It’s an interesting barometer of their optimism and outlook on various teams and conferences.

The aggregate picks are available, and I’ve also compared them to the previously posted KRACH predictions.

Some takeaways:

  • 99.4% of Sioux fans pick UND to beat Yale, a figure 26% higher than KRACH. Lest one conclude that all Sioux fans are always homers, only 65% picked UND to beat UNH last year, despite KRACH calling it a draw.
  • The biggest difference between Sioux fans’ picks and KRACH isn’t the UND game, it’s BC over Alaska. 92.5% of Sioux fans pick BC, while KRACH predicts the game 62%-38% in favor of BC. This is the largest spread between Sioux fans picks and KRACH probabilities in the last three years.
  • Other than their love (fear?) of BC, Sioux fans have no respect for Hockey East. 73% picked Cornell to beat UNH, despite Cornell being 47%-53% underdogs in KRACH; 98% picked Wisconsin to beat Vermont, despite Wisconsin only being 70%-30% favorites in KRACH.

3 thoughts on “Sioux fan tournament prediction bias”

  1. i consider krach and user picks apples and oranges. krach measures the mathematical chances of each team winning in a particular round. even if we could look at non-biased picks, i think we’d find the percentages to be quite different from krach because each person is more likely to pick the favored team, even if that team is only a 60-40 favorite according to krach.

    i would imagine that even in a 60-40 krach split, over 80% of those making non-biased picks would pick the 60 team.

  2. krango – Good point. Probability of a team winning is an entirely different thing from proportion of entrants who picked a team to win, despite both being measured in percentages.

    However, empirical evidence over the 7 or so years we’ve been running this doesn’t agree that people overwhelmingly pick the 60-40 favorite because they’re trying to pick only the winners. They include some likely upsets to make their bracket more unique.

    I’m suggesting that the correlation between who many people pick each upset and the KRACH probability of a particular team winning is interesting, but you’re absolutely right that they’re completely different beasts.

Leave a Reply